What actually matters in Meta attribution now that the easy tracking era is over?
After iOS 14.5, Meta-reported conversions dropped by roughly 40% for many advertisers. That wasn’t just a reporting issue. For a lot of ecommerce brands, it changed budget decisions, creative strategy, and how quickly teams reacted to performance swings. But some brands recovered faster than others.
At Positive Sparks, we reviewed 10 DTC client accounts across different spend levels and product categories to see what separated the recovery stories from the accounts that stayed stuck in attribution fog. The pattern was surprisingly consistent: the strongest accounts were not tracking everything. They were tracking a few signals really well.
1. View-through attribution (VTA)
Why start here when so many advertisers still distrust view-through data?
Because the better accounts didn’t use VTA as vanity reporting. They used it as an optimization signal. In our 10-account review, the accounts with VTA enabled and monitored at campaign level showed stronger alignment between Meta-reported demand and eventual Shopify revenue. Accounts with VTA turned off had 28% lower reported ROAS and, more importantly, 19% lower real revenue.
That gap matters. If somebody sees your ad on mobile, doesn’t click, then buys later through branded search or direct traffic, the click-only model misses a meaningful assist. You know, that missing assist can lead us to kill creative that is quietly doing useful work higher up the funnel.
Actionable takeaway: Keep VTA on, review it weekly beside Shopify sales, and use it to inform creative and audience decisions rather than treating it as a decorative metric.
2. Conversions API match quality
What’s the clearest technical health check in a post-cookie environment?
It’s server-side match quality. Across the same 10 DTC accounts, those with Conversions API match quality above 8.0/10 achieved 2.3x higher attributed ROAS than accounts sitting below 5.0. That’s a huge spread, and it tells us Meta’s system performs better when it can confidently match events back to real users.
This usually comes down to sending a stronger event payload: email, phone, IP, user agent, external ID, and clean deduplication between browser and server events. When those fields are incomplete or inconsistent, Meta loses confidence, attribution gets thinner, and optimization quality drops.
Actionable takeaway: Open Events Manager, check event match quality, and aim for 8.0+. If you’re below that, audit your Shopify or server-side integration before touching budgets.
3. 7-day click + 1-day view window
Is the default reporting setup giving you the cleanest picture?
Not always. In the accounts we reviewed, switching from a 7-day click default to 7-day click + 1-day view increased attributed conversions by 34% on average. More importantly, that reporting window was more accurate when cross-referenced with Shopify data.
That makes sense for DTC brands. A customer may see an ad during the day, remember the product later, and convert without clicking the original impression. A pure click window undercounts that behavior. Bear in mind, we’re not saying every view-through conversion is incremental. We’re saying this window gave a closer operational view of reality for the brands we analyzed.
Actionable takeaway: Compare 7-day click against 7-day click + 1-day view for the last 30 to 60 days, then validate both against Shopify revenue before making scaling calls.
The takeaway
So what should we ignore?
For day-to-day Meta decision-making, a lot of the noise sits in secondary metrics like brand lift, reach, and engagement. Those can be useful in specific tests, but if we’re trying to understand whether Meta is seeing enough signal to optimize profitably, the big three are VTA, CAPI match quality, and the right attribution window.
Get those three right and you’ll have 90% of the picture most ecommerce teams actually need.
So, over to you: Are you currently checking VTA, match quality, and attribution window side by side with Shopify, or is one of those still being overlooked in your account?
