Docs

Metrics and definitions

How Signal over Noise defines and counts every figure it reports, and the full list of named metrics an operator or agent can ask for.Read this page as Markdown
Signal over Noise reports figures as named metrics, each with one definition that has been checked against the platforms’ own reports. Pulse and Ask ask for a figure by its name wherever one exists, so the same question gets the same number wherever it is asked. A question no named metric can answer is answered from the underlying tables instead, without that check.

The choices these definitions rest on

Where a definition below says otherwise, the definition wins.
  • Shopify is the reference. Its definitions win, and other backends are bent to them. Per-store exceptions are workspace config, never code.
  • Reconciliation is the test. Daily figures must agree within 1% of the platforms' own reports, so where a definition could go either way, the one the platforms use wins.
  • A completed day is final. A refund lands on the day it was refunded, as both platforms report it, so yesterday's net sales never changes tomorrow.
  • Money stays in the store's own currency, using the platform's own converted amounts. Nothing is converted to a workspace currency.
  • A day starts at midnight in the store's own timezone, read from the platform rather than from workspace settings.

Key definitions

How the main figures are counted, and where they differ from the report your store or ad platform shows.
MetricDefinitionHow it differs from the platform's report
Net salesSum of order value after discounts, excluding tax, shipping and fees, minus refunds issued that day.Calculated as Shopify Analytics calculates net sales. Against WooCommerce Analytics it can be lower: fee lines such as import duties are left out, because they are not the store's to keep, and an order counts on the day it was placed, where WooCommerce Analytics counts it on the day it was paid. A refund counts on the day it was issued, so a past day never changes.
Order lifetime netNet sales with each refund charged back to the day its order was placed.Unlike net sales, a refund is charged back to the day its order was placed rather than the day it was issued.
OrdersCount of orders placed that day.Counted as Shopify counts orders: a cancelled order stays in unless it was refunded, and test orders never count. On WooCommerce, processing, on-hold, completed and refunded orders count, test orders are left out where the store has said how to recognise them, and an order counts on the day it was placed, where WooCommerce Analytics counts it on the day it was paid.
Average order valueNet sales divided by orders, for any period a ratio of the period's sums.
RefundsMerchandise refunded that day, excluding tax and shipping, whatever day the order was placed.Counted on the day the refund was issued, as the store reports it, whatever day the order was placed.
CustomersDistinct customer keys that ordered that day.
Active customersDistinct customers who ordered in the period asked for, counted at query time.
New customersCustomers whose first order in retained history fell that day, where that history is complete for them.
Unclassified customersOrders that look like a customer's first where nothing can decide: the platform publishes no lifetime order count, and the order falls on the first day of retained history.
SignupsAccounts registered that day, wherever the platform publishes a registration date.
Order indexWhich of a customer's retained orders an order is: 1 their first held, 2 their second.
Retained customersCustomers who placed a REPEAT order in a given 30-day window after the one they were won in.
Cohort customersChances to come back: one per customer per 30-day window their own history has reached.
Retention rateRetained customers divided by cohort customers.
Net revenue retentionRepeat revenue in a window divided by that cohort's own first-order revenue, both on the order lifetime net basis.
Ad spendWhat the connected ad platforms spent, in the store's currency at the European Central Bank rate for the spend's own day. Read from each ad platform directly: Meta Ads and Google Ads.Converted into the store's currency at the European Central Bank rate for each day, so it can differ from what an ad platform shows in its own currency. Only ad platforms connected directly are included.
Blended CACAd spend over new customers, for any period a ratio of the period's sums.Uses only the connected ad platforms' spend, so it understates the cost wherever another channel is paid for. It is not an ad platform's own cost per purchase.
Paid CACAd spend over the new customers whose first order the store's own attribution credits to paid channels, by order tags first and then discount codes.
Platform-reported purchasesPurchases an ad platform attributes to itself, and their value.The ad platform's own claim under its own attribution window, which it can restate: up to 28 days on Meta, and on Google for the account's conversion window. Two platforms can claim the same order, so it is never the store's order count.
Paid efficiencyCost per click and click-through rate over the platforms' own clicks and impressions; platform cost per purchase, platform return on ad spend and platform conversion rate over their own purchase claims. Each a ratio of the period's sums.Each platform counts clicks its own way, so compare a ratio within one platform. The purchase-based ratios are the platform's own claim.
Attributed orders, revenue and new-customer orders (Triple Whale)What Triple Whale credits to each channel, one figure per attribution model.Triple Whale's own attribution, shown beside the store's own and never added to it.
SessionsSite sessions as Google Analytics counts them, by the property's own day and the channel the session arrived by.
Site conversion ratePurchases Google Analytics' tagging reported over its sessions, for any period a ratio of the period's sums.Both sides come from Google Analytics, whose tagging sees only part of the store's orders, so it is not the store's orders over sessions.
Landing page and source conversionGoogle Analytics' sessions and purchases by the page a session landed on and by its source and medium, each with its conversion rate.
Purchase funnelHow many times shoppers viewed an item, added to cart, began checkout, added shipping details, added payment details and bought, as Google Analytics counted the events, by device, channel and landing page, with the rate from each step to the next.
Tracking healthThe share of tracked purchases whose session could not be credited to any channel or source, and the tracked purchases linked to an order over the store's orders.
Tracked purchases and their credit (Google Analytics)Each purchase the tagging reported, once, with the source, channel, campaign and landing page of its session, linked to the store's order by its transaction ID; and the same purchases as the property's attribution model credits them.Revenue is what the tagging sent, converted like ad spend, and is never the store's revenue.
Page frictionSessions in which visitors met repeated frustrated clicks, clicks on things that did nothing, clicks that triggered an error, or script errors, as Microsoft Clarity records them, across the site and per page, with each as a share of the sessions counted.Microsoft Clarity's own count of sessions, which is never Google Analytics' sessions.
Page friction issueA page with at least 30 sessions, at least 5 of them meeting one kind of friction, and a share meeting it at least twice the whole site's: on a day, or, over the bar that day but short of the floors, over the 7 then 14 days ending with it.
Fulfilment backlogPaid, uncancelled orders unfulfilled for more than three days, as at the end of each day: an order cancelled or refunded in full later still counts on the days it was waiting. Asked over a period, the backlog on its last day: a stock is never summed or averaged across days.
Average days to fulfilWhole days from an order's day to its fulfilment day, both on the store's own clock, over the orders fulfilled in the period.On WooCommerce an order counts as fulfilled when it was marked complete, which some stores do in batches.
Units per order and average unit priceProduct units over orders, and product line value over product units.
Order value before refundsProduct line value over orders, before refunds. It is units per order multiplied by average unit price, exactly.
Refund rateMoney refunded on the orders placed in a period, whenever refunded up to the day the figures are as of, over those orders' net sales.A refund is moved back to the day its order was placed, so a recent period keeps rising for weeks as refunds arrive.
Discount to salesDiscount given over net sales, both before tax: on a tax-inclusive Shopify order the discount took its share of tax off with the price, and that share is not counted.
CRM revenue per recipientCRM-reported revenue over recipients.The CRM's own attribution and currency, and still counting for recent sends.

Every named metric

The 283 metrics below can be asked for by name. A metric is available once the tool it is built from is connected and your workspace’s nightly build has read it: see Integrations. An agent asks for a metric by the name in the second column.
MetricNameWhat it means
Active customersactive_customersDistinct customers who ordered in the period. A customer who bought three times is one. Not the same as customers, which counts customers the store knows by when it first knew them.
Ad clicksad_clicksClicks of any kind on an ad.
Campaigns and ad setsad_entitiesCampaigns and Meta ad sets on the period's last day, by whether they can be changed. Asked split by ad_entity_day__entity_level, since a campaign and its ad sets are not one count; split by ad_entity_day__verdict and ad_entity_day__hold_until, with the campaign's or ad set's name, to say which can change and which wait until when.
