Ad attribution is the process of assigning credit to the ads and marketing channels that a customer interacted with before completing a conversion. When someone buys from your store, fills out a form, or signs up for a trial, attribution asks: which ad or channel should get credit for that conversion?
The answer to that question determines how your ad spend is reported, how platforms evaluate your campaigns, and where you allocate budget next. It's also frequently wrong.
Attribution assigns credit for a conversion back to the ads that preceded it
How ad attribution works
When a user visits a website after clicking an ad, that platform (Meta, Google, TikTok) stores a record of the interaction. It might be a cookie on the user's browser, a mobile device identifier, or a server-side event record. If that same user converts within a defined time window, the platform attributes the conversion to the ad they clicked.
The "credit" translates into reported conversions and revenue in your ad dashboards. Your Meta Ads Manager shows 140 purchases this week. Your Google Ads shows 110 purchases. Those numbers are the output of attribution systems running independently on each platform.
Attribution models
An attribution model decides how credit is distributed when a customer interacts with multiple touchpoints before converting. The main models:
Last-click gives 100% of credit to the final ad click before conversion. Common, simple, and heavily biased toward channels that appear late in the funnel (like branded search and retargeting).
First-click gives 100% of credit to the first interaction. Overcredits top-of-funnel channels like prospecting campaigns and organic discovery.
Linear splits credit equally across all touchpoints. Treats every interaction as equally important, which is rarely accurate.
Time-decay gives more credit to touchpoints that happened closer to the conversion. The logic is that recent interactions are more likely to have influenced the decision.
Data-driven uses machine learning to assign fractional credit across touchpoints based on patterns in conversion data. Sounds sophisticated and is more accurate than simple rules, but still operates within the limitations of what the platform can see.
Attribution windows
An attribution window is the time period after an ad interaction during which a conversion gets credited to that ad. If someone clicks a Facebook ad and buys within the window, Facebook claims the conversion. If they buy after the window closes, they don't.
Common defaults:
- Meta: 7-day click, 1-day view. Any purchase within 7 days of a click or 1 day of an impression is attributed to Meta.
- Google Ads: 30-day click. Any conversion within 30 days of a click is attributed to Google.
- TikTok: 7-day click, 1-day view.
The window determines which conversions get counted and has a large effect on reported performance. A brand that switches from a 30-day click window to a 7-day click window will see its reported conversions drop significantly, even if actual business performance is unchanged.
The problem with ad attribution
Platforms attribute conversions they didn't necessarily cause. If someone saw a Facebook ad on Tuesday and bought on Thursday after searching Google and clicking a branded search ad, both Facebook and Google will claim the conversion. Facebook claims it via a view-through or click attribution. Google claims it via the search click. Both numbers appear in both dashboards.
This double-counting is called attribution overlap. It's why the sum of all your platform-reported conversions reliably exceeds the actual number of orders in your backend. It's not a rounding error: it's structural.
The overlap happens because:
- Each platform's attribution system only sees its own touchpoints. Meta doesn't know about the Google click. Google doesn't know about the Facebook impression.
- Attribution windows overlap with each other. A 30-day Google window and a 7-day Meta window will often cover the same purchase.
- View-through attribution credits an ad impression even when the user took no action on the ad itself.
A common real-world example: a brand's Meta Ads Manager reports $180,000 in attributed revenue. Their Google Ads reports $140,000. Their Shopify backend shows $200,000 in total orders. The platforms together are claiming 60% more revenue than actually occurred.
What ad attribution can and can't tell you
Attribution can tell you which channels and campaigns people interacted with before converting. It can tell you the sequence of those interactions within a single platform's visibility. It can give you relative performance signals within a single platform (ad A outperformed ad B, measured by the same attribution logic).
Attribution cannot tell you which interactions caused the conversion. A customer who clicked a Facebook ad, then clicked a Google Shopping ad, and then converted might have bought regardless of either ad. Attribution models assume that exposure led to conversion. That assumption is wrong a significant portion of the time.
Attribution is descriptive, not causal. It describes what was present in the customer journey. It doesn't determine what drove the decision.
Platform attribution vs. independent attribution tools
Each ad platform's native dashboard only sees its own touchpoints. Meta sees Meta touchpoints. Google sees Google touchpoints. These siloed views make cross-channel comparison impossible without a shared data layer.
Independent attribution tools (GA4, Northbeam, Triple Whale, Rockerbox) collect first-party data via a pixel or tag installed on your site. They build a cross-channel view of the customer journey because they see the full sequence of a user's visits across all sources, not just the interactions on one platform.
These tools provide more complete data than any individual platform dashboard. They still face fundamental limitations: cross-device journeys, incognito browsing, iOS privacy changes, and offline interactions remain invisible. Studies suggest that 30-50% of customer interactions aren't trackable by any pixel-based system.
The alternative: incrementality testing
Incrementality testing measures whether your ads actually caused conversions, rather than just being present in the journey before the conversion happened. The method: randomly split your audience into two groups, show ads to one group and suppress ads to the other, then compare conversion rates between the groups in your backend data.
The difference in conversion rates is the incremental lift caused by your advertising. This is a causal measurement, not a descriptive one. It's also the only way to know if the conversions your attribution system is reporting would have happened without your ads.
Incrementality testing is slower (4-8 weeks per test) and can't cover every channel simultaneously. But it answers the question attribution can't: not which touchpoints appeared, but which ones actually changed behavior.