Attribution

Multi-touch attribution built on your own click data

ROASaurus records every ad click on your domain and credits each order to the clicks that led to it. Compare last click, first click, linear, data-driven and Total Impact models on the same orders, with spend synced from your ad accounts.

Twelve models

From last click to data-driven Markov and Shapley models, Total Impact and Clicks & views, switchable on the same data.

Your own click data

Every landing with a click id or UTM is stored as a touch, collected on your domain.

Survey-blended credit

A post-purchase survey adds the channels that clicks do not show.

Order journeys

Every click and storefront event before each purchase, with its credit.

Campaign profit

Contribution margin, POAS and LTV:CAC per campaign under any model.

Spend joined

ROAS and CPA from campaign spend synced from 13 ad platforms, converted to your store currency.

How attribution works

Every visit that lands with a click id or UTM is stored as a touch for that visitor. When an order comes in from your store, ROASaurus takes the visitor it was matched to and credits the touches inside your lookback window. Shares of one order always add up to one. If the visitor has no browser touch, for example because of an ad blocker or a missing script, ROASaurus falls back to the click ids and UTMs in the order's landing URL. Orders with no touch at all count as direct.

The models

ModelHow credit is splitWhat it shows
Last clickAll credit to the most recent touch in the windowWhich ads close sales
First clickAll credit to the first touch in the windowWhich channels bring new people in
LinearSplit evenly across all touches in the windowEvery channel that played a part
Total ImpactThe survey answer gets 50% and the clicks share the other 50% evenly. With only a survey answer, it gets 100%. With no answer, the clicks share it evenly.Channels that clicks miss, such as podcasts, word of mouth or ads people saw but did not click

Eight more: last non-direct click (organic search and referral visits count too), linear paid (only paid ad clicks), position-based (40% first, 20% between, 40% last), time decay (recent clicks count more), data-driven Markov chain and Shapley value (trained on your store's own journeys, including clicks that led to no order; with too little data they split credit linearly and say so), any click (every clicked channel gets the whole order, so its rows add up to more than the total), and Clicks & views (the clicks plus the purchases the ad platforms report after an ad view without a click, modelled onto orders and labelled as an estimate).

Total Impact uses a post-purchase survey, "How did you hear about us?", on Shopify's Thank-You page. Answers such as Facebook or Instagram, TikTok, Google search, YouTube, a friend, email or a podcast are mapped to channels. Other platforms can send answers to the surveys API.

Lookback window

Each store has its own window, from 1 to 90 days, with 7 days as the default, or an unlimited lookback that counts every click since the visitor's first visit. The report switches between these, so you can see how much credit moves as the window grows. Total Impact has its own survey lookback, and the attribution and campaign profit reports can put sales on the day of the credited click instead of the order date.

What the reports show

  • Channels and campaigns: orders, revenue, new customers, ad spend, ROAS and CPA under the chosen model.
  • Orders and journeys: each order with its last click, the number of ad clicks and how it was matched. The journey page lists every click and storefront event before the purchase, with the credit each model gives it.
  • Campaign profit: contribution margin, profit after ads, POAS, CAC, the 90-day LTV of the customers each campaign acquired, and LTV:CAC. See profit analytics and lifetime value.

Why the numbers differ from Ads Manager

Each ad platform reports conversions with its own windows, and Ads Manager also counts view-through conversions: people who saw an ad and bought later without clicking. Every platform claims the sale for itself, so platform totals added together are usually higher than your real order count. ROASaurus counts orders it can tie to a click on your site, plus survey answers in Total Impact and, if you choose, modelled ad views, and gives each order one unit of credit. That makes it a fair way to compare channels against each other. More in how attribution works and what ROAS means.

Who it is for and setup

Brands that advertise on more than one channel and need one count of orders to compare them by. Attribution runs on the click data from server-side tracking, so there is nothing extra to install. For campaign-level ROAS on Meta, set the ad URL parameters to utm_source=facebook&utm_medium=paid&utm_campaign={{campaign.name}}&utm_id={{campaign.id}}&utm_term={{adset.name}}&utm_content={{ad.name}} and connect the ad account for spend (Meta, TikTok). For Total Impact, enable the post-purchase survey.

Limits

  • Spend, and so ROAS and CPA, syncs from Meta, TikTok, Google Ads, Microsoft Advertising, Pinterest, Snapchat, Reddit, X, LinkedIn, Amazon Ads, AppLovin, Criteo and The Trade Desk once you connect the ad account. Taboola and offline spend are entered by hand.
  • View-through conversions count only under Clicks & views, or with "Clicks + modelled views" switched on, and they are modelled from what the ad platforms report, not tracked.
  • The data-driven models need at least 30 orders with clicks from two channels in 90 days. Below that, they split credit linearly.
  • Spend in another currency is converted to the store currency at ECB rates.
More features

Built on the same first-party data

Server-side trackingFirst-party click capture, order matching and Conversions APIs for 15 ad platforms, deduplicated with the browser pixel.Profit analyticsNet profit after every cost, POAS and break-even ROAS, a full P&L and profit per product, order and channel.Lifetime valueCohorts, CAC payback, predicted LTV up to 24 months, repurchase timing and RFM segments synced to Klaviyo.Ask AI & MCPAn assistant on Claude that answers from your own reports, and an MCP server with read-only tools for Claude Code and other clients.Creative analyticsHook rate, hold rate, fatigue and ROAS per creative, hook, creator and format for Meta and TikTok.Marketing mix & budgetsA marketing mix model with a budget optimiser, geo-lift and on/off tests, spend curves and daily scale, hold or cut advice per campaign.AudiencesAudiences from orders, predicted LTV, RFM, visitors and leads, kept in sync with ten ad platforms and Klaviyo.Actions & AutopilotAn ads manager for nine platforms with an approval inbox, rules with guardrails, an AI Autopilot and one-click rollback.AI briefingsAnomaly alerts, AI digests grounded in your own numbers, a weekly action plan with a decision log and scheduled questions in Slack.Competitors & AI visibilityCompetitors' Meta ads and emails, how often AI assistants name your store, AI content drafts and AI focus groups.InfluencersAn influencer CRM with deals, code and link attribution, and ROI and profit per deal and influencer.AgenciesClient workspaces, a multi-client overview, white label on your own domain, PDF reports and a revenue share.
FAQ

Questions, answered

What is multi-touch attribution?

A way of crediting one order to more than one marketing touch. In ROASaurus, the linear model splits each order evenly over every ad click in the lookback window, and Total Impact blends those clicks with what the customer said in a post-purchase survey.

Why does ROASaurus show fewer sales than Meta Ads Manager?

Ads Manager counts view-through conversions and uses its own attribution windows, and each platform claims the sale for itself. ROASaurus counts orders it can tie to a click on your site, and each order is credited once across all channels. Only the Clicks & views model adds the views the platforms report, as a modelled estimate.

Can I change the lookback window?

Yes. Each store has a window from 1 to 90 days, 7 by default, or an unlimited lookback. The attribution report also switches between 1, 7, 14, 28, 30, 60 and 90 days or unlimited.

Do I need the post-purchase survey?

Only for Total Impact. The other models work from click data alone. Without survey answers, Total Impact splits credit evenly across the clicks.

Every sale, traced to the ad that earned it

Server-side tracking, profit and lifetime value in one app. 14-day free trial, no credit card.

Start your free trial →