Hold rate (Meta)ad_hold_rateThruPlays over three-second views for a video ad: of the people its opening stopped, how many watched it through. Read beside ad_three_second_view_rate, which says how many it stopped. Empty for an ad with no three-second view.
Ad impressionsad_impressionsTimes an ad was shown.
Ad-level clicksad_level_clicksClicks on Meta's and Google's ads, of any kind.
Ad cost per purchase (Meta's claim)ad_level_cost_per_purchaseAn ad's spend over the purchases Meta attributed to it. Meta's own claim, over Meta's ads alone.
Ad click-through ratead_level_ctrClicks over impressions for an ad. Asked by ad_day__ad and by day or week, a falling rate at the same spend is an ad wearing out.
Ad-level impressionsad_level_impressionsImpressions of Meta's and Google's ads.
Ad-level purchases (Meta's claim)ad_level_purchasesPurchases Meta attributed to its ads, under its own window. Meta's own claim; Google's ads carry none at this grain.
Ad-level spendad_level_spendSpend through Meta's and Google's ads, in the store's currency; split by ad_day__ad_platform for one. A part of ad_spend, never added to it.
ThruPlays (Meta)ad_level_thruplaysPlays of an ad's video to its end, or for 15 seconds.
Three-second views (Meta)ad_level_video_viewsMeta's three-second video views.
Ad link clicksad_link_clicksClicks on a link in an ad.
Ad set clicksad_set_clicksClicks of any kind on an ad in an ad set.
Platform cost per purchase by ad setad_set_cost_per_purchaseAn ad set's spend over the purchases its platform attributed to it, summed over the period. The platform's own measure, asked split by platform. Empty where an ad set claimed no purchase.
Ad set impressionsad_set_impressionsTimes an ad was shown through an ad set.
Platform-reported purchases by ad setad_set_platform_purchasesPurchases an ad platform attributed to its own ad sets, under its own window and restated by it. Never a count of customers or orders. Asked split by platform, because two platforms claim the same orders.
Ad set spendad_set_spendWhat was spent through ad sets (Google's ad groups are ad sets here), in the store's currency. For comparing ad sets with each other. It is not total spend: a Google Performance Max campaign has no ad groups, so this can be less than ad_spend, and it is the same money, never added to it.
Ad spendad_spendWhat the connected ad platforms spent, in the store's currency. Only the platforms that are connected: it understates total marketing spend wherever another channel is paid for.
Three-second view rate (Meta)ad_three_second_view_rateThree-second views over impressions for a video ad: how often its opening stops the scroll.
ThruPlay rate (Meta)ad_thruplay_rateThruPlays over impressions for a video ad: how often it is watched through.
Average days to fulfilaverage_days_to_fulfilAverage whole days, on the store's own clock, from an order's day to its fulfilment day, over the orders fulfilled in the period. On WooCommerce "fulfilled" is when the order was marked complete, which some stores do in batches.
Average first response (hours)average_first_response_hoursAverage hours from a contact opening to the first reply from the merchant's side, an automatic reply included, over the contacts that had one. Counted on the day the contact opened.
Average first human response (hours)average_human_first_response_hoursAverage hours from a contact opening to the first reply from a person, over the contacts a person replied to. Counted on the day the contact opened, so a recent day's figure rises as slower replies land.
Average order valueaverage_order_valueNet sales divided by orders. A ratio of two reconciled figures rather than an average of a per-row average, which is why it is declared here instead of summed from a column of its own.
Average resolution (hours)average_resolution_hoursAverage hours from a contact opening to its latest close, over the contacts closed now whose close is not before they opened. A ticket reopened and closed again counts to its later close. Counted on the day the contact opened.
Average satisfaction scoreaverage_support_ratingThe average satisfaction score, 1 to 5, over the scores given in the period.
Average unit priceaverage_unit_priceWhat a product unit sold for, after discounts: merchandise sales over merchandise units. Moves with prices, discounts and mix.
Blended CAC (connected ad sources)blended_cacBlended customer acquisition cost: connected ad spend over every new customer, paid or not. Blended across the connected ad platforms only. Summed spend over summed customers for the period, never an average of days. Ask it only over days the ad history covers, and read it beside unclassified_customers.
Brand share of search purchases (Google)brand_search_purchase_shareThe share of a campaign's search-term purchases that came from searches for the store's own name: people who would likely have bought anyway, so a high share flatters the campaign's return.
Brand search purchases (Google's claim)brand_search_purchasesPurchases Google attributed to searches for the store's own name.
Brand search spend (Google)brand_search_spendSpend on searches for the store's own name.
Brand share of search spend (Google)brand_search_spend_shareThe share of search-term spend on searches for the store's own name, between 0 and 1. Asked by search_term_week__term_campaign.
Cancel flow sessionscancel_flow_sessionsVisits to the cancel flow, on the store's day each started. Ask by cancel_flow_session__cancel_reason for the merchant's own reasons, cancel_flow_session__status for its status, as the system spells it, with what each value means unproven, and cancel_flow_session__offer for what it showed. A V3 flow keeps only its status and type. A reason the subscriber typed never appears.
Click-through rateclick_through_rateAd clicks over ad impressions. Split by ad_campaign_day__ad_platform, it is each platform's own, for the same reason as cost_per_click.
Cohort customerscohort_customersChances to come back: one per customer per 30-day window their own history has reached. The denominator of retention_rate, and not a customer count.
Cohort first order netcohort_first_order_netWhat a cohort's first orders were worth, on the lifetime-net basis. Counted once per window the customer is eligible for, so it is the denominator of revenue retention rather than a revenue figure. Use net_sales to state revenue.
Cohort retained netcohort_retained_netWhat a cohort's repeat orders in a window were worth, on the lifetime-net basis with refunds charged back to the order that caused them.
Contacts closed without a personcontacts_without_human_replyContacts closed with no reply from a person: answered by automation alone, or closed unanswered.
Cost of goods soldcost_of_goods_soldWhat the merchandise sold cost the merchant: each line's units at its variant's cost as of the order's day, from the store's own catalogue. An order from before the first stock read takes the earliest read's cost (cost_line__is_cost_backdated). Only lines with a cost count, so read it beside costed_sales_share: a variant the merchant set no cost for adds nothing here, and is not free. Empty for a period with no costed line, never nought.
Cost per clickcost_per_clickAd spend over ad clicks. Platforms count clicks differently (Meta's are clicks of any kind), so a figure across platforms blends two measures; split by ad_campaign_day__ad_platform, it is each platform's own.
Costed merchandise salescosted_merchandise_salesMerchandise sales on the lines that have a cost, after discounts and before refunds: the revenue gross_margin is earned on. Nought for a period with no costed line, so costed_sales_share is nought there too.
Share of sales with a costcosted_sales_shareThe share of merchandise sales whose variant has a cost, between 0 and 1: how much of the revenue gross_margin and cost_of_goods_sold cover. Under one where the merchant set no cost for some variants, or a line has no variant. Nought also where the store's cost had not been read by then, or its read was refused or failed: before saying the merchant set no costs, check core_stock_reads for a complete read.
CRM bouncedcrm_bouncedMessages that bounced.
CRM campaign recipientscrm_campaign_recipientsMessages the store's CRM sent in campaigns, on every channel, on each campaign's send day. Asked beside orders and net_sales by metric_time__day, it sets a send cut against the same days' sales in one query. Sends are on UTC days and sales on the store's own, so an evening send can sit a day from its orders. Split by channel, it is a second query, crm_recipients by crm_message_day__message_kind and crm_message_day__crm_channel, since orders have no channel.
CRM clickscrm_clicksClicks on CRM messages.
CRM conversionscrm_conversionsConversions the CRM attributes to its messages.
