The Short Answer
- People rarely buy from the first ad they see. In the large accounts we work across, the typical person saw a brand's ads 4.6 times a month, and those accounts had dozens of different ads live at once.
- Ads Manager gives each conversion to one ad. It's widely reported by practitioners that the most recent ad clicked gets the credit, and if there was no click, the most recent ad viewed. Meta's current help pages describe the attribution windows but don't spell out that rule.
- That makes your closer look like your best ad. A sale static is often the last thing someone clicks, so it collects the purchases. The videos that got them interested in the first place show up looking expensive.
- Meta changed attribution in 2026. Click-through now means link clicks only, and saves, shares and other engagements count as "engage-through."
- Don't judge every ad on last-click ROAS. Use leading indicators for upper-funnel creative, check blended results, and run lift tests when the stakes are high.
People See Several Ads Before They Buy
Very few purchases happen on the first impression. Someone sees a video while scrolling, keeps going, sees a different ad from the same brand a few days later, saves it, checks the site, and buys a week after that.
Our own data shows how much repetition is normal at scale. Across the 23 highest-spending US accounts we tracked from June to August 2026, the median monthly frequency was 4.6, meaning the typical person reached saw that brand's ads about four or five times a month. DTC brands ran even higher, at a median of 5.8. See our Meta ads benchmarks.
And those impressions aren't all the same ad. The median large account in our data had 75 creatives live at once. See how many ads to run on Meta. So a typical buyer may see a UGC video, a motion ad, a testimonial and a sale static from the same brand before they purchase.
Meta's own research points the same way. A Facebook IQ study of 11 brand campaigns found that lift in ad recall and purchase intent rose with frequency, and that the lift in purchase intent didn't start to slow until a frequency cap of about 1.5 a week. It also said there's no "magic frequency level" that works for every brand.
You'll often hear the "rule of 7," the idea that people need to see an ad seven times before they buy. There's no solid source behind the number. The point it makes is real, though: repeated, varied exposure is how most people get to a purchase.
How Meta Attribution Works
Attribution is how Meta decides which conversions to count for an ad. Two things control it: the attribution setting (which windows count) and the attribution model.
Attribution Settings
Per Meta's attribution help page, there are three ways a conversion can be attributed to an ad:
| Type | What counts | Window |
|---|---|---|
| Click-through | A purchase after someone clicks a link in your ad | 1 or 7 days |
| Engage-through | A purchase after a non-link interaction, like a save, share or like, or after watching a video ad for 5 seconds | 1 day |
| View-through | A purchase after someone saw your ad without interacting | 1 day |
Most advertisers use 7-day click and 1-day view, the default Meta set in 2021. You set this at the ad set level, and you can compare settings side by side using the "Compare attribution settings" columns in Ads Manager.
There's also a conversion count setting: all conversions (the default) counts every purchase in the window, including repeat purchases from the same person, while first conversion counts only the first one. If Ads Manager shows more purchases than you have new customers, this is often why.
Attribution Models
Meta now offers a choice of model:
- Standard counts every conversion that falls inside your attribution windows.
- Incremental uses machine learning to predict which conversions were actually caused by the ad. More on it below.
- Custom lets you send attribution data from your own analytics tool back to Meta.
Meta Incremental Attribution
Standard attribution counts every conversion that follows an ad, including people who would have bought anyway. Incremental attribution tries to count only the conversions your ads actually caused, and it optimizes delivery toward those. Meta's example: if a campaign has 100 conversions and Meta considers 70 of them incremental, the Results column shows 70.
It's available for campaigns with a sales, leads or engagement objective that optimize for website conversions. You can't use bid controls with it, and you can't switch models after a campaign is published. Because it reports fewer conversions, expect results to look lower than under standard attribution, even when the campaign is doing the same job.
Which Ad Gets the Credit?
Here's where most confusion comes from. A buyer might interact with four or five of your ads, but Ads Manager reports each purchase under one ad, not split across all of them.
The rule practitioners have long reported, including Meta ads expert Jon Loomer, is last touch: the most recent ad clicked gets the credit, and if there was no click, the most recent ad viewed. Meta's current help pages don't publish that tie-breaking rule, but it matches what advertisers see in their accounts.
Picture one customer's week:
- Monday: watches 6 seconds of a UGC video about the product, keeps scrolling.
- Wednesday: saves a motion ad explaining how it works.
- Friday: clicks a testimonial ad, browses the site, leaves.
- Saturday: clicks a static ad for 20% off and buys.
Under a 7-day click, 1-day view setting, Ads Manager credits the sale static with the purchase. The UGC video, the motion ad and the testimonial all helped, but none of them get the sale in Ads Manager. They don't disappear, though. They still show their impressions, clicks, saves and engagement. They just don't show the purchase.
The Bottom-of-Funnel Trap
This is what to watch out for. Your bottom-of-funnel ads, like a static promoting a sale, are usually the last thing someone clicks before buying. So in Ads Manager they look like your best performers: cheap cost per purchase, high ROAS.
Meanwhile, the top-of-funnel videos that got people interested look expensive. They brought people in, but the sale got credited somewhere else.
The natural reaction is to cut the "expensive" videos and move budget to the static. For a while, nothing changes, because people already warmed up keep buying. Then the account stalls. The static was closing sales that the videos started, and now nothing is starting them.
We walked through a real version of this, an $800 stroller where the demo video looked unprofitable while driving huge save and share volume, in why we track ad saves and shares.
