Ecommerce Facebook Ads Attribution: Fix CAPI; Expect 1.2–1.6x GA4

Facebook ads attribution is the system Meta uses to decide which ad gets credit for a sale, based on the clicks and views that happened before it. Meta’s practical default is a 7-day click window plus a 1-day view window, meaning a purchase gets credited to an ad if someone clicked it within about a week or saw it within the last day as documented by Jon Loomer. The immediate takeaway: expect Meta’s dashboard to report more conversions than Google Analytics 4, and plan your measurement stack around that gap rather than chasing false parity.

TL;DR:

• Meta’s attribution model credits conversions primarily based on a seven-day click window and a one-day view window, often inflating reported numbers compared to GA4.
• Using a 7-day click and 1-day view window is suitable for most ecommerce scenarios, but shorter or longer windows should align with customer buying behavior.
• Pairing your Pixel with the Conversions API is crucial to recover lost signals and improve attribution accuracy amid increasing privacy restrictions.
• Differences between Meta and GA4 figures typically stem from Meta’s view-through modeling and tracking opt-outs, with ratios of 1.2 to 1.6 being normal, and larger gaps indicating misconfiguration.
• Regularly monitoring event match quality, deduplication, and ratio between platforms is essential to maintain a trustworthy measurement stack and optimize ad performance effectively.

Table of Contents

• How Meta’s attribution model assigns credit to your ads
• Attribution windows explained: click, view, and engaged-view
• Signal loss and how CAPI, AEM, and SKAN try to fix it
• Why Meta’s numbers never match GA4 (and how to reconcile them)
• Choosing attribution settings for real ecommerce scenarios
• Building a measurement stack you can trust
• Practitioner recommendations and common pitfalls I see working with ecommerce accounts
• My perspective: platform reporting versus cross-channel truth
• How my team at Cosma can help with attribution and CAPI
• Sources
• FAQ

How Meta’s attribution model assigns credit to your ads

Meta’s model works on a lookback basis. Any click or view that falls inside the chosen window makes an ad eligible for credit if a conversion happens afterward, regardless of how many other ads or channels the person interacted with in between. This is why two overlapping campaigns can each claim the same sale in their own reporting.

There are three ways Meta counts an interaction:

• Click-through: someone clicks a link in the ad and converts within the click window.
• View-through: someone sees the ad without clicking and converts within the view window, currently 1 day by default.
• Engaged-through: someone interacts with a video or carousel (without a link click) and converts within a short window tied to that engagement.

The setting you choose does more than change your reports. It tells Meta’s delivery system what a “conversion” looks like, which shapes who the algorithm targets and how it spends your budget, according to Jon Loomer’s analysis of Meta’s attribution mechanics. Widen the window and you will likely see more conversions credited to view-through interactions, which can shift optimization toward broader reach rather than direct response. That is the tradeoff most advertisers do not realize they are making when they toggle this setting without testing it first.

Attribution windows explained: click, view, and engaged-view

Each window answers a different question about when credit should count. Here is how the common options break down for a typical ecommerce funnel:

1. 1-day click: counts a purchase only if it happens within 24 hours of a link click. Best for flash sales or impulse purchases where the buying decision is fast.
2. 7-day click: counts a purchase up to a week after a click. This is the default click window and fits most standard ecommerce funnels where people compare, leave, and return later.
3. 1-day view: counts a purchase within 24 hours of an ad impression, even with no click. Useful for catching brand-driven purchases but prone to inflating credit for ads that were simply seen alongside other marketing.
4. Engaged-view: counts interactions with video or carousel content that lead to a fast conversion, relevant mainly for video-heavy campaigns.

For most ecommerce accounts, 7-day click plus 1-day view is the right starting point because it captures the realistic consideration window without overcrediting passive impressions, a default Meta itself applies as of mid-2026. Shorter windows undercount slower-considered purchases. Longer or view-inclusive windows overcount casual scrollers who bought for unrelated reasons.

