Skip to content

Attribution

Triple Whale for ecommerce stores

Triple Whale pulls spend from the ad platforms, orders from the store and its own visitor data into one place, then offers several ways of assigning credit for a sale. It is popular with direct to consumer brands because it answers the question an operator asks each morning, which is whether yesterday paid for itself. We use it for direction and speed, and we keep the store's own order data as the record that settles any disagreement.

Triple Whale

Attribution · by Triple Whale. A text wordmark, not a logo: we are not affiliated with the vendor and make no claim to its trademarks.

What we do with it

The work inside Triple Whale

Blended numbers as the primary view

Total spend against total revenue for the same period cannot be double counted by anyone. We set the blended ratio as the headline the team steers by, add contribution after cost of goods and shipping where the data supports it, and treat platform-level attribution as the explanation underneath rather than the target.

One attribution model, used consistently

Several models are available, and switching between them changes which channel looks successful. We choose the model that matches how the brand actually buys media, document that choice, and compare periods on the same basis, because a comparison across two models tells you nothing about performance.

Post-purchase survey as an independent read

Asking customers where they heard about the brand gives a second signal that no pixel can produce, particularly for channels that influence without a click, such as podcasts, creators and word of mouth. Response rates and self-reporting mean it is directional, and it is most useful where it disagrees with the tracked data.

Creative reporting the paid team can act on

Performance grouped by creative, angle and format is what changes next week's production plan. We set the naming conventions and the reporting groups so the question is which concepts to make more of, rather than which video identifiers performed best last month.

Cohorts and payback alongside the daily view

Daily numbers drive tactical decisions and cannot answer whether acquisition is affordable. Cohort views by acquisition month, repeat purchase timing and the time it takes to recover acquisition cost put the daily ratio in context, especially for brands with a genuine second purchase pattern.

When it fits

Triple Whale is the right tool when

  • Brands spending across several platforms at once

    When paid social, search, shopping and creators all run together, every platform claims the same orders. A blended view is the only place the total makes sense.

  • Operators who make decisions daily

    Teams adjusting budgets each morning need one dashboard with spend, orders and margin in view. That workflow is what these tools are built around.

  • Stores where creative volume drives results

    If the main lever is how much creative gets produced and tested, reporting that groups performance by concept rather than by ad identifier is worth the subscription on its own.

What to watch

Where accounts drift

Facts about defaults and costs, not criticism. These are the settings we check on every account.

  • Attribution is an estimate, presented precisely

    Any modelled figure is an informed allocation, not a measurement. Reading a return figure to two decimal places invites more confidence than the underlying data supports, so decisions should hinge on clear differences rather than small ones.

  • Tracked coverage depends on consent and blockers

    Visitor-level data is affected by consent choices, browser restrictions and privacy tools, and the share that goes unobserved varies by market and audience. That is why the blended view remains the anchor for spending decisions.

  • It does not replace the accounting record

    Revenue reported for finance comes from the store platform and its payment records. An analytics platform is for allocation and speed, and reconciling the two once a month keeps both conversations grounded.

  • Subscription cost has to earn out

    Pricing scales with order volume and features, so the value has to show up in decisions actually made differently. A store with two channels and a weekly reporting rhythm may get the same answers from its existing analytics.

Questions

Triple Whale questions

By CartKernel · Last reviewed

Will an attribution platform make the ad platforms agree with each other?

No, and nothing will. Each platform counts conversions on its own rules and windows, which is why their totals overlap. What a shared view gives you is one consistent basis for comparing channels, plus a visible gap between what platforms claim and what the store recorded.

Should budget decisions use these numbers or the analytics property?

Use the blended ratio for how much to spend in total, and a consistent model for how to split it between channels. Analytics remains the place for on-site behaviour and landing page performance. Problems arise when a team switches sources depending on which one looks better.

Is post-purchase survey data reliable enough to act on?

It is directional. People misremember, and only a portion answer. It is still the only signal available for channels that create demand without a trackable click, so we read it as a trend over months and use it to challenge tracked data rather than to replace it.

Do we still need a properly configured analytics property?

Yes. Site behaviour, landing pages, search performance and funnel analysis all live there, and it is the free, platform-independent record that survives a change of vendor. The two answer different questions and a store running paid media at scale usually wants both.

How do you make budget decisions when attribution is imperfect?

Set the total by what the blended ratio and margin can support, allocate within it using one consistent model, and test changes at a size where the effect is visible in total revenue. Holdouts and geographic tests answer incrementality questions that no dashboard can settle.

Works alongside

All technology
AnalyticsGA4GA4 for ecommerce stores: complete purchase events, property settings that fit a store, channel groups set correctly and revenue reconciled to orders monthly.OpenShopify trackingElevarElevar for ecommerce stores: a Shopify data layer built to spec, checkout purchases captured server side, destinations monitored and matched to orders.OpenServer-side taggingStapeStape for ecommerce stores: hosted server-side tagging on a custom domain, Shopify webhook purchases, consent passed through and request volume monitored.OpenPaid socialMeta AdsMeta Ads agency for ecommerce brands: pixel and Conversions API events, a feed that matches the site, creative on a schedule, results reconciled to orders.OpenData warehouseBigQueryBigQuery for ecommerce stores: the GA4 export joined to orders and ad cost, cohort and repeat purchase tables, queries built to keep scan costs sane.OpenComparisonGA4 vs third-party attribution toolsGA4 vs third-party attribution tools: what each one measures, where the extra spend earns its place, and what has to be true before either can be trusted.OpenAnswerWhich attribution model should an ecommerce store use?No attribution model is correct. Use one consistently for reporting, platform data for bidding, and holdout tests for budget. How to build that stack.OpenAnswerHow do you measure incrementality for ecommerce ads?Incrementality is measured by withholding advertising and comparing, not by attribution. The test designs that work for a store and how to size them.OpenAnswerWhy does Meta report more purchases than my store?Meta counts purchases against the ad interaction inside an attribution window and can double count undeduplicated events. The causes, ranked, and the fixes.OpenAnswerHow do you track new versus returning customer revenue?Track new versus returning revenue from the order record, not from analytics sessions. How to define a new customer and the joins that break the split.Open

Find the leak.

A free Growth Analysis ranks what your store should fix first, by revenue at stake.