# Gap between shop orders and measurement tool conversions

> A gap between real orders and conversions measured by GA4 or the Meta pixel is a known structural phenomenon: declined consent, ad blockers, multi-device journeys, orders placed outside the web checkout. It isn’t automatically a bug, but that doesn’t excuse skipping the check for an isolable technical cause on top.

- Source canonique : [https://allaux.fr/en/services/tracking/ecart-commandes-boutique-outil-mesure](https://allaux.fr/en/services/tracking/ecart-commandes-boutique-outil-mesure)
- Langue : EN
- Dernière mise à jour : 2026-09-30

## Direct answer

> Before hunting for a fault, take out of the count every order that never goes through the web funnel: entered in the back office, taken by phone, paid by manually validated bank transfer, plus cancelled and refunded orders. Then compare over an identical period and attribution window: a gap that is stable over time is structural, a gap that appeared overnight is technical.

## The situations I handle

- The back-office order count clearly exceeds purchases counted in GA4
- The gap seems to be widening over recent months
- The gap is more pronounced on mobile than desktop
- An ad channel looks underperforming even though its orders genuinely exist
- Some orders come from the phone or a payment outside the web checkout
- No way to tell if the observed gap is normal or reveals a real problem

## Telling the structural part apart from the abnormal part

1. **1. Isolate orders that can’t technically be measured** — I exclude phone orders, back-office entries, and offline payments (manually approved bank transfer) from the calculation: they never go through the front-end checkout and can’t generate an event. False gaps, not tracking losses.
2. **2. Compare the gap over a stable period, not one day** — I calculate the real orders / conversions ratio over several weeks with no known change. A stable gap over time points to a structural cause. A gap jumping sharply on a specific date points to a technical trigger around that date.
3. **3. Isolate the gap by acquisition channel** — I compare the ratio for direct, paid (Google Ads, Meta Ads) and organic traffic. Uniform across channels: likely structural. One channel abnormally worse: likely channel-specific, a badly placed pixel or broken tracking parameter.
4. **4. Isolate the gap by device type** — I compare mobile and desktop separately. A higher gap on mobile is often partly explained by journeys starting on one device and finishing on another, breaking attribution without any fault to fix.
5. **5. Check a specific checkout step, not just the final conversion** — In GTM preview, I check each checkout event fires in order (view_item, add_to_cart, begin_checkout, purchase). An event systematically missing at one step signals an isolable technical cause to fix on its own.
6. **6. Quantify the remaining structural share without aiming for zero** — Once technical causes are fixed, a residual gap is still expected: declined consent and ad blockers normally prevent sending data for those visits. Aiming for zero isn’t realistic.

## What’s structural, what isn’t

Part of the normal structural gap, not something to fix technically:

Consent declined by part of European visitors.

Ad blockers and privacy-hardened browsers.

Multi-device journeys, which break attribution without real data loss.

Orders placed outside the web checkout.

Points to an isolable, fixable technical cause instead: a missing event at a specific step, a channel with an out-of-line gap, or a sharp break date in the ratio.

## A gap of several tens of a percent can be normal

> Depending on European traffic share, ad blocker use and the site’s own multi-device journeys, a gap of that order between real orders and measured conversions remains known, not necessarily a sign of a problem.

## Go further

- **Consent mode in Europe** — How visitor consent affects data sent to Google. ([/services/tracking/mode-consentement-europe](/services/tracking/mode-consentement-europe))
- **Audience measurement without consent (CNIL)** — The criteria for exempted measurement, distinct from consent mode. ([/services/tracking/mesure-audience-sans-consentement-cnil](/services/tracking/mesure-audience-sans-consentement-cnil))
- **My conversions aren’t showing up** — The ordered method to check each technical step. ([/services/tracking/conversions-ne-remontent-pas](/services/tracking/conversions-ne-remontent-pas))
- **Ecommerce events that matter** — The expected events at each checkout step. ([/services/tracking/evenements-ecommerce-qui-comptent](/services/tracking/evenements-ecommerce-qui-comptent))

## FAQ

### What gap is considered normal?

No universal threshold: it depends on European traffic share, ad blocker use and the site’s own multi-device journeys. Several tens of a percent remain a known structural phenomenon.

### Should I aim for a zero gap?

No, that isn’t realistic given consent, ad blockers and multi-device journeys that structurally escape measurement.

### How do I know if part of the gap is still a bug?

By isolating the gap by channel, device and checkout step: uniform everywhere is structural; concentrated on one point is technical.

### Do phone orders count towards the gap?

They artificially inflate it if not excluded, since they can’t technically generate any measurement event.

### If the gap jumps sharply on one date, is that necessarily a bug?

That’s the most reliable signal of a technical cause to look for around that date: a theme update, a consent banner change, a new module.