Why Admaxxer and Meta/Google Disagree on Conversions

TL;DR: Admaxxer measures with a deterministic first-party pixel + server-side payment events. Meta, Google, and TikTok measure with their own logged-in graphs, modeled conversions, view-through windows, and cross-device fingerprinting. The two will not match exactly — and they should not. A 5–20% gap in either direction is normal. This guide explains why, and gives you a five-step reconciliation that ends with a defensible answer to "which number is right?"

Why this happens — five real causes

1. Attribution windows are different by default

Each platform counts a conversion as theirs based on when, in time, an ad touch happened relative to the purchase. The defaults are not the same:

Meta defaults to 7d-click + 1d-view in Ads Manager. A purchase that happens 5 days after a click counts; a purchase that happens 4 days after only seeing the ad does not (unless it's within the 1-day view window).
Google Ads defaults to 30d-click with data-driven attribution. Search and YouTube can claim a conversion up to 30 days after the click.
TikTok defaults to 7d-click + 1d-view in TikTok Ads Manager.
Admaxxer uses a blended first-party pixel with last-non-direct touch by default, and lets you toggle 1d / 7d / 30d click windows on every report. Our blended pixel does not deduplicate against view-through events at all (we do not see them) — so for the same window setting, our number is typically lower than Meta's because we never count view-throughs.

If you compare a Meta dashboard at 7d-click + 1d-view against an Admaxxer dashboard at last-click 7d, the platforms are measuring different things. Always align windows first before reading the gap.

2. Modeled conversions are not real events

A growing share of every platform's conversion count is modeled — statistical estimates that fill the gap left by lost signals (iOS 14.5+ ATT, browser tracking restrictions, ad blockers). Modeled conversions are real conversions that happened — the platform just did not directly observe them, so it estimates them from cohort behavior.

Meta CAPI + AEM (Aggregated Event Measurement) sends server-side events for iOS 14.5+ users so Meta can rebuild attribution. Match rate matters here — see our CAPI Match Rate guide.
Google Enhanced Conversions hash-matches user PII (email, phone) at conversion time so Google can join those events back to logged-in users it lost track of.
Admaxxer's pixel never models. Every event in our dashboard is a real, deterministic event from the user's browser or your server. We round nothing. We estimate nothing. The trade-off: our raw number is "more true" but typically lower, because we silently lose anyone the pixel could not reach (CSP-blocked, ad-blocked, hostile networks).

3. Cross-device attribution we cannot replicate

If a user clicks an Instagram ad on their phone, then converts on their laptop the next day, Meta can stitch those two devices together because both are logged into the same Facebook account. Admaxxer cannot. Our pixel uses first-party cookies + fingerprint hashing, but we have no logged-in social graph. If you see Meta reporting a purchase that has no matching session in Admaxxer, cross-device is a likely cause.

This shows up most in:

Mobile-discovery, desktop-purchase categories (apparel, furniture, B2B SaaS).
iOS users with ATT denied — Meta uses logged-in graph + AEM modeling; Admaxxer sees only the device where the pixel fired.
Multi-user households (someone sees the ad, their partner buys it).

4. iOS 14.5+ App Tracking Transparency (ATT)

When a user denies the ATT prompt in the Facebook or Instagram app, Meta can no longer pass the device's IDFA back to the conversion event. Meta compensates with AEM modeling — but that modeling is, by definition, an estimate. Same store, same week, same ads:

Pre-iOS 14.5: Meta and Admaxxer typically agreed within ±5%.
Post-iOS 14.5: Gaps of 10–25% are common, with Meta's number usually higher (more modeling = more attributed conversions).

5. CAPI / Enhanced Conversions Match Rate

Server-side events are only useful if the platform can match them back to a known user. CAPI Match Rate (Meta) and Enhanced Conversions match-rate (Google) measure the percentage of purchase events the platform was able to join with one of its users. A low match rate (below 60%) means the platform is starved of signal and will lean even harder on modeling. Improving match rate — by passing email, phone, FBC, FBP, and a clean external_id — closes the gap between platform-reported and Admaxxer numbers. See CAPI Match Rate in the Metric Glossary for the exact definition and how Admaxxer measures it.

Step 1 — Align the time window

Before anything else, make sure both reports cover the same window in the same time zone. Common silent mismatches:

Meta defaults to the Ad Account's time zone (often Pacific). Admaxxer defaults to the Workspace time zone.
"Last 7 days" in Meta Ads Manager means rolling 7 days ending yesterday in Meta's reporting clock; Admaxxer's "Last 7 days" follows your dashboard's chosen timezone.
Click vs conversion timestamping: if you compare Meta's "conversions on day X" to Admaxxer's "conversions on day X", that's apples-to-apples. If you compare Meta's "spend on day X" to Admaxxer's "conversions on day X", you'll always see drift on weekend / Monday boundaries.

Pick a single 7-day window in the same timezone. Run both queries again. Often the "discrepancy" disappears here.

Step 2 — Align the attribution window

In Meta Ads Manager, set the comparison window column to 7-day click only (drop the 1-day view). In Admaxxer, set the dashboard window to last-click 7d. Now you are measuring the same thing on both sides.

Numbers should be much closer. The remaining gap is real measurement disagreement — not a configuration mismatch.

