Why Facebook and Google Metrics Disagree (And What to Do About It)

Your Facebook Ads Manager shows 80 purchases. Your Google Ads shows 60 conversions. Your Shopify store shows 75 total orders.

80 + 60 = 140. But you had 75 orders. Who's right?

The frustrating answer: They're all "right" by their own rules. This guide explains exactly why the numbers differ and what you can actually do about it.

Four main reasons:

• Different attribution windows: Different time periods for claiming credit
• Different attribution models: Click vs view, last vs multi-touch
• Different tracking mechanisms: Pixels vs cookies vs modeling
• Self-serving attribution: Each platform wants to claim credit

The "attribution window" is how long after an ad interaction a platform will claim credit for a conversion.

Customer clicks a Facebook ad Monday, clicks a Google ad Wednesday, buys Friday.

• Facebook claims the sale (within 7-day click window)
• Google claims the sale (within 30-day click window)
• Both report 1 conversion, you had 1 order

Customer clicked the ad, then later converted. More reliable—shows clear intent.

Customer saw the ad (didn't click), then later converted. Much more questionable:

• Did the ad actually influence them?
• Or were they going to buy anyway?
• Meta's 1-day view attribution is often criticized

If someone sees your Meta ad, then searches your brand on Google and buys:

• Meta claims it (view-through, 1-day)
• Google claims it (click-through, brand search)
• Both get credit for 1 sale

View-through attribution is the biggest source of overcounting across platforms.

• ~80% of iOS users opted out of tracking
• Facebook can't track post-click behavior reliably
• Conversions are now "modeled" (estimated) not measured
• Attribution windows shortened

• Meta often underreports or overreports by 20-50%
• The direction (over vs under) varies by account
• Modeling quality depends on data volume
• Google was less affected (more search-based)

Pre-iOS14, discrepancies were 10-20%. Post-iOS14, 30-50% gaps are common. Meta's modeling is inconsistent, while Google's search data remained more stable.

Perfect attribution is impossible when customers use multiple devices, multiple platforms, and privacy settings block tracking.

• If Meta shows ROAS dropping 20% week-over-week, the direction is probably right
• If Google shows a campaign improving, it's probably improving
• Don't obsess over whether ROAS is 3.2 or 3.5

• Use Meta data to optimize within Meta (which ads, audiences)
• Use Google data to optimize within Google
• Don't try to compare across platforms with platform numbers

• Total revenue (from store) ÷ Total ad spend (all platforms)
• This MER/blended ROAS can't be gamed by platform attribution
• If blended metrics are healthy, the overall mix is working

• Look at platform metrics directionally (improving/declining)
• Optimize within each platform using its own data
• Don't reallocate budget based on platform-reported ROAS alone

• Use MER/blended ROAS as primary guide
• Run incrementality tests (turn off a channel, measure impact)
• Look at new customer acquisition, not just attributed sales

• Report platform metrics + blended metrics together
• Note that platform numbers will exceed actual sales
• Focus stakeholders on overall business metrics