Why the same order shows two different numbers
Every order that flows through your store is one real event. Facebook Ads Manager and GA4 each try to describe that event, but they answer different questions. Facebook answers "did my ad influence this sale?" GA4 answers "what happened in a trackable browser session?"
Those are not the same question, so the answers rarely line up. Below are the four mechanisms doing most of the damage, grouped by whether you can fix them or only understand them.
Attribution models pull in opposite directions
GA4's default is last non-direct click layered with data-driven attribution, which splits one conversion fractionally across several touchpoints. So a single order might show up as roughly half a conversion under "Paid Social" in GA4, with the rest scattered across organic and direct.
Facebook uses its own attribution window instead. If its ad was clicked or seen inside that window, Facebook claims the whole order — one full conversion, no splitting. As Ruler Analytics explains, Facebook self-attributes while GA4 defaults to last non-direct click, so a customer who clicks your ad but later returns via a brand search gets credited to Facebook in one tool and to Google in the other.
View-through conversions: Facebook counts them, GA4 doesn't
This is the biggest single reason Facebook's order count runs high. Facebook's current default window is 7-day click plus 1-day view, according to Jon Loomer's 2026 attribution guide and Foreplay's attribution breakdown. That "1-day view" half means Facebook claims any purchase made within a day of someone seeing your ad without clicking.
GA4 has no concept of a view. It only records what a user did in a session on your site. So every view-through order Facebook counts is an order GA4 attributes to whatever channel the buyer actually arrived through — usually direct or organic.
You can see how large this credit window is by shrinking it. Switching a campaign from the default down to a strict 1-day-click window can cut Facebook's reported conversions by roughly 40%, per TrackBee's testing — same real sales, narrower credit.
Client-side data loss shrinks GA4's number
While Facebook inflates, GA4 deflates. GA4 runs on client-side JavaScript that has to load and fire before a purchase is recorded. Ad blockers, Safari and Firefox tracking prevention, and cookie-consent declines stop that tag cold. Elevar and Audiense estimate that ten to twenty-five percent of users fall into these blocked or opted-out buckets.
The result is a structural undercount. GA4 typically records fifteen to thirty percent fewer purchases than the store's server-side order record, according to BlueFrog Analytics and Consentmo. A buyer who closes the tab before the confirmation page loads is a completed sale your store keeps and GA4 never sees.
Timezone and click-date reporting desync your days
Even when totals eventually agree, daily views won't. Facebook reports a conversion on the date of the click or view that earned credit, not the purchase date. A Monday click that converts Thursday lands on Monday in Facebook and Thursday in GA4.
Timezones make it worse — GA4 reports in the property's configured timezone while Facebook uses the ad account's, as OWOX notes in its discrepancy guide. Always compare on trailing seven- to fourteen-day windows, never single days.
What a "normal" gap actually looks like
Some gap is healthy. A large gap means something is broken. Here's how to read yours — the same diagnostic logic behind reconciling all of your ecommerce data.
- Facebook higher than GA4 by a wide margin: expected. Facebook adds view-through and modeled conversions; GA4 adds none. This gap is structural.
- GA4 fifteen to thirty percent below your store's order count: normal client-side loss, per BlueFrog.
- GA4 thirty to forty percent below: investigate your tag setup.
- GA4 forty percent or more below: a genuine tracking break — a misfiring tag, a broken consent banner, or a missing purchase event.
The goal is a stable ratio between the two tools, not equality. If your Facebook-to-GA4 ratio holds steady week over week, your tracking is fine even though the raw numbers differ. A sudden ratio change is your real alarm.
A worked example: one week, two very different counts
Say your store logs 100 real orders in a week. Here's how each tool describes that same 100.
Facebook Ads Manager reports about 78 "purchases." Of your 100 buyers, 55 clicked an ad within seven days and 15 only saw one within a day — that's 70 by window. Facebook then models roughly 8 more it couldn't directly observe (iOS opt-outs, blocked pixels), reaching about 78. It reports each on the click date, so a chunk lands in the prior week.
GA4 reports about 72 "purchases," fractionally split. It loses roughly 20 buyers to ad blockers, consent declines, and closed tabs, recovers a few through modeling, and lands near 72. Under data-driven attribution it credits "Paid Social" only a slice of each — maybe 48 conversions show against Facebook, with the rest spread across organic and direct.