CRM deliveredcrm_deliveredMessages delivered. Final as soon as sent, so safe to compare at once.
CRM failedcrm_failedMessages that failed to send.
CRM flow message pausescrm_flow_message_pausesTimes a live flow message was set to manual, draft or disabled, on the UTC day each is dated by, whether or not its flow stayed live. A message paused inside a live flow sends nothing while the flow reads live. Split by flow and message to say which. History starts the first night flow steps were read. On Omnisend an automation is its own message, so a switched-off automation counts here and in crm_flow_pauses alike: never add the two.
CRM flow pausescrm_flow_pausesTimes a live CRM flow was paused, on the UTC day each pause is dated by: set to manual or draft on Klaviyo, switched off on Omnisend. A flow that stops being listed is not a pause; it is a status change to empty.
CRM flow recipientscrm_flow_recipientsMessages the store's CRM sent from flows, on every channel, on the day each sent. Asked beside orders and net_sales by metric_time__day like crm_campaign_recipients.
CRM flow status changescrm_flow_status_changesTimes a CRM flow's status changed, on the UTC day each change is dated by. Split by flow and by the status it changed to and from to say which flow went where. History starts with the nightly reads, so a flow changed before them shows no change. A flow missing from the latest finished run counts as a change to empty (no longer listed), and that change disappears if the next night lists the flow again.
CRM flow trigger changescrm_flow_trigger_changesTimes a Klaviyo flow's trigger or profile filter changed, on the UTC day each is dated by: a different list, segment or metric, a changed trigger filter, or a changed profile filter. A flow no run lists any more counts once, as a change to no trigger. Split by flow to say which. History starts the first night triggers were read.
CRM openscrm_opensEmail opens. SMS and push have none.
CRM recipientscrm_recipientsMessages the store's CRM sent, by campaigns and flows, on every channel.
CRM reported revenuecrm_reported_revenueRevenue the CRM platform attributes to its messages under ITS OWN attribution window, which differs by platform, in the account's currency, or in NO currency (unconverted mixed currencies, tax and shipping included) where the run_currencies table gives crm as XXX; then never compare it with net_sales, and never compare it across periods or between messages either, because the total moves with which currency customers paid in. A SUBSET of store revenue: never add it to net_sales. Counted on a campaign's send day, so recent campaigns are still counting.
CRM revenue per recipientcrm_revenue_per_recipientCRM-reported revenue per person sent to: which campaigns and flows earn for the attention they take. The CRM's own figure, in its account's currency, and still counting for recent sends. Where the run_currencies table gives crm as XXX, its numerator is a sum of unconverted mixed currencies, so this is in NO currency: then never compare it across sends, messages or periods, and never rank by it. Otherwise it is comparable across sends of different sizes.
CRM spam complaintscrm_spam_complaintsSpam complaints on CRM messages.
CRM unsubscribescrm_unsubscribesUnsubscribes from CRM messages.
CustomerscustomersDistinct customers the store knows, keyed by platform id or hashed email. A customer who bought twice is one customer.
Sessions with clicks on things that did nothingdead_click_sessionsSessions across the site that met clicks on things that did nothing.
Share of sessions with clicks on things that did nothingdead_click_shareThe share of the site's sessions that met clicks on things that did nothing: those sessions over the sessions counted, summed over the period.
Spend on dead pagesdead_destination_spendSpend sent to a page on a day it was dead: a link whose check found an error status, or a product or collection redirected to the home page, search or an all-products page; or a product the store had deleted, archived, left in draft or not published. The spend at stake.
Destination page match ratedestination_page_match_rateThe share of spend on the store's own pages that went to a page Clarity or Google Analytics recorded that day, between 0 and 1. Read before setting any page's spend beside its friction or conversion, by ad_destination_day__destination_platform: a low share means the ads' paths are spelled other than the site's, or the site's sources are not connected, and that platform's page reading is withheld.
Spend by destinationdestination_spendSpend by the page it sent people to, in the store's currency. On a day a platform's spend by product was read it is all of that platform's spend, the part with no known page kept as none. Otherwise a part of ad_spend, never added to it: Meta's links begin with the first night its destinations were read, and Google reports a landing page only where it can name one.
Discount givendiscount_givenMoney given away by discount codes.
Discount to salesdiscount_to_salesDiscount given as a share of net sales: how much of the period's revenue was bought with discounts. Split by market or new versus repeat through the order.
Discounted ordersdiscounted_ordersOrders a discount code was redeemed on.
Sessions with clicks that triggered an errorerror_click_sessionsSessions across the site that met clicks that triggered an error.
Share of sessions with clicks that triggered an errorerror_click_shareThe share of the site's sessions that met clicks that triggered an error: those sessions over the sessions counted, summed over the period.
Sessions friction was counted infriction_counted_sessionsSite sessions on the days whose friction counts are sessions and that reported every kind: what a site-wide share of sessions is taken over. No answer for a day read means its friction was counted in events and cannot be stated as a share.
Sessions on the page that met the frictionfriction_issue_affected_sessionsOf the page's sessions, those that met the friction.
Sessions on the pagefriction_issue_page_sessionsThe page's sessions over the days it was judged on: the day it stood out, or the window page_friction_issue__window_days names.
Share of the page's sessions that met the frictionfriction_issue_shareThe share of the page's sessions that met the friction over the days it was judged on.
Sessions on the site that met the frictionfriction_issue_site_affected_sessionsOf the site's sessions, those that met the friction anywhere.
Sessions on the sitefriction_issue_site_sessionsThe whole site's sessions over the same days.
Share of the site's sessions that met the frictionfriction_issue_site_shareThe share of the whole site's sessions that met the same friction over the same days: what the page is measured against.
Page friction issuesfriction_issuesPages whose friction stood out from the rest of the site on a day, or over the window ending with it, counted once per page, day and kind of friction. A day with nothing standing out has no row, and neither has a day not read: friction_sessions says which days were read.
Sessions (page friction)friction_sessionsSessions by people across the whole site, as the page friction source records them. Not the web analytics source's sessions, which count differently. Known only for the days read.
Fulfilled ordersfulfilled_ordersOrders fulfilled in the period, on the day they were fulfilled.
Fulfilment backlogfulfilment_backlogOrders waiting longer than the store's threshold, on the last day of the period asked for. Split by metric_time__day it is the backlog day by day.
Backlog orders on holdfulfilment_backlog_heldBacklog orders the store's platform held, for any reason, that were not preorders. Empty, not zero, where the backlog could not be split: see split_basis.
Backlog late ordersfulfilment_backlog_lateThe rest of the backlog: orders past the threshold and within stale_after_days (a week) of being placed, neither preorders nor held. What is genuinely stuck. Where split_basis is age_only it is told by age alone and holds that platform's preorders and held orders too. Empty, not zero, where the backlog could not be split: see split_basis.
Backlog preorders not overduefulfilment_backlog_preordersBacklog orders that are preorders not known to be overdue: inside their promised date, promised none, or tagged as preorders and older than the read of fulfilment orders reaches, so of unknown date. On the last day of the period. Empty, not zero, where the backlog could not be split: see split_basis.
Backlog preorders overduefulfilment_backlog_preorders_overdueBacklog orders that are preorders waiting past the day they were promised for: a broken promise to the customer. Empty, not zero, where the backlog could not be split: see split_basis.
Backlog preorders with no datefulfilment_backlog_preorders_undatedBacklog orders that are preorders still inside their date and were promised no ship date at all, as the read held them, on the last day of the period. Empty, not zero, where the backlog could not be split.
Backlog stale ordersfulfilment_backlog_staleBacklog orders waiting more than stale_after_days (a week) since they were placed that are neither preorders nor held: most likely statuses nobody closed rather than parcels stuck. Where split_basis is age_only it is told by age alone and holds that platform's preorders and held orders too. Empty, not zero, where the backlog could not be split: see split_basis.
