Attribution Windows and Lookback Periods, Explained
An attribution window is the fixed span of time after an ad interaction during which a later conversion still gets credited to that ad. A lookback period is the same idea read backwards: at conversion time, how far back the platform looks to find an ad touch worth crediting. Same clock, opposite direction. Every number in your ad dashboards depends on where those clocks are set.
Here's the fraud case that made me care about this in the first place. A gaming client was buying installs from a smallish network, and their 7-day click window looked perfectly healthy on the surface. Underneath, the network was click-flooding: firing fake click events for users who were already going to install organically, then sitting inside that 7-day window to harvest the credit. Conversion rate from that source ran near 90 basis points of clicks converting when a clean source ran 30 to 40. The window wasn't the fraud. The window was the hiding place. Shorten it to 24 hours and most of that stolen credit evaporates, because you can't harvest a conversion you didn't have time to sit in front of.
So windows aren't a settings-page footnote. They decide which channel gets paid.
Click, view, and re-engagement windows
There are three flavors, and they don't behave the same way.
A click window (CTA, or click-through attribution) starts when someone taps the ad. If they convert inside the window, the ad gets credit. This is the strongest signal of the three, because a tap is a deliberate act.
A view window (VTA, or view-through attribution) starts when someone sees the ad without clicking. No tap. They scroll past, and if they convert later inside the view window, the platform still claims it. View windows are shorter than click windows almost everywhere, and for good reason. The causal link is weaker. Somebody who saw your ad for 300 milliseconds and bought two days later may or may not have been moved by it. Treat VTA numbers as suggestive, not proof.
A re-engagement window applies to users who already have your app installed. The install is done, so you're not attributing a first-open. You're attributing a return session or a re-purchase to a retargeting or deep-link touch. Re-engagement windows tend to be tighter still, because the intent-to-action gap for an existing user is short. Meta's engage-through component, rolled out from February 2025 per AdAmigo's rules writeup, is essentially this: crediting a conversion to an engaged view rather than a hard click.
The practical rule I use: the weaker the signal, the shorter the window should be. Click can breathe. View should be kept on a leash.
Default and maximum windows by network
This is the table I wish someone had handed me on day one. Values are current as of the platforms' 2025-2026 documentation. They move, so treat this as a starting map, not scripture.
| Network | Default click window | Max click window | Default view window | Max view window | Notes |
|---|---|---|---|---|---|
| Meta | 7-day click | 7 days | 1-day view | 1 day | Default is 7-day click + 1-day engage-through + 1-day view; click caps at 7 days |
| Google Ads | 30-day click | 90 days | up to 30 days | 30 days | Preset options only: 1, 7, 30, 60, 90 days depending on source |
| TikTok | 7-day click | 28 days | 1-day view | 7 days | Click options 1/7/14/28; view options off/1/7 |
| SKAN 4.0 (iOS) | n/a (measurement windows) | 35 days total | n/a | n/a | Windows 0-2d, 3-7d, 8-35d; postbacks delayed by randomized timers |
A few things worth staring at.
Meta caps the click window at 7 days. That's short compared to Google, and it's deliberate. Meta's default of 7-day click plus 1-day view is confirmed in Jon Loomer's 2026 attribution breakdown, and the platform recommends it for most advertisers. If your product has a genuinely long consideration cycle, Meta simply won't credit the top-of-funnel touch that happened three weeks ago. That's not a bug you can configure away.
Google is the generous one. Per Google Ads Help, the default click window is 30 days and you can push click-through out to 90. View-through, though, maxes at 30. So Google will look a lot further back for a click than Meta will, which means when you compare the two dashboards, you're not comparing like with like. You're comparing a 30-day memory against a 7-day memory.
