Blog
Attribution: why Ads, Meta and your shop disagree
Google Ads reports 12,000 euros in conversions this month. Meta reports 9,000. Your shop backend says you took in 15,000. If the platforms are right, you sold 21,000. Nobody is lying, and yet at least two of the three numbers are wrong for budget decisions. This is the attribution problem, and it is structural, not a bug. Here is why every ad platform overreports, what their numbers actually mean, and how to get back to one neutral baseline.
David Hildering · Published
Three dashboards, three different revenues
Ask any shop owner who runs ads on more than one platform, and you will hear the same story: Google Ads and Meta together claim more revenue than the shop actually took in. Performance agencies see it with every client. The numbers are not invented, they are measured, and they still do not add up.
The reason is simple and rarely spelled out: each platform counts a sale as “its” conversion whenever the sale happens within its own attribution window after someone interacted with one of its ads. Neither platform can see the other’s data. Both take full credit. If a customer saw a Meta ad on Monday, clicked a Google ad on Thursday and ordered on Friday, both platforms report that order, and so does your shop. One sale, three reports.
How each platform counts
The counting rules are documented, and they explain most of the gap:
- Google Ads counts a conversion when it happens within the conversion window after an ad interaction, 30 days after a click by default. Depending on your settings, it may count every conversion in that window or only the first one, and modeled conversions fill in what direct tracking misses.
- Meta uses a 7-day click or 1-day view window by default. The view-through part is the quiet inflator: a sale is reported even if the person never clicked an ad, as long as they saw one in the window. Nobody clicked anything, and the sale still appears in Meta’s results.
- Analytics tools such as GA4 work differently again: they look at the whole journey and split the credit across channels, so a campaign can show 100 conversions in Google Ads and a fraction of that in GA4, without either being broken.
Add the different time zones, deduplication rules and consent states, and it becomes obvious: these systems were never designed to agree. Each answers its own question. Only one of them answers yours: what did my shop actually sell?
What inflated numbers do to your decisions
The real cost is not cosmetic. Budget decisions built on platform-reported revenue systematically favor the wrong things:
- ROAS looks better than it is. Return on ad spend calculated from platform-reported revenue is inflated by every sale the platform merely touched. Campaigns get scaled on numbers the bank account never sees.
- Claimed revenue exceeds 100 percent. Add up what Google Ads, Meta and Microsoft Ads claim, and the total regularly exceeds your actual revenue. Every platform reports a share larger than its true, unmeasurable contribution.
- Organic and newsletter get starved. When platforms claim every sale within their windows, the customers who would have bought anyway are counted as ad-driven. Channels without a reporting self-interest lose budget they earned.
None of this means ads do not work. It means platform-reported revenue is a claim, not a measurement, and claims should not be the basis for budget allocation.
The neutral baseline: measure what actually happened
Two sources deserve to be called neutral: your order backend, which knows exactly how much money arrived, and first-party analytics on your own site, which records visits and orders as they happen without modeling them toward any channel.
This is the approach MetriXs takes. Orders are recorded as real events on your site (or sent server-side from your shop backend), so reported revenue is the sum of orders that actually happened, not a modeled claim. Revenue is shown in your currency, per source, per campaign and per landing page, so you can see which channels visitors really arrived through. Visitors who land with a Google Ads, Meta or Microsoft Ads click ID are labeled accordingly in a dedicated ad view of the Sources panel, without any conversion being exported back to the platforms and without modeled fills. The measurement stays on your side.
That gives you a fixed reference point: the platforms’ claimed revenue on one side, the orders your shop actually recorded on the other, and the real arrival channels in between.
A practical workflow for budget decisions
You do not have to stop reading platform dashboards. They are useful for optimizing inside a platform. For budget decisions across channels, use a simple split:
- Inside a platform: its own numbers are fine for comparing its own campaigns against each other, since the same inflation applies to all of them.
- Across channels: use what your shop and your first-party analytics recorded. Compare revenue by source and by campaign week by week, not platform claims side by side.
- Watch the delta: the gap between what platforms claim and what your shop sold is not noise. It grows with broad targeting and view-through counting, and it tells you how much of your reported performance is accounting, not selling.
For the technical side of recording orders without cookies, including server-side setups for hosted checkouts, see our guide on conversion tracking without cookies.
Conclusion
Google Ads, Meta and your shop report three different revenues because they measure three different things, and only one of them measures your shop. Attribution windows and view-through counting are not errors, they are how the platforms are designed, and the gap will not close on its own. Put a neutral baseline in place: record what actually sells, attribute arrivals honestly, and treat platform revenue as what it is, a directional claim. MetriXs gives you that baseline, cookieless and EU-hosted, from the free plan up.
Frequently asked questions
- Why does Google Ads report more conversions than my shop?
- Google Ads counts a conversion when it happens within the conversion window after an ad interaction, 30 days after a click by default. It cannot see whether another channel or a direct visit actually delivered the customer, and modeled conversions add estimates on top. Every sale that merely follows an ad interaction is reported as a Google Ads conversion.
- What is view-through attribution?
- Meta’s default attribution is 7-day click or 1-day view. View-through means a sale is credited to an ad when the person saw the ad but never clicked it, as long as the purchase happens within the window. Since most people buy without clicking ads, this silently inflates reported results.
- Is one of the tools broken?
- No. Each tool answers a different question. Google Ads asks how many sales happened within its window after ad interactions, GA4 splits credit across the whole journey, and your shop counts money received. The disagreement is structural. Compare like with like: same period, same order definition, same treatment of refunds and taxes.
- How do I find out what my ads really earn?
- Strictly measuring incremental revenue from ads requires experiments, such as geo or holdout tests. As a practical baseline, record what your shop actually sold and which channels visitors arrived through, and treat platform-reported revenue as directional. The delta between claim and reality is itself useful information.
- Does MetriXs send my conversions to Google Ads or Meta?
- No. MetriXs does not export conversions to ad platforms and does not add modeled conversions. It records visits and orders on your side, labels visitors who arrive via Google Ads, Meta or Microsoft Ads click IDs, and breaks revenue down by source. Nothing is shared with the platforms, and no new subprocessor is involved.
One neutral baseline for your revenue
MetriXs records orders as they happen, breaks revenue down by real source and arrival channel, cookieless and EU-hosted. Free plan, paid tiers from 6 €/month.
Start measuring →Related: Track conversions without cookies · Calculate your consent loss · Shopify analytics without cookies · Google Analytics alternatives in 2026