Fulfilment daysfulfilment_daysDays from order to fulfilment, summed over the orders fulfilled.
Add-to-cart rate (web analytics)funnel_add_to_cart_rateItems added to the cart over items viewed, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Add-to-carts (web analytics)funnel_add_to_cartsTimes shoppers added an item to the cart, as web analytics counted the add_to_cart event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales.
Checkout completion rate (web analytics)funnel_checkout_completion_ratePurchase events over checkouts begun, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Cart-to-checkout rate (web analytics)funnel_checkout_rateCheckouts begun over items added to the cart, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Checkouts begun (web analytics)funnel_checkouts_begunTimes shoppers began checkout, as web analytics counted the begin_checkout event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales.
Item views (web analytics)funnel_item_viewsTimes shoppers viewed an item, as web analytics counted the view_item event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales.
Payment details added (web analytics)funnel_payment_info_addedTimes shoppers added payment details in checkout, as web analytics counted the add_payment_info event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales.
Shipping-to-payment rate (web analytics)funnel_payment_step_ratePayment details added over shipping details added, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Purchase events (web analytics)funnel_purchase_eventsPurchase events the tagging sent, as web analytics counted the purchase event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales. The tagging's count, never the store's orders; for tracked purchases ask web_purchases.
Payment-to-purchase rate (web analytics)funnel_purchase_step_ratePurchase events over payment details added, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Shipping details added (web analytics)funnel_shipping_info_addedTimes shoppers added shipping details in checkout, as web analytics counted the add_shipping_info event. Split by web_funnel_day__device_category, web_funnel_day__funnel_channel or web_funnel_day__landing_page. Empty, never zero, before the store's tagging first sent the event; zero on a quiet day after that, a broken tag's included, so set a fall to zero against the store's orders before reading it as lost sales.
Checkout-to-shipping rate (web analytics)funnel_shipping_step_rateShipping details added over checkouts begun, as web analytics measures them, summed over the period and only over days both steps were recorded. Web analytics' own measure, never set against the store's orders. Empty where the store's tagging does not send one of its steps. Read it beside its two counts, because a slice with few events has a rate that means nothing; over a period in which a step began to be recorded, its count also sums days the rate leaves out.
Gross margin (costed lines)gross_marginWhat the costed lines sold for, after discounts and before refunds, less what their units cost: costed_merchandise_sales less cost_of_goods_sold. Only over the lines that have a cost, so state costed_sales_share beside it, and empty for a period with none. Before refunds, shipping and fees.
Gross margin share (costed lines)gross_margin_shareGross margin as a share of what the costed lines sold for, between 0 and 1. What a unit of revenue leaves after the goods are paid for, on the lines that have a cost; state costed_sales_share beside it.
Sales by groupinggroup_salesLine value in a product grouping, after discounts and before refunds. NOT ADDITIVE across groupings and not reconciled. Use it to compare groupings; use net_sales to state revenue.
Units by groupinggroup_unitsUnits sold in a product grouping, before refunds. NOT ADDITIVE across groupings: a product in three categories is counted in all three.
Item salesitem_salesLine value after discounts and BEFORE refunds, for ranking what sold. Not the same as net sales on a day with returns, and not reconciled against the store's own report. Use net_sales to state revenue and this to rank products or groups.
Known destination shareknown_destination_shareThe share of a platform's spend whose page is known, between 0 and 1, over the days its spend by product was read, when its rows add up to what it spent. Read by ad_destination_day__destination_platform: the rest is how much the brief cannot place on a page, and a claim about where the money went covers only this share. Empty for a platform whose spend by product is not read.
Spend with a known destinationknown_destination_spendSpend whose page is known, in the store's currency: a link's path, or a product the store holds.
Conversion rate by landing page (web analytics)landing_page_conversion_rateA landing page's conversion rate as web analytics measures it: its purchases over its sessions, summed over the period. Ask it split by web_landing_page_day__landing_page, and read it beside the page's sessions, because a page with few sessions has a rate that means nothing.
Purchases by landing page (web analytics)landing_page_purchasesPurchases the tagging reported in sessions that began on a page. The tagging's count, never the store's orders.
Sessions by landing pagelanding_page_sessionsSessions that began on a page, by web_landing_page_day__landing_page. For the site's sessions ask web_sessions, which is reconciled against the source's own total; this report is not.
Leading ad's share of spendlead_ad_spend_shareThe share of an ad set's week its leading ad carried, between 0 and 1. Asked by ad_week__week_ad_set: near one, the ad set rests on one ad.
Spend on pages the site's sources recordedmatched_page_spendOf store_page_spend, the spend sent to a page Clarity or Google Analytics recorded that day.
Merchandise salesmerchandise_salesProduct line value after discounts and before refunds, excluding shipping and fees. For the basket, not for stating revenue.
Merchandise unitsmerchandise_unitsProduct units sold, excluding shipping and fee lines.
Net revenue retentionnet_revenue_retention_rateRepeat revenue over the cohort's own first-order revenue. Can exceed 1, because returning customers may spend more than they did first time. On the lifetime-net basis throughout, which is not the basis net_sales uses.
Net salesnet_salesRevenue after discounts, excluding tax, shipping and fees, net of refunds on the day they were refunded.
New customersnew_customersCustomers won in the period: the order was their first and the history reaches back far enough to prove it. Counted on the order, so it splits by channel, country, sales surface, UTM tag and referring domain. Read it beside unclassified_customers, which carries the ones nothing can classify.
New heavy file bytesnew_heavy_resource_bytesThe bytes a heavy file that arrived on a measured page transferred (page_speed_new_heavy_resources): a file among the page's ten largest, of at least a megabyte, that the page's previous measured night on the same device did not load. Asked split by device, since a file a page loads on both devices is a row on each; split also by page, file and night to name it.
Spend on searches that bought nothing (Google)non_converting_search_spendSpend on search terms in weeks they bought nothing. Asked by term, it names them. Not all are negative keywords to write: a term matching a product the store is selling is demand, so read search_term_week__term_range_check before calling one.
Sales of orders placedorder_sales_by_order_dayNet sales of the orders placed in the period, on the day each was placed. The denominator of refund_rate.
Order value before refundsorder_value_before_refundsProduct line value per order, before refunds: merchandise sales over orders. Exactly units_per_order times average_unit_price, which average_order_value is not, because that has the period's refunds taken off and counts every line. For explaining a basket movement, not for stating order value.
OrdersordersHow many orders the store took.
Page sessions friction was counted inpage_counted_sessionsA page's sessions on the days whose friction counts are sessions: what the page's shares are taken over. Asked split by page.
Page sessions with clicks on things that did nothingpage_dead_click_sessionsSessions on a page that met clicks on things that did nothing. Asked split by page.
Page share of sessions with clicks on things that did nothingpage_dead_click_shareThe share of a page's sessions that met clicks on things that did nothing, summed over the period. Asked split by page.
Page sessions with clicks that triggered an errorpage_error_click_sessionsSessions on a page that met clicks that triggered an error. Asked split by page.
Page share of sessions with clicks that triggered an errorpage_error_click_shareThe share of a page's sessions that met clicks that triggered an error, summed over the period. Asked split by page.
Page INP (real visitors)page_field_inpReal Chrome visitors' 75th percentile interaction to next paint on the page over the 28 days before, in milliseconds. A level, asked split by page and device; a period answers with its last day.
Page LCP (real visitors)page_field_lcpReal Chrome visitors' 75th percentile largest contentful paint on the page over the 28 days before, in milliseconds. Slow to move; empty where Google has too few visitors. A level, asked split by page and device; a period answers with its last day.
Page image bytespage_image_bytesImages the page transferred in the night's run, in bytes. A level, asked split by page and device; a period answers with its last day.