Signal loss and how CAPI, AEM, and SKAN try to fix it

Apple’s App Tracking Transparency (ATT) prompt cut off a large share of the browser and device signal Meta’s Pixel used to rely on, since many people decline tracking. In response, Meta built Aggregated Event Measurement (AEM) to model outcomes for opted-out users and SKAdNetwork (SKAN) to handle app-install attribution with heavily limited, delayed data. Both are workarounds, not replacements: they smooth over gaps but never restore the granularity of direct, matched events.

The real fix is pairing your Pixel with the Conversions API (CAPI), which sends event data directly from your server to Meta instead of relying only on a browser signal that ad blockers and privacy settings can strip out. This is the highest-priority technical investment for any advertiser spending meaningful budget, since attribution-driven bidding is only as good as the events feeding it, per AdLibrary’s 2026 attribution tracking guide.

CAPI can recover a significant proportion of events lost to browser-side signal loss, which directly raises Event Match Quality, according to AdLibrary. That single fix often does more for reporting accuracy than any attribution window change.

Once CAPI is live, prioritize these checks in Events Manager:

• Confirm server events fire for every key action, not just purchases.
• Set a matching event_id on both browser and server events to enable deduplication.
• Fill every available parameter (email, phone, external ID) to boost match rate.
• Watch for event gaps or sudden drops, which usually flag a broken integration.

Why Meta’s numbers never match GA4 (and how to reconcile them)

Meta and GA4 are not measuring the same thing, so exact agreement should never be the goal. Meta counts view-through conversions that GA4 ignores, models purchases for people who opted out of tracking, and uses ad-interaction timestamps rather than GA4’s session-based logic, as explained by AdAdvisor’s breakdown of attribution windows. Time zone settings and reporting-window boundaries add further drift between the two platforms.

A simple reconciliation workflow catches real problems without chasing noise:

• Pull Meta and GA4 numbers for the same date range and campaign set.
• Filter both to the same conversion event, such as purchase.
• Divide Meta’s reported conversions by GA4’s to get a ratio.
• Investigate any sudden jump in that ratio week over week.

Pro Tip: Treat a Meta-to-GA4 ratio in the 1.2 to 1.6 range as normal; if it climbs past 2x, check for duplicate Pixel firing or a broken GA4 tag before touching your budget.

That 20 to 40 percent premium for Meta’s reported numbers over GA4 is a commonly observed pattern among practitioners, and a much larger gap almost always points to misconfiguration rather than real performance, per AdAdvisor’s analysis.

Choosing attribution settings for real ecommerce scenarios

The right window depends on how your customers actually buy, not on what feels safest to leave on default. Match the setting to the buying behavior you are trying to measure.

1. Prospecting campaigns targeting cold audiences: keep 7-day click and 1-day view. Cold buyers often need several days to decide, and cutting the click window here will undercount real demand you generated.
2. Retargeting campaigns aimed at cart abandoners: 1-day click alone is often enough, since the intent signal is already strong and a shorter window keeps reporting tighter to actual retargeting influence.
3. Impulse-purchase products under $50: 1-day click, 1-day view. These buyers decide fast, so a longer window mostly adds noise from unrelated later purchases.
4. Subscription or considered-purchase products: 7-day click, drop the 1-day view. Subscriptions involve research, and view-through credit tends to overstate the role of passive impressions in a longer decision cycle.

Before locking in a change across live campaigns, use Meta’s Compare Attribution Settings feature to see how reported conversions would shift under different windows without actually changing delivery. If you decide to change a live setting, run it as a controlled test for 7 to 14 days and judge the result against both Meta and GA4, a practice Jon Loomer recommends for any attribution change, since a setting change also alters what the algorithm optimizes for, not just what gets reported, as Loomer notes separately.

Pro Tip: Never change attribution windows and creative strategy in the same week. You will not know which one moved your numbers.

Building a measurement stack you can trust

Attribution health is not a one-time setup. It needs a short list of numbers checked on a regular cadence, plus a plan for when platform reporting alone is not enough to justify a budget decision.

Minimum monitoring in Events Manager includes:

• Event Match Quality score for each key event.
• Deduplication rate between Pixel and server events.
• Gaps or sudden drops in event volume.
• Diagnostic alerts Meta surfaces automatically for broken setups.

