Measurement
Part of Does ecommerce SEO measurement need dashboards or evidence?
Five ecommerce SEO key metrics and their data sources
The five ecommerce SEO key metrics worth reporting, the system each figure comes from, and the label that stops a modelled number passing as observed.
What to take away
- Report five figuresorganic sessions, non-brand organic revenue, conversion rate, average order value and assisted conversions.
- Give every figure a named source system and an extraction date.
- Platform revenue is the sales record; GA4 explains the sessions around it.
- Search Console shows demand, not sales, and only for queries Google reports.
- Label modelled figures as modelled, and show the consent rate beside them.
Organic sessions: the visible share of visits
Organic sessions come from GA4, and they count only the visits that analytics could see. Consent mode and cookie banners suppress part of the traffic that older Universal Analytics accounts counted freely, so the figure is a floor rather than a total.
GA4 models some of that gap, and modelled figures are estimates, not observations. Consent rates also vary by audience and device. Report the consent rate next to organic sessions so readers can judge how much of the total is visible.
Non-brand organic revenue: the headline figure
Non-brand organic revenue is the figure to lead with. Brand queries often grow on their own, and mixing them into one line flatters performance. Split the two before you claim progress, and keep brand out of every growth claim.
Take the revenue itself from your platform. Server-side revenue reflects the order as recorded, regardless of cookies, so it is usually the most reliable sales figure. GA4 adds the session context: source, landing page and device. Where the two disagree, state both and explain the gap instead of picking the flattering one.
Conversion rate and average order value as a pair
Read conversion rate and average order value together. A rising conversion rate with a falling basket size usually means discounts, not stronger demand.
Either figure alone can mislead. Before anyone reports a gain in one, check what happened to the other, and write both beside the same period in the pack.
Assisted conversions and the attribution setting
Assisted conversions show which pages help before purchase. They depend on attribution settings, so document the model you use next to the figure.
A change of model moves the number without any change in what shoppers did. That is why the model name belongs in the report header, not in a footnote nobody opens.
Search Console: what it can and cannot add
Search Console supplies impressions, clicks and average position, but only for queries Google chooses to report. Treat it as directional evidence about demand, not a complete sales ledger.
Impressions show how often a page appeared, and clicks show how often it was chosen. Average position summarises where it appeared. None of the three says what the visitor bought, which is why revenue always comes from the platform.
One source and one limit for each figure
Keep the reporting pack short enough that someone actually reads it. The table below sets each figure against the system it comes from and the thing it cannot tell you.
Core Ecommerce SEO Metric Set
Metric
- Organic sessions
- GA4 acquisition
- Non-brand organic revenue
- GA4 brand filter
- Conversion rate
- GA4 ecommerce events
- Average order value
- Store platform or GA4
- Assisted conversions
- GA4 attribution paths
Source
- Organic sessions
- Non-paid visibility growing?
- Non-brand organic revenue
- Did search sell?
- Conversion rate
- Traffic matches buying intent?
- Average order value
- Right basket size?
- Assisted conversions
- Which pages help before purchase?
What it answers
- Organic sessions
- Non-brand organic revenue
- Conversion rate
- Average order value
- Assisted conversions
| Figure | Source system | What it cannot tell you |
|---|---|---|
| Organic sessions | GA4 | Visits hidden by consent choices |
| Non-brand organic revenue | Platform revenue, with GA4 for context | Brand demand, which is split out |
| Conversion rate and average order value | Platform orders and GA4 sessions | Whether a lift is discounting or demand |
| Assisted conversions | GA4, under the attribution model in use | A cause: the number moves when the model does |
| Impressions, clicks, average position | Search Console | Sales, or queries Google does not report |
Lawful measurement under UK GDPR and PECR
Analytics cookies and similar technologies fall within UK GDPR and PECR duties, so a measurement setup is a compliance topic, not only a marketing one. The ICO finalised its storage and access technologies guidance on 29 April 2026, and it covers cookies and tracking pixels.
The ICO also describes a statistical purposes exception, which applies when the sole purpose is collecting statistical information about the use of a service.
Record what is stored, why, and for how long. Choose a retention period for event data that matches your privacy notice. If the notice promises fourteen months, your reports should not rely on three years of history.
Dating the figures and keeping the evidence
Record the extraction date for every figure you quote. A number without a date cannot be checked or reproduced. Keep the raw exports behind the period you report on, because a screenshot of an aggregate loses the detail needed to rebuild a disputed total.
For market context, the business population estimates for 2024 put the number of UK private sector businesses at 5.6 million. That figure frames how crowded a category may be. It is context, not a target. Compare the store against a written baseline, built as in defensible benchmark research.
After a measurement reset
A reset splits brand from non-brand, records the consent rate beside sessions, puts platform revenue beside GA4, fixes a dated monthly cut-off and names an analyst for each metric. Write the reset date into the report header so later readers know which rules applied.
Measurement Reset: Before and After
Before
- Brand and non-brand
- Combined in one line
- Consent
- Assumed full traffic
- Revenue source
- GA4 only
- Reporting date
- Ad hoc
- Ownership
- Unclear
After
- Brand and non-brand
- Split, brand excluded
- Consent
- Consent rate recorded
- Revenue source
- Platform plus GA4 context
- Reporting date
- Fixed monthly cut-off
- Ownership
- Named analyst per metric
Expect the first month after a reset to look worse. Cleaner data usually is. Agree in advance who owns each metric and what happens when a figure breaks its expected range. For change logs, attribution and dashboards, see the measurement and reporting guide.
Common questions
How many metrics should a monthly report carry?
Five to seven core figures, with supporting detail available on request. Longer packs tend to hide the two numbers that matter.
Can I report modelled GA4 revenue as fact?
No. Label it as modelled or estimated, and show the consent rate that produced it. Observed platform revenue should sit beside it.
Do I need consent for analytics on a UK store?
Generally yes, unless the storage is strictly necessary for a service the visitor requested. The ICO guidance covers the exemptions in detail.
What if my figures disagree across tools?
Publish the discrepancy and your reconciliation method. A documented gap is more credible than a single tidy number nobody can reproduce.
