Of real Shopify orders recorded
Reconciled against orders that actually happened, not modelled.
Case study · Incapto
GA4 was not measuring less. It was measuring another business.
Same site, same days, two tools running at once on Incapto's Shopify store. First a reconciliation against real orders, then six differences — and the last one changes where the media budget goes.
incapto.comOf real visits missing from GA4
64,501 visits that appeared in no report.
Of real pageviews missing from GA4
212,422 pageviews, concentrated in the traffic it never saw.
The orders were real.
The traffic behind them was not.
Incapto knew exactly how many orders its Shopify store had taken. What it could not establish was how much of the traffic that produced those orders was reaching its analytics — and Consent Mode makes that gap impossible to size from inside GA4, because the traffic that is not measured leaves no trace to count.
So the team stopped arguing about percentages and ran both tools on the same site for the same days. The first question was not which one measured more. It was which one could be checked against something that indisputably happened.
“Consent Mode left us with a structural blind spot: we knew there was traffic we were not seeing, but we had no way to size it.”
Rosa Tomàs · B2C Acquisition Manager · Incapto
00
Before believing anything: the numbers match the till.
Real orders and revenue from the Shopify online store · 14 Jun → 31 Jul 2026
96%
of real orders were recorded
97%
of real revenue was recorded
Shopify orders happened. They are not estimated and not modelled. For every 100 real orders, Sealmetrics recorded 96. That is what makes it usable as the reference for everything that follows.
01
GA4 was not seeing 29 of every 100 visits.
Visits · 14 Jun → 31 Jul 2026 (48 days)
More than 64,000 visits that appeared in no report. That is the equivalent of having analytics switched off for 14 of those 48 days.
02
And it was not seeing 45 of every 100 pages.
Pageviews · 14 Jun → 31 Jul 2026
It is short 29 of every 100 visits, but 45 of every 100 pages. The missing visits are not ordinary visits.
03
The traffic it cannot see is the traffic that browses most.
Average pages per visit · 14 Jun → 31 Jul 2026
212,422 pageviews spread across the 64,501 visits GA4 did not record — twice the browsing depth of the visits it does record, measured as an aggregate ratio across the period.
04
The loss is not the same in every channel.
Extra traffic Sealmetrics sees, channel by channel · 28 Jul → 6 Aug 2026
Direct
+11%
Referral
+24%
Email
+26%
Paid campaigns
+37 to +52%
Organic search
+62%
Affiliate
+73%
Organic social
+133%
Channels carrying people who already know the brand barely move. The ones bringing new people in from an external click lose between three and twelve times more. This is not fixable by multiplying the reports by a correction factor.
05
And the “no idea where this came from” bucket disappears.
Visits with no origin you can decide on · 28 Jul → 6 Aug 2026
In GA4 that is 14 of every 100 visits: nine assigned to no channel at all, and five more in residual channels that point at no actionable origin. In Sealmetrics it is 3 in every 1,000. These are visits that exist, but that you cannot decide anything with.
06
The result: two different pictures of one business.
Where the traffic appears to come from · 28 Jul → 6 Aug 2026 · rounded
GA4
50%18%10%8%14%
Sealmetrics
62%16%13%9%
And this is the GA4 bar put back on the real scale: the same channels, but calculated over all the traffic that existed, not just the traffic GA4 managed to see.
GA4, on the real scale
41%15%8%6%11%19%
- Paid campaigns
- Direct
- Organic search
- Email, social, affiliate, referral, AI
- Unknown origin
- Traffic GA4 never measures
For GA4, campaigns are half the business. For Sealmetrics, close to two thirds: 12 points of difference exactly where the budget is spent. Direct is not growing — it loses less than everything else, which makes it look more important than it is. Put back on the real scale, the GA4 picture has a 19% hole in it; add the 11% with no known origin and close to a third of the real traffic supports no decision at all.
Method
Reconcile against the till,
then read the difference.
Nothing here depends on trusting one vendor over another. The method is to anchor both tools to a number that is not produced by either of them — the orders the store actually took — and then inspect where the two diverge.
- 01
Take the real order total
Use the eCommerce platform's own orders for the period. Online store only: exclude subscriptions, physical retail and manual admin orders, which have no web visit behind them.
- 02
Measure both tools in parallel
Leave the existing analytics in place and run the second measurement layer over the same days on the same site.
- 03
Reconcile before comparing
Check each tool against the order total first. A tool that cannot match the till is not a reference for anything else.
- 04
Break the gap down by channel
The loss is not uniform. Find which channels it concentrates in, because those are the ones whose budget is being decided on the wrong number.
Result
Same 48 days.
A different investment case.
The reconciled view moves paid campaigns from 50% to 62% of measured traffic — a 12-point difference in the one line of the report that determines media allocation.
Incapto did not change its stack to get a nicer number. It changed the base the number is calculated on, and the channels that were being under-credited are the ones bringing new customers in.
Before you ask
This is what was measured, not what was earned.
There is no ROI figure here, no incremental sales, no revenue attributed to Google Ads clicks. None of those appear because none of them were measured. What these six comparisons change is the base on which the investment decision is made — not the return it produced.
Methodology
- Figures are rounded in the charts. Unrounded: 157,844 and 222,345 visits · 256,005 and 468,427 pageviews in the June–July window; 40,426 GA4 sessions and 50,069 Sealmetrics entries in the July–August window.
- Points 00 to 03 cover 14 Jun → 31 Jul 2026 (48 days). Points 04 to 06 cover 28 Jul → 6 Aug 2026, after the channel-grouping adjustment Incapto made. The two periods are not compared against each other.
- Paid traffic is shown aggregated and as a range because GA4 groups part of the spend under Cross-network, which Sealmetrics splits between search and social. The lower bound assumes the calculation most favourable to GA4.
- The Shopify reconciliation uses the Online Store channel only: recurring subscriptions, physical retail and manual admin orders are excluded, because none of them has a web visit behind it. Sealmetrics reconciles 95.71% of orders and 96.53% of revenue.
- Sealmetrics and GA4 use different measurement methodologies. This case does not claim that Sealmetrics measures 100% of traffic — it reports what each tool recorded over the same days, against an order total neither of them produces.
Compare with your data
You already have
the two numbers this needs.
Take the real orders from your eCommerce platform and the ones your analytics reports for the same period. If they do not match, the next question is no longer how much traffic you are missing — it is which channels you are missing it from.