Page LCP (lab)page_lab_lcpA page's largest contentful paint in the night's Lighthouse run, in milliseconds: how long its main content took to show on that device. One run is noisy; a file that arrived (page_speed_new_heavy_resources) is the evidence. A level, asked split by page and device; a period answers with its last day.
Page blocking time (lab)page_lab_tbtHow long the night's Lighthouse run found the page's main thread blocked, in milliseconds. A level, asked split by page and device; a period answers with its last day.
Page video and audio bytespage_media_bytesVideo and audio the page transferred in the night's run, in bytes. A level, asked split by page and device; a period answers with its last day.
Page performance scorepage_performance_scoreLighthouse's performance score for the page on that device, 0 to 1. A level, asked split by page and device; a period answers with its last day.
Page sessions with repeated frustrated clickspage_rage_click_sessionsSessions on a page that met repeated frustrated clicks. Asked split by page.
Page share of sessions with repeated frustrated clickspage_rage_click_shareThe share of a page's sessions that met repeated frustrated clicks, summed over the period. Asked split by page.
Page script bytespage_script_bytesScripts the page transferred in the night's run, in bytes. A level, asked split by page and device; a period answers with its last day.
Page sessions with script errorspage_script_error_sessionsSessions on a page that met script errors. Asked split by page.
Page share of sessions with script errorspage_script_error_shareThe share of a page's sessions that met script errors, summed over the period. Asked split by page.
Page sessionspage_sessionsSessions that viewed a page. NOT ADDITIVE across pages: one session views several, so it is asked split by page.
Page speed readspage_speed_readsPages the night's PageSpeed Insights read tried: one per page, device and night, whether Google measured it or not. Split by page_speed_day__measured and page_speed_day__unmeasured_reason to say which pages could not be measured and why; no answer means the store's pages are not read.
Page weightpage_weightEverything the page transferred in the night's run, in bytes. A level, asked split by page and device; a period answers with its last day.
Paid CACpaid_cacPaid customer acquisition cost: the same connected ad spend over paid new customers only. A second basis beside blended, never a replacement for it. Empty where the workspace states no attribution policy.
Paid new customerspaid_new_customersNew customers the store's own attribution put in its paid bucket. Empty where the workspace states no attribution policy.
Platform conversion rate (per click)platform_conversion_rateThe purchases an ad platform attributed to itself over its ad clicks, summed over the period: the platform's own conversion rate, under its own window and restated by it. Asked split by platform, because two platforms claim the same orders. Google's clicks are clicks on the ad; Meta's are clicks of any kind, a reaction or a profile visit included, so for Meta ask platform_link_click_conversion_rate. Not the store's conversion rate, which is site_conversion_rate.
Platform cost per purchaseplatform_cost_per_purchaseAd spend over the purchases the platform attributed to itself: what the platform says each purchase cost, under its own window. Asked split by platform, because two platforms claim the same orders. Not an acquisition cost, which is blended_cac.
Platform conversion rate (per link click, Meta)platform_link_click_conversion_rateThe purchases Meta attributed to itself over the clicks on a link in its ads, summed over the period: Meta's own conversion rate, under its own window and restated by it. Meta's alone, because only Meta reports link clicks. It can pass 1, since Meta also credits a purchase to an ad that was seen and never clicked. Not the store's conversion rate, which is site_conversion_rate.
Platform-reported purchase valueplatform_purchase_valueWhat an ad platform says the purchases it claims were worth, in the store's currency. Its own attribution, never the store's revenue. Asked split by platform, for the same reason as platform_purchases.
Platform-reported purchasesplatform_purchasesPurchases an ad platform attributed to itself, under its own window and restated by it. The platform's own measure: it is not a count of customers or of the store's orders, and it overlaps other channels' claims. Acquisition cost never reads it. Asked split by platform, because two platforms claim the same orders and their sum counts an order twice.
Platform return on ad spendplatform_return_on_spendWhat the platform says its purchases were worth, over ad spend: its own return on ad spend, under its own window. Asked split by platform. Never set against store revenue as though it were measured the same way.
Sessions with repeated frustrated clicksrage_click_sessionsSessions across the site that met repeated frustrated clicks.
Share of sessions with repeated frustrated clicksrage_click_shareThe share of the site's sessions that met repeated frustrated clicks: those sessions over the sessions counted, summed over the period.
Refund raterefund_rateShare of the period's order value later refunded. Only comparable between periods old enough for their returns to have landed.
RefundsrefundsMoney refunded, on the day it was refunded.
Refunds on orders placedrefunds_on_orders_placedMoney refunded on the orders placed in the period, whenever it was refunded. Keeps growing for weeks after the period closes, so a recent period always looks low.
Retained customersretained_customersCustomers who placed a repeat order in a 30-day window after the one they were won in. A customer who returned in three separate windows counts three times, which is what makes it the correct numerator for retention_rate.
Retention rateretention_rateThe share of a cohort that bought again. Split by cohort_cell__won_in_month and cohort_cell__months_since_first_order it is the retention curve; split by window alone it is the curve pooled across cohorts; unsplit it is the store's repeat rate. All three are right, because the denominator counts customer-windows rather than repeating a cohort size. A customer appears only once their own window has been reached.
Sessions with script errorsscript_error_sessionsSessions across the site that met script errors.
Share of sessions with script errorsscript_error_shareThe share of the site's sessions that met script errors: those sessions over the sessions counted, summed over the period.
Search impressions lost to budgetsearch_budget_lost_impressionsEligible impressions lost because a campaign's budget ran out.
Search eligible impressionssearch_eligible_impressionsThe impressions Google Search and Shopping campaigns were eligible for, as Google estimates them.
Search impression sharesearch_impression_shareThe share of the searches Google Search and Shopping campaigns were eligible for that they showed on, as Google estimates it. Google states it for those campaign types only. Ask it by ad_campaign_day__campaign_name to compare campaigns.
Search impressions (impression share known)search_impressionsImpressions of Google Search and Shopping campaigns on days Google stated their impression share: the numerator of search_impression_share.
Search impression share lost to budgetsearch_lost_to_budget_shareThe share of those eligible searches a campaign did not show on because its budget ran out: the sign of a campaign its budget holds back. High on a campaign whose cost per customer is good means budget would buy more; high on one whose cost is poor means the budget is already capping waste.
Search impression share lost to ranksearch_lost_to_rank_shareThe share of those eligible searches a campaign did not show on because its ad rank, its bid and quality, was too low: the sign of a campaign its bid or target holds back rather than its budget.
Search impressions lost to ranksearch_rank_lost_impressionsEligible impressions lost to too low an ad rank.
Search term clicks (Google)search_term_clicksClicks on a search term. Asked beside its spend and purchases, it tells a term that had the chance to buy from one that did not.
Search term cost per purchase (Google's claim)search_term_cost_per_purchaseA term's spend over the purchases Google attributed to it, asked by search_term_week__search_term. Google's own claim.
Search term purchases (Google's claim)search_term_purchasesPurchase conversions Google attributed to its search terms. Google's own claim.
Search term spend (Google)search_term_spendWhat Google's search terms cost, in the store's currency: a part of search spend, never added to ad_spend, as Google leaves out terms below its privacy threshold.
Weeks a search term was seen (Google)search_term_weeks_seenHow many weeks a search term cost money or was shown. A term seen once may never return, so a negative keyword would not have stopped its spend.
Shopping spend mapped to the store's products (Google)shopping_mapped_spendShopping spend on offers that map to a product the store holds.
Shopping spend mapped to products, share (Google)shopping_mapped_spend_shareThe share of Shopping spend whose offer maps to a product the store holds, between 0 and 1. Read before Shopping spend by product: a low share means the feed's offers are not the store's own ids.
Shopping purchases by product (Google's claim)shopping_product_purchasesPurchase conversions Google attributed to each product. Google's own claim.