Aim for an EMQ around 6 to 8 and a deduplication rate above 80 percent on purchase events once Pixel and CAPI are both correctly implemented, a target range outlined in LionelZ’s guide to Meta attribution models.

When the ratio between Meta and mobile measurement partners like AppsFlyer drifts widely, mismatched lookback windows are a common culprit since AppsFlyer’s windows are configurable from 1 to 30 days while Meta’s click default is 7, per AppsFlyer’s own documentation on Meta ads discrepancies. Reserve incrementality testing or media-mix modeling for bigger questions: whether a channel is truly adding sales or just claiming credit for demand that existed anyway. Meta’s own Incremental Attribution option uses randomized holdouts to isolate that lift directly inside Ads Manager, an approach worth running before any major budget reallocation.

Practitioner recommendations and common pitfalls I see working with ecommerce accounts

When I audit an account, I follow the same sequence every time: check CAPI implementation first, then EMQ, then deduplication rate, then compare attribution windows, then reconcile against GA4. Skipping straight to window changes without fixing the data feeding them wastes the exercise.

The most common mistakes I run into:

• Missing or mismatched event_id values, which causes double counting or silent event loss.
• Duplicate purchase events firing from both browser and server with no dedup logic.
• Wrong event priority set in Events Manager, so a lower-value action gets optimized ahead of purchase.

In practice, checking deduplication and event_id pairing first catches most of these issues before they ever require touching a bid strategy or attribution window.

Pro Tip: Run a single test purchase after any Pixel or CAPI change and confirm it shows up once, not twice, in Events Manager before trusting the data again.

My perspective: platform reporting versus cross-channel truth

I use Meta’s own numbers to guide day-to-day optimization, since that is what the algorithm is actually reacting to, but I lean on GA4 for cross-channel reconciliation and save incrementality testing or media-mix modeling for real budget decisions. At Cosma, my team spends more time fixing measurement inputs than adjusting bids, because a clean data feed usually solves what looked like a targeting problem.

— Stefano Mazzei

How my team at Cosma can help with attribution and CAPI

If your Meta numbers and GA4 keep drifting apart, my team at Cosma can run an attribution audit, fix your CAPI setup, and test incrementality before you touch your budget. Book a call through Cosma’s booking page to get started.

Sources

• Meta Ads Attribution (2026) — Jon Loomer
• Meta Ads Attribution: Windows, Why Meta and GA4 Never Agree, and Which Number to Trust | AdAdvisor
• Meta ads discrepancies — AppsFlyer Support
• Facebook ads attribution tracking: 2026 guide — AdLibrary

FAQ

What is attribution in Facebook ads?

Attribution in Facebook ads is the process Meta uses to decide which ad interaction gets credit for a conversion. Meta’s practical default credits a purchase to an ad if it followed a click within 7 days or a view within 1 day, as reported by Jon Loomer.

What does 7-day click, 1-day view attribution mean?

It means Meta credits a conversion to an ad if the person clicked it within the past 7 days or simply saw it within the past 24 hours, even without clicking. This is Meta’s current default setting for most ad accounts as of mid-2026, according to Jon Loomer.

How does Meta’s attribution setting affect ad delivery?

The attribution setting does not just change reporting, it changes what Meta’s algorithm treats as a success signal for optimization. Changing the window can shift delivery toward different audiences or behaviors, which is why Jon Loomer recommends testing changes with a holdout before rolling them out across spend.

What is the best attribution setting for Facebook ads?

For most ecommerce accounts, 7-day click plus 1-day view is the safest starting point, since it matches typical consideration times without overcrediting passive views. Faster impulse purchases or tight retargeting audiences often perform better measured on a 1-day click window instead.

Why do Meta and GA4 report different conversion numbers?

Meta counts view-through conversions and models purchases for people who opted out of tracking, while GA4 relies on session-based attribution that ignores view-through entirely. A Meta-to-GA4 ratio between 1.2x and 1.6x is a commonly observed and acceptable range, per AdAdvisor’s analysis, while anything beyond 2x usually signals a tracking misconfiguration.