Google Ads attribution reports are produced by a company that earns revenue when you spend more on Google Ads. That is a structural incentive, not an accusation. But it should inform how much you trust those reports and how you use them to make decisions.
Google's attribution is not designed to give you a neutral view of where your customers came from. It is designed to show you the value of Google's products within Google's measurement framework. Understanding the difference between those two things will save you a meaningful amount of money.
Google Ads claims credit for all 1,000 conversions, but only ~400 were incremental
What Google Ads actually measures
Google can see Google. It can see clicks on Search ads, Display impressions, YouTube views, and Shopping interactions, when those touchpoints are tagged with the Google click ID. It cannot see your Meta ads. It cannot see your email campaigns. It cannot see your influencer posts, your podcast sponsorships, or your organic word-of-mouth.
Google's conversion tracking works by firing a conversion tag when a customer completes a goal action (a purchase, a lead form, a phone call). The tag looks back at that customer's recent Google touchpoints and assigns credit according to the selected attribution model.
The result is a partial picture of the customer journey, limited to Google's slice of it, presented as if it represents the whole.
The branded search problem
Branded keyword campaigns (ads that appear when someone searches your brand name) convert at very high rates. A campaign targeting "YourBrand", "YourBrand shoes", and "YourBrand review" will almost always show excellent ROAS in Google Ads, often 10x or higher.
Those conversions are real. People search your brand and then buy. The question is whether the ads are causing those purchases or capturing demand that was going to arrive anyway.
Incrementality tests on branded search consistently find that 70-90% of branded conversions are organic demand. These are people who already know your brand, already intended to buy, and were going to find you through the organic search result or by typing your URL directly. The paid ad appeared in front of them, they clicked it instead of the organic result, and Google logged a conversion.
You paid for a click that replaced a free click. The ROAS is excellent. The incremental value is close to zero.
This does not mean you should stop bidding on branded terms entirely. Competitive defense (preventing competitors from appearing on your brand searches) can justify some branded spend. But evaluating branded ROAS as evidence that paid search is generating incremental revenue is a mistake.
The attribution window inflates display and video
Google Ads defaults to a 30-day click attribution window for most campaign types. A customer who clicked a Google Display ad 28 days ago and purchased today gets attributed to that display campaign at full value.
Display advertising has very low click-through rates, typically 0.1% or below. Most of the people who click a display ad are not in active purchase mode at the moment of the click. When they eventually buy, weeks later, the display click sitting in the attribution window claims the conversion.
This structurally inflates display ROAS beyond what display advertising actually achieves in incremental terms. Video (YouTube) has similar dynamics, compounded by the fact that most YouTube conversions are attributed to view-through, meaning someone watched an ad for a few seconds and later bought, even without clicking.
Data-driven attribution does not fix the core problem
Google moved most accounts to data-driven attribution (DDA) by default. DDA uses machine learning to distribute credit across multiple Google touchpoints based on observed conversion path patterns. Within Google's ecosystem, this is genuinely an improvement over last-click.
But DDA only distributes credit across touchpoints Google can see. It cannot account for the Facebook prospecting ad that introduced your customer to the brand three weeks before they searched your name. It cannot credit the email newsletter that moved them from consideration to intent. It does not know those touchpoints exist.
DDA is a more sophisticated way of allocating credit within Google's bubble. The bubble is still a bubble.
How to calibrate your Google Ads data
Separate branded and non-branded campaigns and hold them to different standards. Branded campaigns should be evaluated on competitive defense value, not attributed ROAS. Non-branded campaigns are doing more incremental work and their attributed ROAS is a more meaningful (though still imperfect) signal.
Run a geo experiment on your branded search spend. Use matched geographic markets: maintain branded search in test markets, pause it in control markets for four to six weeks. Measure the difference in total backend revenue. Most brands find very little incremental impact. Some find none. The experiment will give you a defensible answer for whether the spend is justified.
Use MER as a cross-channel sanity check. Your Google-reported revenue and your actual backend revenue should move in the same direction over time. If Google is claiming $200k in attributed revenue per month and your backend shows $120k in total revenue, someone is doing the math wrong.