A Shopify merchant wanted to understand which numbers to trust when evaluating paid acquisition. Store reporting, analytics and ad platforms appeared to tell different stories. A low blended conversion rate made the store look weak, while strong platform ROAS suggested advertising was performing exceptionally well.
The question was practical: which differences reflected measurement problems, and which came from the way each system counted or assigned credit?
What I reviewed
I reviewed Shopify orders, sales definitions and traffic-quality reporting; GA4 implementation and ecommerce reporting; Meta browser/server Purchase signals and deduplication; and Meta and Pinterest attribution breakdowns.
The main commercial comparison used an aligned four-week reporting period. Historical Shopify traffic quality and the later GA4 anomaly check used separate windows. Pinterest reporting was in UTC, so its totals supported decision-level comparison rather than an exact transaction-by-transaction timezone match.
1. The blended conversion rate hid the traffic-quality problem
In the historical Shopify window, roughly 82% of sessions were classified as bots by Shopify. That segment recorded no completed checkouts. Including it in the denominator substantially changed how the conversion rate looked.
Same period, different denominators. The chart does not show an improvement caused by the audit.
This was evidence from Shopify’s own classifier. It did not independently identify the network origin of the sessions, and it did not mean a similar percentage of revenue or ad spend had been lost.
2. Healthy purchase transmission did not make Meta attribution causal
Meta’s server-side Purchase count matched the Shopify order count in the reviewed period. Browser/server identifiers, value coverage and deduplication diagnostics did not support a material duplicate-purchase inflation finding.
That aggregate alignment was useful evidence about transmission. It was not proof that Meta advertising caused every purchase, nor a transaction-by-transaction certification.
Meta’s standard attributed ROAS was approximately 40x, compared with approximately 17x in the seven-day-click view. More than half of the standard attributed value came from the one-day-view component. The reporting rules explained an important part of the apparent performance story.
Decision: retain the healthy purchase implementation and review standard attribution alongside the click-based view. Neither view establishes incremental return on its own.
3. Pinterest’s headline ROAS depended heavily on views
In the aligned Pinterest report, roughly 98% of attributed order value came from view-through attribution. The headline ROAS was about 18x, while platform-reported click-through order value divided by spend was about 0.4x.
Those figures describe different attribution views. They do not prove that Pinterest contributed no value, and the click-based ratio is not an audited incremental ROAS. Applying an arbitrary percentage discount would have hidden the actual breakdown.
Decision: show total, click-through and view-through value side by side before using the platform’s report in budget discussions.
4. GA4 was useful, but incomplete
GA4 recorded approximately 95% of the Shopify order count. The audit documented the missing purchase-count coverage without assigning a cause that the available evidence could not establish. Revenue definitions also differed, so GA4 purchase revenue was not forced to match Shopify Total or Net sales.
A geographic traffic anomaly persisted after a measurement-ID change. The report distinguished the client’s account of an earlier ghost-traffic issue from independently observed post-change reporting. It did not establish that the GA4 anomaly was the same population as Shopify’s bot-classified sessions, or that Cloudflare had failed.
A clear role for each source
Shopify
The commercial control for completed orders and explicitly defined sales metrics.
Meta
An optimization and attribution signal, with standard and click-based results shown together.
A channel report that requires a visible click-through and view-through split.
GA4
A behavioral and acquisition cross-check, with purchase gaps and traffic anomalies flagged.
The delivered report included a measurement baseline, cross-platform reconciliations, prioritized findings and a recommended weekly scorecard. It also identified the questions that still needed investigation.
One rule was essential: do not add Meta and Pinterest attributed revenue and label the result unique or incremental store revenue. The same sale can receive credit in more than one platform.
What this case demonstrates
A measurement audit can separate an implementation question from an attribution question before a team changes healthy tracking or makes a budget decision from an ambiguous headline number. Here, the documented result was a clearer basis for decisions and a focused investigation plan.
This case does not claim increased sales, realized savings, resolved bot traffic or a completed GA4 repair. The precise cause of the GA4 purchase gap and continuing traffic anomaly remained unresolved at report delivery.
Source: the engagement’s final forensic measurement and attribution report. Client identity, dates, account identifiers and commercial amounts are withheld; selected figures are rounded. The original report is confidential and is not published here.
Do your dashboards tell different stories?
Start with a free 30-minute Shopify Measurement Diagnostic. We can clarify the reporting concern and identify what needs validation first.