Step 3 — Inspect CAPI Match Rate

Open Dashboard › Analytics › CAPI Match Rate. Targets:

≥80% — you are at or near the ceiling. Meta's modeling is well-fed; their numbers will be close to ours.
60–80% — typical. Expect 5–15% Meta-over-Admaxxer drift from modeling.
<60% — bad. Meta is leaning hard on modeling. Improve identity passing (email, phone, fbc/fbp, hashed external_id) before you trust either number.

See Metric Glossary › CAPI Match Rate for the calculation. Hyros markets this as their core differentiator — Admaxxer ships the same metric for free in every plan.

Step 4 — Check for "spend without sessions"

If Meta reports spend on a campaign that produced zero sessions in Admaxxer, the pixel is not firing for that traffic. This usually means:

The ad clicks are landing on a page that does not have the Admaxxer script (a partner microsite, a UTM-stripped subdomain, or a redirect chain that drops the click ID).
The destination page has a CSP that blocks admaxxer.com. See CSP errors blocking the pixel.
The user is on a privacy-extreme browser (Brave, Firefox + uBlock, Safari with Hide IP) and the pixel never executes.

For each campaign with a high "spend without sessions" rate, fix the install before reading attribution numbers. A 0% session rate makes the gap to Meta look enormous; a 95% session rate makes it look normal.

Step 5 — Triangulate against Shopify (or your source-of-truth)

The customer's checkout system is the only deterministic ground truth — every purchase must pass through it. Compare:

Total orders in Shopify for the window.
Total purchases in Admaxxer for the window.
Total purchases in Meta for the window.

If Shopify says 51, Meta says 47, Admaxxer says 52, the truth is roughly 51 — and now you can read the gaps:

Admaxxer is 1 over Shopify. This is normal — sometimes the pixel fires twice (browser back-button, retry on slow network); Admaxxer dedupes by order_id when present, but pixel-only events without a server-side echo can stutter.
Meta is 4 under Shopify. This is also normal — Meta only attributes purchases it can tie back to its ad-clicks; some of Shopify's 51 came from organic, email, direct, or another channel Meta cannot see.

If Admaxxer is materially under Shopify (more than ±5% off), see Sales mismatch troubleshooting. If Meta is reporting purchases that Shopify cannot find at all, Meta is over-modeling — and you should pull back trust in Meta's column for that campaign.

Worked example

A real reconciliation we ran for a DTC supplement brand, week of 2026-04-21:

Source	Purchases	Window	Notes
Meta Ads Manager	47	7d-click + 1d-view	Default. Includes view-throughs and modeled events.
Meta Ads Manager (aligned)	38	7d-click only	Stripped view-throughs. Closer to Admaxxer.
Admaxxer	52	last-click 7d	Includes purchases attributed to non-Meta last-clicks (email, organic).
Admaxxer (Meta only)	40	last-click 7d, channel=meta	Filtered to last-click attributable to Meta. Now apples-to-apples with row 2.
Shopify orders (paid)	51	order date in window	Ground truth. All orders, any source.

Reading the table: the real story is 51 paid orders. ~40 of those had Meta as their last paid touch (Admaxxer + Meta agree at this filter). The other 11 came from email (Klaviyo), organic search, and direct. Meta's headline 47 included 9 view-throughs + modeled conversions that we cannot independently verify — they may be real but we are not counting them.

The brand chose to optimize on the aligned 7d-click number (38 / 40 — nearly identical) because that's the most defensible signal across both platforms. View-through and modeled adds give Meta's bidder a fuller signal, so we did not turn them off in Ads Manager — we just stopped reporting against the inflated number.

When to trust which source

Trust Shopify (or Stripe / Paddle / your checkout) for total orders and total revenue. Always.
Trust Admaxxer for first-party session-level behavior, last-click attribution, blended MER, cohort LTV, and channel-mix. We see every traffic source on equal footing, including organic, email, and direct.
Trust Meta / Google / TikTok for in-platform optimization signal — feed CAPI / Enhanced Conversions, let their bidder model what it needs to model. Don't use their headline conversion count as your reporting number; use it as input to the bidder.
When all three disagree, your source-of-truth ladder is: checkout › Admaxxer (last-click) › platform (modeled).

Diagnostic queries

Run these against your Admaxxer workspace to surface gaps faster:

# Sessions per Meta campaign vs spend reported by Meta
# (in Admaxxer dashboard) Filter: source=facebook, last 7d
# Compare against Meta Ads Manager, same campaign IDs, same window.

# Spend-without-sessions rate per campaign
SELECT
  campaign_id,
  SUM(spend) as spend,
  SUM(CASE WHEN session_count = 0 THEN spend ELSE 0 END) as spend_without_sessions,
  ROUND(100.0 * SUM(CASE WHEN session_count = 0 THEN spend ELSE 0 END) / NULLIF(SUM(spend), 0), 1) as pct
FROM ads_meta_with_attribution
WHERE date >= today() - 7
GROUP BY campaign_id
HAVING spend > 0
ORDER BY pct DESC;

When to escalate

Open a support thread (/support) when:

Aligned windows still show >30% gap between Admaxxer and the platform.
CAPI Match Rate is <40% and you've already ensured email/phone are being passed.
You're seeing Meta-attributed purchases that have no corresponding Shopify order at all (this is platform over-attribution and should be raised with your Meta rep, not us — but we can help compose the diagnostic).

Include screenshots of both reports with timestamps, the windows used, and your CAPI Match Rate. We will help you isolate which of the five causes above is dominant.