Should I trust Meta ads conversions?
Trust Meta's conversions as Meta's claim about which sales its ads touched, not as sales the ads caused. They usually hold up for comparing ads under one attribution setting, and mislead when added to other platforms' numbers or used to set budget.
By Joris van Huët, Founder & CEOUpdated 7 min read
Run the numbers for your store: the free break-even ROAS calculator.
Trust them as Meta's claim, not as a count of sales Meta caused. They usually hold up for comparing ads that share one attribution setting. They mislead if you add them to other platforms' numbers or use them to set budget. Check them against Shopify orders, then against people who saw no ads.
What one store's data shows
The usual answers pull in opposite directions: "Meta inflates everything" and "Meta sees what GA4 misses". One store's export shows why trust depends on the question you put to the number.
One store's anonymised GA4 export, 1 January 2024 to 21 August 2026. It holds shares of revenue only: no ad spend, no order counts.
| What the export shows | Share of revenue | Source cell |
|---|---|---|
| Paid Social, in last click, first click and touched views | 0.0% | Channels sheet, Paid Social row |
| All channels in the touched view, added up | 110.4% | Channels sheet, Touched column total |
| Journeys with 1 touch (0.5 days to buy) | 79.5% | Journeys sheet, 1 touch row |
| Journeys with 10+ touches (16.0 days to buy) | 3.0% | Journeys sheet, 10+ touches row |
Trust needs a second witness, and in this store GA4 cannot be one. Paid Social sits at 0.0% of revenue on the Channels sheet, in last click, first click and touched alike. If this store ran Meta ads, no row in the export could confirm a single Meta purchase. Trust would rest on Meta's count alone. The export has no spend column, so it cannot even say whether ads ran.
Next, the touched view. On the Channels sheet it adds up to 110.4%, because a journey that touched two channels counts once for each. One tool, watching every channel, already hands out more credit than there was revenue. Meta's report runs on that rule for its own ads. Meta credits a purchase made within a set number of days after someone viewed or clicked an ad. Nothing in that rule asks what else the buyer clicked. So a Meta purchase can be real and still be your email's sale too.
Journeys of 1 touch carry 79.5% of revenue on the Journeys sheet and close in 0.5 days. Most of this store's revenue came from people who arrived once and paid within half a day. Buyers that quick may have decided before they ever arrived. If one of them saw a Meta ad that morning, Meta's 1-day view-through setting can count the sale. Whether the ad changed anything is a question no count can answer.
At the other end, journeys of 10 or more touches hold 3.0% of revenue on the Journeys sheet and take 16.0 days to buy. If a Meta link click opened a journey like that, a 7-day click window would shut long before the purchase. The same report that over-credits quick buyers can miss slow ones.
What the export cannot show: Meta's own purchase count, ad spend, orders, or whether any ad caused a sale. Treat it as one store's record, not a benchmark for yours.
Why does "Meta inflates everything" mislead?
It treats trust as one switch. Meta's count answers some questions well and others badly, so set the switch per question.
Which ad sold more? Usually a fair fight, if both ad sets use the same attribution setting. One rule applies to both, so the comparison holds. Meta's own help page warns against comparing results across ad sets with different attribution models. Even then, retargeting flatters itself: people who already filled a cart may buy with or without the ad.
How much revenue did Meta bring in? Here the count is weak. It credits any purchase inside its window, whoever else touched the buyer. Part of it is an estimate, too: Meta says it models conversions where data is missing or partial. On Shopify, the purchase value Meta receives is the order's total price, including duties, taxes and discounts. Set that against your net sales and Meta looks richer than it is.
The count can also run low. Meta notes that its pixel may not fire when a shopper uses an ad blocker. That leaves Meta's reports short of your own records.
Should I spend more? The count cannot answer this one at all. That takes a comparison with people who saw no ads.
What can Meta's count not tell you?
Whether the sale would have happened without the ad. A purchase inside the window shows timing, nothing more. Cause needs a group that could not see the ads.
Meta offers two routes. Its incremental attribution model uses machine learning to predict whether an ad caused each conversion. That is a better question, still marked by the seller of the ads.