Shopping return by product (Google's claim)shopping_product_return_on_spendA product's Google-attributed purchase value over its ad spend, asked by product__product_title or another product dimension. Google's own claim.
Shopping spend by product (Google)shopping_product_spendWhat each product's Google ads cost, in the store's currency.
Shopping purchase value by product (Google's claim)shopping_product_valueWhat Google says those purchases were worth, in the store's currency.
SignupssignupsAccounts registered, on the day they were registered rather than the day they first bought. What creates one is the platform's: on Shopify a checkout, a newsletter signup and an account registration alike; on WooCommerce an account registration only, never a guest checkout. Empty on a store whose platform publishes no registration date, and empty for a week in which nobody registered.
Site conversion rate (web analytics)site_conversion_rateSite conversion rate as measured by web analytics: purchases the tagging reported over sessions, both from the same source, summed over the period and never an average of days. It is not the store's orders over sessions, because the tagging misses some orders.
Stock availablestock_availableStock on hand, as units available to sell at the end of the period's last day: what the store holds less what is committed to orders not yet shipped, read from the store's stock, with a location sold below nought counted as nought. Products not on sale are left out. A level, never summed across days. Ask by product__product_title, variant_stock_day__sku or variant_stock_day__variant for one product's or variant's. Empty on a day the stock was not read.
Units sold a day, recentstock_daily_sales_rateUnits sold a day over the stock_sales_rate_days (28) store days ending the period's last day: the recent pace that days of cover is measured at. Merchandise units before refunds, for the variants whose stock is read.
Days of coverstock_days_of_coverDays of cover: how many days the stock available at the end of the period's last day lasts at the units sold a day over the stock_sales_rate_days (28) store days ending it. Nought when out of stock and still selling; empty where nothing sold in those days, in stock or not, or the stock was not read. Asked without a split it is the whole store's cover, which a slow seller with deep stock lengthens: ask by product__product_title, variant_stock_day__sku or variant_stock_day__variant to find what will run out. A stockout in those days, or a variant on sale for fewer, lowers the pace, so cover reads long just after a restock or a launch.
Units sold a day before running outstock_sales_rate_before_outA sales fall that coincides with a stockout: for variants out of stock at the end of the period's last day, the units they sold a day over the stock_sales_rate_days (28) store days before they ran out. Ask it beside stock_daily_sales_rate by variant_stock_day__variant or product__product_title: what sold before the stockout and stopped since is supply lost, not demand. variant_stock_day__out_of_stock_since says when each ran out. Empty for a variant in stock, and where the day before it ran out was not read.
Units sold while out of stockstock_units_sold_while_outUnits sold on days a variant began with nothing available: sold while out of stock, on a variant the store keeps selling when none is left, so they are owed to customers before stock arrives.
Variants with low coverstock_variants_low_coverVariants in stock with fewer than stock_low_cover_days (14) days of cover at the end of the period's last day: selling at a pace their stock will not keep up with. Empty on a day the stock was not read.
Variants out of stockstock_variants_out_of_stockVariants with nothing available to sell at the end of the period's last day. Empty on a day the stock was not read.
Variants that ran out of stockstock_variants_ran_outStockouts: variants whose stock reached nought in the period, each counted on the day it did, after a day read with some. Set beside units sold by day, or ask stock_sales_rate_before_out beside stock_daily_sales_rate by variant, to tell a sales fall that is a stockout from one that is lost demand. Not counted where the day before was not read.
Spend on the store's own pagesstore_page_spendLink spend sent to a page on the store's own site, in the store's currency: the denominator of destination_page_match_rate. Leaves out a host no check placed on the store's domain, a withheld path, an ad that names no page, a feed ad's link, and spend on product pages, whose paths come from the store. Split destination_spend by site_page__page for the spend a page's friction and conversion can be set beside, products included.
Cancellation reason concentration (28 days)subscription_cancellation_reason_concentrationThe share of the reasons given that the most common one carries, between 0 and 1, over the subscriptions cancelled in the 28 days to the period's last day: subscription_leading_reason_cancellations over subscription_cancellations_with_reason. Near one, a single reason explains the cancellations. Ask by subscription_cancellation_reason_day__leading_cancel_reason for which reason. A share of the reasons given, never of all cancellations: where subscription_cancellations_without_reason is the larger count, most cancellations gave none and the share says little. Empty where none gave a reason.
Cancellations with a reason (28 days)subscription_cancellations_with_reasonSubscriptions cancelled in the 28 days to the period's last day that gave a reason in the merchant's own wording: the reason on the subscription's newest cancel-flow visit. A count of 28 days whatever the period asked, never summed across days. Read it beside subscription_cancellations_without_reason, which says how many gave none.
Cancellations without a reason (28 days)subscription_cancellations_without_reasonSubscriptions cancelled in the 28 days to the period's last day with no reason in the merchant's wording: the dunning process cancelled it, the subscriber skipped the cancel flow or typed their own reason, or the store runs a V3 flow, which keeps none. A count of 28 days, never summed across days.
Subscription churn ratesubscription_churn_rateSubscriptions cancelled in the period over those that could have cancelled in it: the ones cancelled plus the ones still active on its last day, by the subscription system's own daily count. Between 0 and 1. Ask it over a window long enough for a subscription to bill once, such as 28 days. Empty for a period with no cancellation, and where the system counted no day in it. One cancelled and later reactivated is not counted.
Subscriptions keptsubscription_cohort_keptOf subscription_cohort_size, those not cancelled when the window closed. A paused subscription is kept.
Subscription cohort sizesubscription_cohort_sizeSubscriptions measured, one per 30-day window closed in the period. The denominator of subscription_retention_rate, never a count of subscriptions in its own right. Ask by subscription_cohort_cell__started_in_month and subscription_cohort_cell__months_since_start.
Sales recovered from dunningsubscription_dunning_recovered_salesNet sales, before refunds, of the renewals that ended dunnings started in the period: the one renewal each, so the least the recovery was worth, since a subscriber kept goes on to pay more.
Dunning recovery ratesubscription_dunning_recovery_rateOf the dunnings started in the period, the share a renewal has since ended. Ones still open count against it, so compare windows that have had the same time.
Dunnings lostsubscription_dunnings_lostDunnings started in the period where the dunning process has since cancelled the subscription.
Dunnings recoveredsubscription_dunnings_recoveredDunnings started in the period that a renewal has since ended. Newer ones have had less time, so compare windows that have had the same time.
Dunnings startedsubscription_dunnings_startedDunnings, on the day of their first failed charge: a subscription's run of failed charges, one however many retries, until a renewal or a cancellation ends it. Ask by subscription_dunning__outcome for how they stand: recovered, lost, cancelled or open.
Involuntary churn sharesubscription_involuntary_churn_shareOf the subscriptions cancelled in the period, the share the dunning process cancelled once its retries ran out, between 0 and 1: billing failed, and the subscriber never chose to leave. Nought where there were cancellations and none was the dunning process's. Empty for a period with no cancellation.
Cancellations for the leading reason (28 days)subscription_leading_reason_cancellationsOf subscription_cancellations_with_reason, those that gave the most common reason. Ask by subscription_cancellation_reason_day__leading_cancel_reason for which reason that is. A count of 28 days, never summed across days.
Net subscription flowsubscription_net_flowSubscriptions started in the period less subscriptions cancelled in it: above nought the base grew, below it the base shrank. Both sides count subscriptions, so someone who starts two counts twice. Empty for a period with neither. One cancelled and later reactivated is not counted as cancelled.
Subscription payment failuressubscription_payment_failuresFailed charges on subscriptions, each retry counted. A rise with no rise in subscriptions_with_failed_payment is the store's retry settings, not more failures. Empty in a period with none, so an empty answer does not say a store has a subscription system: subscriptions_in_dunning does, and it also shows a system that names a failed charge differently.
