Playbook + practice template · Free to use
Ecommerce SEO measurement playbook
Build a comparable page-group report and turn uncertain performance changes into testable next steps.
Practice datasets are fictional and labeled as such. Real store captures are dated observations with their own sources and evidence limits.
Write the decision the report must support
Use this playbook to assess a release or choose the next content or technical task. State the decision first: did the repaired category group regain relevant search visits, or should the team investigate another cause? Bring Search Console exports, consented analytics, catalog availability and a dated change log.
Choose one market, device scope and page group. Define the exact comparison periods before looking for a favorable result. Include days of week and seasonal context. An all-site average can hide gains in categories and losses in products; a percentage with a tiny denominator can exaggerate an unimportant change.
Keep each evidence source in its own lane
Search Console describes search exposure and clicks; analytics describes observed visits and events under its collection rules. Neither dataset is a perfect substitute for the other. Do not force their totals to match by silently excluding records. Document consent, attribution, export and aggregation limits.
Preserve raw exports and add a mapping from canonical landing URLs to page groups. Keep unknown URLs in a review group. Record currency before aggregating revenue and distinguish orders from item quantities. A change in product mix or availability can alter revenue even if search visits are stable.
| Input | Keep with the record |
|---|---|
| Search export | Property, search type, market, device, dates and export limits |
| Analytics | Metric definition, consent/collection limits and attribution scope |
| Catalog | In-stock range and price changes during both periods |
| Release log | Change, affected URLs, publication time and owner |
Calculate changes without hiding zero or missing data
Calculate absolute difference as after minus before. Calculate percentage change only when the baseline is nonzero; otherwise report “new activity” with counts. Calculate aggregate CTR from total clicks divided by total impressions rather than averaging row percentages. Keep missing values distinct from genuine zeros.
Use separate views for products, categories and editorial pages. Compare query themes where the export supports them, but remember that grouped query totals may not reconcile with page totals. A lower CTR can accompany broader exposure; investigate the changed query mix before rewriting all titles.
| Metric | Before | After | Reading |
|---|---|---|---|
| Clicks | 100 | 130 | +30; +30% |
| Impressions | 2,000 | 4,000 | +2,000; +100% |
| CTR | 5% | 3.25% | Lower rate with more clicks; inspect query mix |
| Orders observed | 8 | 8 | No observed increase; not proof that all incremental visits had no value |
Worked decision: clicks rise but sales do not
The fictional category group gains 30 clicks while observed orders stay at eight. During the same period, its most popular size is unavailable and a new information-seeking query contributes impressions. The team should not declare the content a commercial success or failure from those totals alone.
The next task is to segment the changed queries, inspect the relevant landing pages and compare product availability. The report names the limitation: there is no controlled experiment and consented analytics does not observe every visit. The recommendation is a bounded investigation with a named owner, not a forecast of revenue from another paragraph of copy.
Separate incidents, seasonality and attribution
If a page disappears after a release, verify response and indexing controls before treating it as a ranking fluctuation. If all seasonal categories decline together, compare equivalent seasonal periods and catalog timing. Do not combine those explanations into one “SEO dropped” label.
Use annotations for promotions, tracking changes and outages. Check whether a metric definition changed mid-period. Where an experiment has no reliable control, describe the result as an association and retain competing explanations. This makes the report useful for the next decision instead of overstating certainty.
Turn the result into an action record
Practice: a page has zero recorded clicks before and 12 after; the export omits several low-volume queries. Report the change and one justified next step. Worked answer: “12 new recorded clicks; percentage change undefined from a zero baseline. Query coverage is incomplete.” Inspect the visible query themes and the landing experience before expanding the page pattern.
Pass when a colleague can reproduce the calculations, understand the limitations and identify the next owner and check date. Use the workbook’s blank comparison beside its completed example. Revisit the record at the agreed interval; do not keep changing the baseline until the graph looks positive.
Sources and maintenance
Check current platform documentation before implementation. Review these instructions when your platform, catalog behavior or source guidance changes.