Why Meta reports a slightly different event count or value than actual

If the difference is small and inconsistent in direction, your integration is most likely fine. Meta deliberately adds privacy-preserving noise to low-volume event counts and values, so figures for a single ad, day, or breakdown row with only a few conversions are approximate by design. Once you exceed 10-30 events in a report cell, the noise mostly averages out and Meta's numbers approach actual conversion numbers.

Why Meta adjusts small numbers

Meta appears to apply tiny adjustments to its ads reporting so that no individual person can be identified from the Ads manager report. (Note that the report includes social-demographic segmentation, such as sex or age.) Empirically, this appears as a small random amount is added to or subtracted from counts in a way where no Ads Manager report cell can be cross-referenced with a specific conversion event. The adjustment is about the same size whether the true count is 1, 3 or 30, which is why it visibly distorts the lower event counts and is much less visible on volume. For example:

• Individual ads with a handful of conversions show counts that are slightly off (by a few conversions on low events count - where this may appear as as 30-100% difference - and by 5-7% when many events are tracked).
• Purchase value at low volume doesn't match any real order, so single-conversion ROAS figures are unreliable.

Meta doesn't document this behaviour directly. It does say is that some conversions in Ads Manager are statistically modelled, and that its reporting is built on various privacy techniques so no individual can be identified. That's consistent with the observed small privacy adjustments described above, but it's important to add that Meta has never publicly confirmed how exactly these adjustments work; the figures reported above are empirical observations.

How to tell noise from a real tracking problem

Compare at the coarsest level first: the whole account or campaign over a full week or month, against Able's Transaction date attribution report for the same event and period. If the totals are within a few percent, the integration is healthy and the finer-grained gaps are noise.

If the totals are materially off, more volume won't fix it. Noise goes both ways, so a gap that is consistently one-directional across ads points to a tracking issue instead: consistent under-reporting means events aren't reaching Meta or aren't being matched (check your Event Match Quality); consistent over-reporting usually means Pixel and CAPI events aren't deduplicating or view-through attribution is inflating results (see Meta Ads over-reporting conversions).

Also make sure you're comparing like with like: same event name, same attribution window, and the same date basis. Meta reports on click or impression date, while your order system reports on transaction date. (Able can report on both - make sure you choose the right attribution option in the report settings.)

What to do about it

Evaluate performance at the level where conversion counts are at least in the dozens, usually campaign or ad set totals over at least a week, rather than analyzing a 2-versus-4 difference between individual ads. For exact per-event figures, use Able's own reports: Able records every conversion and the Meta click it came from without any adjustment, so you can reconcile at whatever granularity you need while relying on Meta's aggregates for the in-platform view.