Subscription retention ratesubscription_retention_rateThe share of a subscription cohort not cancelled by the close of a 30-day window since each subscription started. Ask by subscription_cohort_cell__started_in_month and subscription_cohort_cell__months_since_start, comparing the same window across cohorts. Not retention_rate, which counts customers coming back after a first order. A window is only here once it has closed, by the subscription source's newest change, so a read that stopped closes no more.
Subscriptions activesubscriptions_activeSubscriptions active on the period's last day, by the subscription system's own daily count. A population, never summed across days. Empty on a store with no subscription system. The day is the system's own, in a timezone Skio does not document.
Subscriptions cancelledsubscriptions_cancelledSubscriptions cancelled in the period, on the store's day each cancelled. Ask by subscription__is_passive_churn to tell the dunning process's cancellations from the subscriber's. One cancelled and later reactivated is not counted.
Subscriptions in dunningsubscriptions_in_dunningSubscriptions in dunning on the period's last day: a charge failed and neither a renewal nor a cancellation has ended it, by the subscription system's own daily count. A population, never summed across days. Set it beside subscriptions_with_failed_payment: subscriptions in dunning with no failed charges mean the store's system names a failed charge differently, not that nothing failed.
Subscriptions pausedsubscriptions_pausedSubscriptions paused on the period's last day, by the subscription system's own daily count. A population, never summed across days. Empty on a store with no subscription system.
Subscriptions startedsubscriptions_startedSubscriptions started in the period, on the store's day each started. Empty on a store with no subscription system.
Subscriptions with a failed paymentsubscriptions_with_failed_paymentDistinct subscriptions with a failed charge in the period: how many are failing, whatever the retries. Nought also on a store with no subscription system, where subscriptions_in_dunning is empty, and on one whose system names a failed charge differently, where subscriptions_in_dunning has a count: read it first.
Support backlogsupport_backlogSupport contacts open at the end of the last day of the period asked for. Split by metric_time__day it is the backlog day by day.
Support contactssupport_contactsSupport contacts customers opened: tickets that are neither spam nor trashed, and not started by the merchant.
Contacts a person replied tosupport_contacts_human_respondedContacts a person has replied to.
Support contacts per 100 orderssupport_contacts_per_100_ordersSupport contacts opened per 100 orders the store took, over the same days. Both are counted on the store's day.
Contacts with a resolution timesupport_contacts_resolution_timedContacts closed now whose close is not before they opened. A ticket the helpdesk stamps as created after its close is resolved, but its resolution has no length.
Contacts resolvedsupport_contacts_resolvedContacts closed now.
Contacts replied tosupport_contacts_respondedContacts that have had any reply from the merchant's side, automation included.
First human response hours (total)support_first_human_response_hours_totalHours to the first reply from a person, summed over the contacts a person replied to. Divide by those contacts for the average.
First response hours (total)support_first_response_hours_totalHours to the first reply from the merchant's side, summed over the contacts that had one. Divide by contacts replied to for the average.
Satisfaction score totalsupport_rating_score_totalThe satisfaction scores given, 1 to 5, summed. Divide by scores given for the average.
Satisfaction scoressupport_ratingsSatisfaction scores customers gave, on the day each was given.
Resolution hours (total)support_resolution_hours_totalHours from opening to the latest close, summed over the contacts closed now whose close is not before they opened. Divide by contacts with a resolution time for the average.
Top three ads' share of spendtop_three_ad_spend_shareThe share of an ad set's week its three biggest ads carried, between 0 and 1. Asked by ad_week__week_ad_set: near one with many ads live, the rest are barely being shown. An ad set of three ads or fewer is always one.
Conversion rate by source (web analytics)traffic_source_conversion_rateA source's conversion rate as web analytics measures it: purchases in the sessions it sent over those sessions, summed over the period. Ask it split by web_source_day__traffic_source, and read it beside the source's sessions.
Purchases by source (web analytics)traffic_source_purchasesPurchases the tagging reported in sessions that arrived from a source. The tagging's count, never the store's orders.
Sessions by sourcetraffic_source_sessionsSessions that arrived from a source and medium, by web_source_day__traffic_source. For the site's sessions ask web_sessions, which is reconciled against the source's own total; this report is not.
Ad spend (Triple Whale's reading)tw_ad_spendAd spend as Triple Whale reads it from every ad platform, in the store's currency. Triple Whale's reading, beside the platforms' own ad_spend and never added to it: for a platform connected directly it is the same money read a second time. It also holds a channel whose platform is not connected. Ask by channel_spend_day__spend_channel for one channel's.
Blended ROAS (Triple Whale)tw_blended_roasTriple Whale's blended return on ad spend: the revenue of every order as Triple Whale counts it, over the ad spend Triple Whale reads from every platform, summed over the period. A multiple, so 3 is three of revenue for one of spend. Triple Whale calls it Blended ROAS; it is MER as Triple Whale reckons it. Triple Whale's opinion on both sides, beside the store's own figures and never in place of them: its revenue is not net_sales and its spend is not ad_spend. The revenue is tw_linear_all_revenue over every channel, paid or not, so it has no channel split. Empty for a period Triple Whale read no spend in.
New-customer orders (First Click, Triple Whale)tw_first_click_new_customer_ordersTriple Whale's first orders credited to each channel under first click: each order to the first channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Orders (First Click, Triple Whale)tw_first_click_ordersTriple Whale's orders credited to each channel under first click: each order to the first channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Revenue (First Click, Triple Whale)tw_first_click_revenueTriple Whale's order revenue, in the store's currency credited to each channel under first click: each order to the first channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
New-customer orders (Last Click, Triple Whale)tw_last_click_new_customer_ordersTriple Whale's first orders credited to each channel under last click: each order to the last channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Orders (Last Click, Triple Whale)tw_last_click_ordersTriple Whale's orders credited to each channel under last click: each order to the last channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Revenue (Last Click, Triple Whale)tw_last_click_revenueTriple Whale's order revenue, in the store's currency credited to each channel under last click: each order to the last channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
New-customer orders (Linear All, Triple Whale)tw_linear_all_new_customer_ordersTriple Whale's first orders credited to each channel under linear: each order shared equally across every channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Orders (Linear All, Triple Whale)tw_linear_all_ordersTriple Whale's orders credited to each channel under linear: each order shared equally across every channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
Revenue (Linear All, Triple Whale)tw_linear_all_revenueTriple Whale's order revenue, in the store's currency credited to each channel under linear: each order shared equally across every channel touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels add up to the store's total.
New-customer orders (Linear Paid, Triple Whale)tw_linear_paid_new_customer_ordersTriple Whale's first orders credited to each channel under linear over paid touches: each order shared across the paid channels touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels never count an order twice, but an order with no paid touch is left out, so they can fall short of the store's total.
Orders (Linear Paid, Triple Whale)tw_linear_paid_ordersTriple Whale's orders credited to each channel under linear over paid touches: each order shared across the paid channels touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels never count an order twice, but an order with no paid touch is left out, so they can fall short of the store's total.
Revenue (Linear Paid, Triple Whale)tw_linear_paid_revenueTriple Whale's order revenue, in the store's currency credited to each channel under linear over paid touches: each order shared across the paid channels touched. Triple Whale's opinion, beside the store's own attribution and never added to it. Its channels never count an order twice, but an order with no paid touch is left out, so they can fall short of the store's total.