The stronger route is a Conversion Lift test. Meta splits your audience into a test group that can see your ads and a control group that cannot. The gap in conversions between the two is the lift. Meta even lists "Does my current attribution model align with the results of a controlled experiment?" among the questions a lift test answers.
As a guide, Meta asks for a campaign started in the past year, with $5,000 USD or more in spend and at least 500 conversions. Below that bar, a geo holdout does the same job with regions in place of people.
What to do this week
- Check the purchase signal Meta works from. In Events Manager, select your dataset, find Purchase and click View details, then open the Event coverage and Event matching tabs. Pass: coverage meets Meta's 75% target and the match score sits above 5, the bar for self-serve lift tests. Fail: either one falls short, so fix the data before you judge any campaign.
- Add up every platform's claim for one week. Take Meta's purchases, Google Ads conversions and any email credit for the same days. Then count Shopify orders: go to Orders in your Shopify admin and use the Date filter. Pass: the claims add up to less than your orders. Fail: they add up to more, so some sales are counted twice and no single platform's number is your revenue.
- See whether you qualify for a Conversion Lift test. Look for a campaign that started in the past year with at least $5,000 USD in spend and 500 conversions. Pass: one qualifies, so create the test in Meta's Experiments tool. Fail: none does, so plan a geo holdout by region instead.
Check the homework. Your GA4 Attribution paths export already holds the evidence. Causality Engine reads that one file and shows what each channel caused next to what last-click gave it, in 1 to 2 minutes, for €99 once (excluding VAT), refundable within 30 days. Check the homework
Sources, 1 October 2026: About actions attributed to your ad (Meta Business Help Center); About attribution models and attribution settings (Meta Business Help Center); About Meta's Modeled Conversions (Meta Business Help Center); Understand how results are sometimes calculated differently (Meta Business Help Center); Facebook data sharing (Shopify Help Center); About conversion count differences between Meta Ads Reporting and third-party reporting tools (Meta Business Help Center); About incremental attribution (Meta Business Help Center); About Conversion Lift (Meta Business Help Center); View server event details in Meta Events Manager (Meta Business Help Center); Filtering orders (Shopify Help Center).
Related answers
Frequently asked questions
Are Meta's modeled conversions real purchases?
Partly. Meta says it models conversions where data is missing or partial, estimating them from people it can measure. The purchases may well have happened, but which ad gets them is Meta's estimate. Ads Manager flags such results with an in-product message, so hover over it before you quote the number.Can I trust Meta's incremental attribution column?
Use it to rank campaigns, not to settle what Meta added. Meta's incremental model uses machine learning to predict whether an ad caused each conversion, so it is still Meta marking its own work. Check its total against a Conversion Lift test or a geo holdout before you budget on it.What does Meta require for a Conversion Lift test?
As a guide, a campaign that started in the past year with $5,000 USD or more in spend and at least 500 conversions. Self-serve tests also need good signal quality, such as Conversions API events with a match quality score above 5. The test itself costs nothing extra.
Go deeper: Incrementality testing, explained.
Sixty-second versions of these ideas: Causality Engine on YouTube Shorts.
Keep reading
Terms in this article
- AttributionAttribution identifies user actions that contribute to a desired outcome and assigns value to each. It reveals which marketing touchpoints drive conversions.
- Attribution ModelAn Attribution Model defines how credit for conversions is assigned to marketing touchpoints. It dictates how marketing channels receive credit for sales.
- Control GroupControl Group is a segment of an audience intentionally not exposed to a marketing campaign, used to measure the campaign's true causal impact.
- ExperimentsExperiments are scientific procedures that test hypotheses or demonstrate facts. In marketing, experiments like A/B tests determine the causal effect of campaign changes, enabling data-driven decisions.
- IncrementalityIncrementality measures the true causal impact of a marketing campaign. It quantifies the additional conversions or revenue directly from that activity.
- Incrementality TestingIncrementality Testing measures the additional impact of a marketing campaign. It compares exposed and control groups to determine causal effect.
- Machine LearningMachine Learning involves computer algorithms that improve automatically through experience and data. It applies to tasks like customer segmentation and churn prediction.
- RetargetingRetargeting is online advertising that targets users who have previously interacted with your website or content. Attribution analysis shows the causal role of retargeting in driving conversions and improving ad spend.