SKAN is a different animal entirely and doesn't fit the click/view table cleanly. Apple's SKAdNetwork 4 doesn't give you a conventional lookback window. It gives you three sequential measurement windows: days 0-2, days 3-7, and days 8-35, per Adjust's SKAN 4 documentation. Then it delays the postback with a randomized timer, 24 to 48 hours for the first postback and 24 to 144 hours for the later two. So you don't just have a shorter window on iOS. You have a window whose results arrive on Apple's schedule, blurred on purpose so you can't reverse-engineer individual users. Anyone who tells you they have real-time SKAN attribution is selling you something.
Why overlapping windows double-count
Now the part that quietly wrecks budget decisions.
Each platform runs its own window, in its own silo, and claims any conversion that lands inside it. There's no referee deduplicating between them. So a single buyer who saw a Meta ad on Monday, clicked a Google search ad on Thursday, and converted Friday gets counted once by Meta and once by Google. Both are telling the truth by their own rules. Both are wrong about the total.
The magnitude here is not subtle. Databox's analysis of the attribution problem found that summing platform-reported conversions across Meta, Google, and LinkedIn routinely lands at 150 to 250% of actual closed customers. Read that again. If your platforms collectively report 175 conversions and you actually closed 95, every ROAS number you're using to allocate spend is inflated by nearly double. You're not making a rounding error. You're steering a budget with a broken speedometer.
Wider windows make this worse, mechanically. A 90-day Google click window overlaps more of every other platform's window than a 7-day one does. So the platform with the longest memory tends to over-claim the most, and it's usually the one you'd least suspect because its dashboard looks so confident.
The fix isn't a clever window setting. No combination of windows deduplicates across silos, because the silos don't talk. You need a source of truth that sits above all of them: your own server-side conversion data or your CRM, matched once, then reconciled against what each platform claims. Ruler Analytics frames it as a trust hierarchy, closed-won CRM data first and platform dashboards dead last, and that ordering is right. Platform-reported conversions are the least trustworthy number in the stack, not the most.
If you want to actually prove a channel drove incremental revenue rather than just claimed a window, attribution windows alone won't get you there. That's a job for holdout testing, which is a different tool with different math. Incrementality testing versus multi-touch attribution is where that argument lives.
Lookback versus window: the same clock, and why people confuse them
People use "attribution window" and "lookback window" interchangeably, and mostly that's fine, but the framing matters when you're debugging.
A window is prospective. You set it on the ad, and it runs forward: click now, credit for the next 7 days.
A lookback is retrospective. You stand at the conversion and search backward for eligible touches. Google Analytics 4's key-event lookback is this: at conversion, how many days of prior touches are in scope for the model to distribute credit across.
Functionally the same interval. But when a conversion goes missing, knowing which direction the tool is reasoning tells you where to look. If a platform runs a forward window and the touch happened before the window opened, it's invisible. If a tool runs a backward lookback and the touch predates the lookback horizon, same result, different cause. When numbers don't reconcile between two tools, nine times out of ten it's because their windows have different lengths, or one is counting clicks while the other counts views.
A short FAQ
What's a good default attribution window? For paid social with fast-moving intent, 7-day click and 1-day view is a defensible default and it's what Meta and TikTok both ship. For search and longer consideration cycles, Google's 30-day click is reasonable. The mistake is leaving a 90-day view window on and then trusting the numbers.
Why do my platform totals add up to more than my actual sales? Overlapping windows. Each platform claims the same conversion under its own rules with no cross-platform dedup. Databox puts the typical over-count at 150 to 250% of real closed customers. Reconcile against server-side or CRM data.
Does a longer window mean better attribution? No. Longer windows credit more distant, weaker touches and increase double-counting across platforms. Match the window to the real intent-to-action gap for your product, then stop.
Can I trust SKAN's attribution timing? Trust the aggregate, not the timing. SKAN 4's randomized postback timers (24-48 hours for the first postback, 24-144 for the rest) mean results arrive delayed and deliberately fuzzed. It's private-by-design, which is the point, and it's the price of iOS measurement now.
The honest summary: a window is a policy choice about how much credit to hand out and for how long, and every platform tunes its own to look as good as it defensibly can. Set yours on purpose. Then never sum across silos and call it a total.