None of this means sale statics are bad. They're doing their job. It means you can't judge a top-of-funnel ad and a closer by the same number.
What Changed in 2025 and 2026
Meta's attribution has changed a lot in the past two years:
- Click-through now means link clicks only. In March 2026, Meta announced that click-through attribution for website and in-store conversions would include only link clicks. Conversions after a share, save or other non-link interaction moved into engaged-view, now renamed engage-through. The same update cut the threshold for an engaged video view from 10 seconds to 5.
- Some view windows were removed from the API. As of January 12, 2026, Meta stopped returning 7-day view and 28-day view attribution in the Ads Insights API.
- Incremental attribution rolled out globally. Meta completed the global rollout in the second quarter of 2025. On its earnings calls, Meta said advertisers testing it saw an average 46% lift in incremental conversions compared with their usual approach, and that a newer model drove a 24% increase in incremental conversions versus standard attribution.
The engage-through change matters for creative. Saves and shares now have their own attribution bucket, which makes it easier to see when an ad is driving purchases through engagement rather than clicks.
How to Measure Ads That Don't Get Last-Click Credit
1. Use Leading Indicators
For top-of-funnel creative, look at signals that show the ad is building interest: cost per save, cost per share and hook rate. They tell you whether an ad is landing even when it isn't the last click. See why we track saves and shares and our hook rate benchmarks.
2. Compare Attribution Settings
Add the "Compare attribution settings" columns and look at an ad's conversions under 1-day click, 7-day click, 1-day view and engage-through separately. An upper-funnel video that looks weak on 1-day click but picks up conversions through views and engagement is doing more than its headline ROAS suggests.
3. Watch Blended Results
Look at total revenue against total ad spend across the business, sometimes called MER (marketing efficiency ratio). If you cut top-of-funnel ads and blended results drop a few weeks later, even while in-platform ROAS looks fine, those ads were doing more than Ads Manager showed.
4. Try Incremental Attribution
For campaigns that qualify, incremental attribution reports conversions Meta's model thinks the ads actually caused, not every conversion that happened to follow an ad. It's a better read on whether a campaign is adding sales or just collecting them.
5. Run a Lift Test
When the stakes are high, test it. Meta's Conversion Lift studies hold back a random group of people from seeing your ads and compare their purchases to people who did. Meta calls incrementality experiments like these "the gold standard in measurement." They need enough budget and time to be reliable, so they're best for big decisions, like whether to keep funding a whole layer of upper-funnel creative.
Plan a Full-Funnel Creative Mix
The fix isn't to stop running sale statics. It's to give every stage of the journey the right ads, and judge each one on its job.
| Stage | The ad's job | Typical creative | Judge it on |
|---|---|---|---|
| Top of funnel | Stop the scroll and create interest | UGC stories, motion explainers, hybrids, entertaining hooks | Hook rate, saves, shares, cost per engaged view |
| Middle of funnel | Build trust and answer objections | Testimonials, comparisons, demos, press and reviews | Clicks, engage-through conversions, site visits |
| Bottom of funnel | Close the sale | Offer statics, sale banners, urgency, bundles | Cost per purchase, ROAS |
Since Meta's Andromeda update, all of these can live in the same broad campaigns. Meta decides who sees what. Your job is to make sure there's a genuinely different ad for each stage, so the system has something to show people at every point in the journey. See creative diversity on Meta.
To see how a better CTR or conversion rate changes your results, try our ROAS calculator.
Build the Whole Funnel With Biddyco
Biddyco is a creative agency for Meta ads with a US-based senior team. We make the UGC, motion, static and hybrid ads that cover every stage of the funnel for brands spending $100K to $5M a month, across 100+ accounts. Book an intro call to talk through your account, or see more data in our Meta Ads Statistics hub.
FAQ
How does Meta attribution work?
Meta counts a conversion for an ad if it happens within the attribution window after someone clicks, engages with or views that ad. You choose the windows at the ad set level: 1 or 7 days for link clicks, and 1 day for engagements and views. Each conversion is reported under one ad.
If someone clicks several of my ads, which one gets credit?
It's widely reported that Meta uses last touch: the most recent ad clicked gets the credit, and if there was no click, the most recent ad viewed. Meta's current help pages don't spell out the tie-breaking rule, but this matches what advertisers see. The other ads still show their clicks and engagement, just not the purchase.
Does Meta use last-click attribution?
In practice, close to it. Ads Manager credits one ad per conversion, and practitioners report it goes to the last ad clicked within the window, or the last ad viewed if there was no click. Meta's incremental attribution model is different: it reports only the conversions it predicts the ads actually caused.
What is engage-through attribution on Meta?
Conversions that happen within 1 day after someone interacts with your ad without clicking a link, like saving, sharing or liking it, or after watching a video ad for at least 5 seconds. Meta introduced the name in 2026, when it moved saves, shares and other non-link clicks out of click-through attribution.
What is Meta incremental attribution?
An attribution model that uses machine learning to predict which conversions were actually caused by your ads, then optimizes delivery and reporting around those. It's available for campaigns with sales, leads or engagement objectives that optimize for website conversions.
How do I measure top-of-funnel ads on Meta?
Look at leading indicators like saves, shares and hook rate, compare attribution settings to see view and engagement conversions, and watch blended results across the business. For big budget decisions, run a Conversion Lift test.
