Why Your Numbers Don't Match Shopify, Meta, and Google

Quick answer

They will never match, and that's expected. Shopify counts orders. Meta and Google count conversions they believe their ads caused. Those are different questions. On top of that, Meta rebuilt its attribution system twice in 2026, which lowered reported conversions across most accounts without any change in actual performance.

Why the platforms disagree with Shopify

Shopify counts an order when it happens. It doesn't care why.

Meta and Google count a conversion when they believe their ad influenced it, inside an attribution window they define. If a customer clicks a Meta ad Monday and buys Thursday, Shopify records one order on Thursday. Meta records one conversion attributed to Monday's ad.

Now add a customer who clicks both a Meta ad and a Google ad before buying. Shopify: one order. Meta: one conversion. Google: one conversion. Three systems, three true answers, and 200% attributed revenue.

This is why Agency AI tracks MER and blended ROAS. Total store revenue divided by total ad spend doesn't care how the platforms argue about credit. It's the only number that can't be double-counted.

What changed on Meta in 2026

Two changes, both large.

January 12, 2026. Meta removed the 7-day view and 28-day view attribution windows. The longest view-through window available is now 1 day. Advertisers relying on longer view windows saw reported conversions fall, with the largest drops in categories with long consideration cycles.

March 3, 2026. Meta narrowed the definition of a click. Click-through attribution now requires an actual link click. Likes, shares, saves, comments, and other non-link interactions no longer count as clicks. They moved to a renamed category, engage-through, which has a fixed 1-day window and now covers all ad formats, including video views of 5 seconds or more.

The consequence is subtle and worth understanding. Under the old system, someone who liked your ad on Monday and bought on Friday counted as a 7-day click-through conversion. Under the new one, that purchase doesn't qualify as click-through (no link click) and doesn't qualify as engage-through (outside the 1-day window). It disappears from attribution entirely.

Across the industry, reported conversions fell noticeably after these changes. Remarketing campaigns were hit hardest, because remarketing audiences engage socially before converting more often than cold audiences do.

Performance did not decline. Measurement changed. Don't compare your pre-March 2026 numbers to your post-March numbers directly. Set a new baseline from mid-March forward.

Meta's current default

7-day click, 1-day engage-through, 1-day view.

A conversion is attributed if the person clicked a link within 7 days, engaged with the ad within 1 day, or saw it within 1 day. When more than one applies, Meta credits click-through first, then engage-through, then view-through.

For e-commerce purchases, this default is generally the right setting. Purchases involve deliberation. People see an ad, think, and come back.

Why Google and Meta don't match each other

Google Analytics uses session-based attribution. Meta uses event-based. GA4 has no equivalent of view-through at all — if someone sees your Meta ad, doesn't click, and buys later, Meta counts it and GA4 has no idea the ad existed. That single difference explains a large share of the gap.

What to actually do about it

Pick one source of truth for business decisions. Shopify revenue and total ad spend. That gives you MER and blended ROAS, and it's the number your bank account agrees with.

Use platform-reported ROAS for relative comparison only. Meta's ROAS on Campaign A versus Campaign B is a useful signal. Meta's ROAS versus Shopify's revenue is not a meaningful comparison.

Don't chase the gap to zero. It won't close. Run both and read them for what each is good at.

Where Agency AI gets its numbers

Agency AI reads your Shopify orders to calculate true revenue, MER, and cost per acquisition, and cross-references first-time versus repeat buyers to separate genuine customer acquisition from repeat purchases. Platform-reported ROAS still appears at the campaign level, because that's the number Meta and Google are optimizing against.