New-customer orders (Total Impact, Triple Whale)tw_total_impact_new_customer_ordersTriple Whale's first orders credited to each channel under Total Impact, Triple Whale's modelled incrementality, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Orders (Total Impact, Triple Whale)tw_total_impact_ordersTriple Whale's orders credited to each channel under Total Impact, Triple Whale's modelled incrementality, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Revenue (Total Impact, Triple Whale)tw_total_impact_revenueTriple Whale's order revenue, in the store's currency credited to each channel under Total Impact, Triple Whale's modelled incrementality, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
New-customer orders (Triple Attribution, Triple Whale)tw_triple_new_customer_ordersTriple Whale's first orders credited to each channel under Triple Attribution, Triple Whale's own model, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Orders (Triple Attribution, Triple Whale)tw_triple_ordersTriple Whale's orders credited to each channel under Triple Attribution, Triple Whale's own model, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Revenue (Triple Attribution, Triple Whale)tw_triple_revenueTriple Whale's order revenue, in the store's currency credited to each channel under Triple Attribution, Triple Whale's own model, which credits several channels for one order. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
New-customer orders (Triple Attribution + Views, Triple Whale)tw_triple_views_new_customer_ordersTriple Whale's first orders credited to each channel under Triple Attribution counting ad views as well as clicks. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Orders (Triple Attribution + Views, Triple Whale)tw_triple_views_ordersTriple Whale's orders credited to each channel under Triple Attribution counting ad views as well as clicks. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Revenue (Triple Attribution + Views, Triple Whale)tw_triple_views_revenueTriple Whale's order revenue, in the store's currency credited to each channel under Triple Attribution counting ad views as well as clicks. Triple Whale's opinion, beside the store's own attribution and never added to it. Not additive across channels: always split it by channel_attribution_day__channel, and never total it.
Spend on unchecked pagesunchecked_destination_spendSpend sent to a page on a day no check could say whether it answered: a bot check, a refusal, a feed ad's link, a page not checked, a product the store does not hold, or spend with no known page. Never counted as dead.
Unclassified customersunclassified_customersOrders that look like a customer's first where nothing can confirm it: no lifetime order count from the platform, and the order falls on the first day of retained history. Counting these as new would inflate acquisition and deflate every cost per acquisition built on it.
UnitsunitsItems sold, before refunds. Counts merchandise and non-merchandise lines alike unless you filter by is_merchandise.
Units per orderunits_per_orderProduct units per order. With average_unit_price it tells which half of an order value movement it was: customers buying more items, or buying dearer ones.
Campaign match rate (web analytics credit)web_campaign_match_rateThe share of a platform's credited purchases that were matched to one of its campaigns, between 0 and 1. Read before any web-credited cost by campaign: a low share means the campaigns' links were tagged with something other than their names, and the campaign figures cover only part of the platform. Asked split by platform.
Campaign spend match rate (web analytics credit)web_campaign_spend_match_rateThe share of a platform's spend on campaigns web analytics had credited by name, between 0 and 1: read beside web_campaign_match_rate, so a platform whose big campaigns are never matched is not read by campaign on the strength of its small ones. Asked split by platform.
Ad spend (days web analytics was read)web_credited_ad_spendAd spend on the days web analytics' credit is known for too, in the store's currency. The numerator of the web-credited costs; not total spend, which is ad_spend.
CAC (web analytics credit)web_credited_cacAd spend over the new customers web analytics credits to that platform: an acquisition cost by Google Analytics' attribution model, data-driven unless the property was set otherwise, beside blended_cac and the platforms' own claims. Asked split by platform, and by campaign where the match rates hold. Summed spend over summed credit for the period, never an average of days. Google Analytics sees no view-through and only the purchases its tagging caught, so it under-credits a platform that works through views, Meta most of all; and a credited purchase with no matching store order counts as no new customer, so the cost reads high where web_order_match_rate is low.
Cost per purchase (web analytics credit)web_credited_cost_per_purchaseAd spend over the purchases web analytics credits to that platform: what a purchase cost by Google Analytics' attribution model, data-driven unless the property was set otherwise, rather than the platform's own claim. Asked split by platform, and by campaign where the match rates hold. Google Analytics sees no view-through and only the purchases its tagging caught, so it under-credits a platform that works through views, Meta most of all, and this cost reads high there.
Credited purchases matched to a campaign (web analytics)web_credited_matched_purchasesCredited purchases matched to one of the platform's campaigns: the numerator of web_campaign_match_rate.
Spend on campaigns matched to credit (web analytics)web_credited_matched_spendSpend on campaigns web analytics had credited by name on or before the day: the numerator of web_campaign_spend_match_rate.
New customers credited to the platform (web analytics)web_credited_paid_new_customersThe same credit, over the purchases that won their customer. Asked split by platform.
Purchases credited to the platform (web analytics)web_credited_paid_purchasesPurchases the web analytics source's attribution model credits to a connected ad platform: data-driven on Google Analytics unless the property was set otherwise. Fractional where it shares one purchase. Asked split by platform.
Revenue credited to the platform (web analytics)web_credited_paid_revenueRevenue the web analytics source credits to a connected ad platform, in the store's currency. Asked split by platform.
Credited purchases (web analytics)web_credited_purchasesPurchases as the web analytics source's attribution model credits them, by source, channel or campaign. Data-driven on Google Analytics unless the property was set otherwise. The source's opinion, beside the store's own attribution and never added to it.
Return on ad spend (web analytics credit)web_credited_return_on_spendRevenue Google Analytics' attribution model, data-driven unless the property was set otherwise, credits to a platform, over its spend. Asked split by platform. Google Analytics sees no view-through and only the purchases its tagging caught, so this return reads low for a platform that works through views, Meta most of all. Never set against the platform's own return as though they measured the same thing.
Credited revenue (web analytics)web_credited_revenuePurchase revenue as that attribution model credits it, in the store's currency.
Engaged sessionsweb_engaged_sessionsSessions that lasted over ten seconds, had a key event, or had two or more page views.
Tracked purchases linked to an orderweb_linked_purchasesTracked purchases whose transaction ID matched a store order. Set against orders it says how much of the store's trade web analytics saw.
Store orders web analytics sawweb_order_match_rateTracked purchases linked to a store order over the store's orders: how much of the store's trade web analytics saw, over the period. Never 100%, because shoppers decline tracking and orders are taken other ways; what matters is a fall, which means a tag broke, consent changed, or the purchase stopped carrying the order's reference.
Purchase revenue (web analytics)web_purchase_revenueRevenue the site's tagging reported with those purchases, in the store's currency. The tagging's figure, never the store's revenue.
Purchases (web analytics)web_purchasesPurchases the site's tagging reported to web analytics. Fewer than the store's orders wherever tracking was declined or blocked, and never a count of orders.
Sessionsweb_sessionsSite sessions, as the connected web analytics source counts them. Known only for the days read; a day before that is unknown, not zero.
Tracked purchasesweb_tracked_purchasesPurchases the site's tagging reported, each transaction counted once, by the source, channel, campaign or landing page of the session it happened in.
Share of tracked purchases with no channel or sourceweb_unattributed_purchase_shareThe share of tracked purchases whose session could not be credited to any channel or source, over the period, never an average of days. When it rises, every channel figure from web analytics is suspect: the tagging is losing track of where visitors came from.
Tracked purchases with no channel or sourceweb_unattributed_purchasesTracked purchases whose session the web analytics source could not credit: a channel of Unassigned or (not set), or a source and medium of (not set). Read as a share of web_tracked_purchases.
Weekly ad spendweek_ad_spendSpend through an ad set's ads in a whole week, in the store's currency.
Weekly frequency (Meta)week_frequencyHow many times each person reached saw an ad set or ad in a whole week: impressions over reach. Asked for one ad set or ad (ad_week_reach_row__reach_entity) and one week; a frequency rising at the same spend while cost per result rises is an ad wearing out.
Weekly impressions (Meta, weekly read)week_impressionsImpressions of an ad set or ad in a whole week, from the weekly read.
Leading ad's weekly spendweek_lead_ad_spendThe spend of the ad that led its ad set that week.
Weekly reach (Meta)week_reachPeople an ad set or ad reached in a whole week. Not additive across ad sets, ads or weeks.
Top three ads' weekly spendweek_top_three_ad_spendThe spend of the three ads that spent the most in their ad set that week.