Cross-Channel Attribution Tools for Ecommerce: A Shortlist

Choose the measurement family before the tool: multi-touch attribution assigns credit to touchpoints, marketing mix modelling relates aggregate spend to sales, and incrementality testing compares people or regions that saw ads with those that did not. Of the three, only incrementality testing starts from a control group, and in 15 U.S. advertising experiments at Facebook the observational methods often failed to produce the same effects as the randomised experiments.

What are the three measurement families?

Multi-touch attribution splits credit across touchpoints. Google's Analytics Help defines attribution as the act of assigning credit for important user actions to different ads, clicks and factors along the user's path, and says a model can be a rule, a set of rules or a data-driven algorithm. GA4's attribution reports offer three models: data-driven, paid and organic last click, and Google paid channels last click. First click, linear, time decay and position-based were removed in November 2023, and direct visits get no credit unless the whole path was direct. Attribution's strength is detail, since credit goes to individual ads, clicks and factors on a path, but nothing in that definition involves a group that saw no ad.

Marketing mix modelling fits aggregate data. Google's Meridian documentation says you provide a target KPI, often sales, media data for each channel and control variables, and that the model is designed to estimate the true causal impact of marketing, which states intent, not a test result. Meta's Robyn page says it needs no personal data or individual log-level data, does not depend on cookies or pixel data, and calibrates models against ground-truth methods such as geo-based tests and Facebook lift studies.

Incrementality testing uses a control group. Google Ads Help says Conversion Lift splits an audience into people who see your ads and people who do not, and the difference in conversions is the lift caused by the ads. Tests can be based on users or on geography, and Conversion Lift is not available for all Google Ads accounts.

How strong is the evidence for each family?

The pages above, written by platforms that also sell advertising, say what each method is built to do, not whether it works on your store. The Facebook study in the opening paragraph used 500 million user-experiment observations, and its observational methods still often failed to match the experiments even after conditioning on extensive demographic and behavioral variables (Marketing Science, 2019). Lewis and Rao report on twenty-five large field experiments with major U.S. retailers and brokerages: the median confidence interval on return on investment was over 100 percentage points wide, and informative experiments can easily require more than 10 million person-weeks, so experiments can be costly and, for many firms, possibly infeasible (Quarterly Journal of Economics, 2015).

Both studies are peer reviewed, and neither abstract names an attribution vendor. No source read here compares the three families on an ecommerce store or reports how accurate marketing mix models are against experiments. Ask any vendor what its tool compares against and which test shows it works.

Which tools sit in which family?

Prices are as each vendor's page showed them on the date given, and they change without notice. An entry with no link has no dated price source, so it has no price.

Multi-touch attribution

• Paid tiers are priced on annual GMV, and its pricing page blocks automated reads, so no paid price is listed.
• Northbeam: machine-learning multi-touch attribution. Its pricing page lists $1,500 for Starter and $3,500 per month for Professional, with Growth and Enterprise by quote (as read on 2026-09-08).
• Polar Analytics: an analytics platform with deterministic multi-touch attribution on its own pixel. Its Shopify App Store listing shows Core from $750 a month, priced on online GMV (as read on 2026-09-26).

Marketing mix modelling

• Fospha: a hybrid of impression-based marketing mix modelling and multi-touch attribution. Its pricing page lists Lite at $1,500 a month for $100k to $500k of monthly media spend, and Pro at $2,000 a month plus a share of media spend (as read on 2026-09-08).
• Prescient AI: machine-learning-based media mix modelling, quoted through a demo. Its pricing page shows no price, says the platform is built for brands spending $100K+ a month on advertising, and lists annual contracts (as read on 2026-09-26).

Incrementality testing

• Haus: geo-lift experiments, with Causal Attribution and Causal MMM calibrated on them. Its pricing page shows four plans, each starting with a demo, and no price (as read on 2026-09-26).

More than one family

• Rockerbox: multi-touch attribution, MMM and control-group incrementality testing, sold to enterprises, with pricing not publicly listed.

From a GA4 export

• Causality Engine: reads a GA4 export and shows what each GA4 channel group caused, with Direct split back to the channels that sent those buyers, next to last-click rebuilt from the same export. It works on channel groups, not campaigns, keywords or creatives, and needs no pixel or SDK. A read costs €99 once, returns in 1 to 2 minutes and is refundable within 30 days; Pro is €299 a month, billed monthly; prices exclude VAT. Direct connections to Shopify, Meta, Google or TikTok are roadmap, not live.

How can I test the numbers before paying for a tool?

1. In Shopify, open Orders, click Export and choose Orders by date for one full month. Count distinct order numbers in the Name column, because Shopify's CSV puts extra line items of an order on separate lines.
2. In each ad platform, export purchases for the same dates and time zone, and note the attribution setting the report uses.
3. Add the platform totals in a spreadsheet. Pass: the total is at or below Shopify's order count. Fail: it is higher, so the platforms claim more purchases than there are orders and some orders are counted more than once. A pass shows only that overlap is not visible in the totals, not what any channel caused.
4. In GA4, click Advertising, go to Attribution, then Attribution models, choose your purchase key event and use the drop-downs in the Attribution model (non-direct) columns to compare paid and organic last click with data-driven. Pass: each channel's credit barely moves between the two. Fail: a channel's credit moves a lot, so its number reflects the method, and neither column says what the channel caused.

Sources, 6 October 2026: Get started with attribution (Google Analytics Help); Key event attribution models report (Google Analytics Help); About Conversion Lift (Google Ads Help); Introduction to Meridian (Google for Developers); Robyn (Meta Marketing Science); A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook (Marketing Science, 2019); The Unfavorable Economics of Measuring the Returns to Advertising (Quarterly Journal of Economics, 2015); Exporting orders (Shopify Help Center); vendor pages, each read as the date in its entry says: Northbeam, Polar Analytics, Fospha, Prescient AI, Haus.

Related answers

• Causal attribution
• Incrementality testing
• Multi-touch attribution
• Marketing mix modeling
• Data-driven attribution