So one week, 100 real orders, and two tools that disagree by six on the total and by thirty on how many Facebook "drove." Neither is lying. Facebook is answering "how many did my ad touch?" and GA4 is answering "how many did I witness in a session?"
The same divergence shows up in your Facebook ROAS versus GA4 and in session counts between the two tools — and it isn't unique to Meta, as anyone comparing Google Ads conversions to GA4 discovers.
How to narrow the gap — and what you can't fix
You can shrink the tracking gaps. You cannot close the methodology gaps.
Tag every paid link with UTMs. Manual utm_source=facebook tagging is what lets GA4 classify the traffic correctly instead of dumping it into direct, as both Ruler and OWOX stress.
Standardize timezones across Facebook, GA4, and your store so day boundaries align.
Add server-side tracking (CAPI and GA4 server-side) to recover blocked-browser events. This narrows GA4's undercount — but note it does nothing for view-through or modeling. Even with flawless tracking, Facebook's structural inflation over a store's own count runs twenty to thirty-five percent, per Vaizle. Anyone promising CAPI will "make the numbers match" is selling you a fix for a problem that isn't fixable.
Compare on trailing windows, not single days, to absorb the click-date offset.
Why this mismatch actually costs you money
Here's the part the ranking articles skip: this isn't an analytics annoyance, it's a profit problem. You calculate return on ad spend from these numbers. If you trust Facebook's inflated order count, your ROAS looks better than reality and you scale a campaign that's quietly unprofitable.
Say Facebook claims 78 orders on $1,000 of ad spend at $40 each — that reads as $3,120 in ad-driven revenue and a healthy 3.1x ROAS. But your store only logged 55 last-click orders from Facebook. And revenue isn't profit: on a print-on-demand mug, a $40 order might carry $16 product cost, roughly $1.46 in Shopify processing fees (about 2.9% plus 30 cents on the Basic plan, per Webgility), and $10 of allocated ad spend — leaving closer to $12.54 in true per-order profit. Attribution decides which campaigns get credit for that $12.54, so the wrong count sends your budget to the wrong ads.
That's the gap between a platform-reported number and the truth of what a sale is worth. It's the same reason merchants moving from Etsy to Shopify suddenly need a single source of profit truth.
PodVector connects Shopify, Meta Ads, Google Ads, Printify, Printful, and Stripe, then computes true per-order profit across all of them — so you're deciding on margin, not on Facebook's self-credited count. Victor, its AI operator, reads that live data, surfaces which campaigns actually earn, and proposes Shopify-side moves for your approval. Victor is not a dashboard and does not touch your ad account — he reads ad data and hands you the decision. See your real per-order profit.
FAQs
Should Facebook Ads orders ever equal GA4 conversions?
No. They measure different things — Facebook credits view-through and modeled conversions on a click-date basis, while GA4 counts only fired sessions with fractional credit. Aim for a stable week-over-week ratio, not equality. If the ratio suddenly shifts, that's your signal to investigate.
Which number is correct, Facebook or GA4?
Neither is the source of truth for how many sales happened — your store's own server-side order record is. Facebook's number tells you how many sales its ads plausibly influenced; GA4's tells you what its tags observed. Use your store platform for order counts and revenue, and treat Facebook and GA4 as influence estimates.
Why is Facebook always higher than GA4?
Facebook adds two things GA4 refuses to count: view-through conversions (sales credited to an ad that was seen but not clicked) and modeled conversions (statistical estimates for buyers it couldn't observe). Its default 7-day-click-plus-1-day-view window, documented by Jon Loomer, is far wider than anything GA4 applies.
Will setting up the Conversions API make them match?
No. CAPI recovers events lost to ad blockers and closed tabs, which narrows GA4's undercount. But it does nothing about view-through, modeling, or last-click-versus-window differences — the methodology gaps that cause most of the mismatch. A jump in your conversion count after adding CAPI usually means a deduplication error, not a real gain.
What gap size means my tracking is actually broken?
For GA4 versus your store's order count: fifteen to thirty percent lower is normal, thirty to forty percent warrants investigation, and forty percent or more signals a real break, per BlueFrog Analytics. For Facebook, there is no "broken" threshold — it's expected to run high, so watch for sudden ratio changes instead of an absolute number.
How do I compare them fairly?
Use trailing seven- to fourteen-day windows to absorb Facebook's click-date reporting offset, align all three timezones, and tag every paid link with UTMs so GA4 classifies traffic correctly. Then compare trends and ratios rather than expecting the totals to reconcile to the same figure.