FICTIONAL EXAMPLE · NOT CUSTOMER RESULTS
Start with a comparison, not a conclusion.
Imagine example.com received 1,248 clicks in a complete 28-day period and 1,110 in the preceding 28 days. That is 138 more clicks, an increase of 12.43%. The change describes search performance; it does not establish what caused it.
Ask for property totals first. Query and page tables can omit rows, so their sums should not replace the property summary.
Find the explanation behind a rising number.
The launch-checklist query in our homepage example rises from 60 to 120 clicks. That alone is not a reason to publish another checklist. Look at the landing page and compare the other metrics:
- More impressions? The site may be appearing more often for that search.
- Higher CTR? More appearances may be turning into visits. Ranking and the result’s presentation can both matter.
- A different position? Check the query and page context; average position is not one fixed rank.
Request each period’s query-and-page rows and match the same keys. A term present only in the current top-row report is not necessarily a new keyword. Country and device breakdowns add context, but separate country and device reports cannot establish how Canadian mobile visitors performed.
Turn findings into a realistic content plan.
| Evidence to investigate | Possible next step |
|---|---|
| A relevant query gains impressions but earns few clicks. | Check intent, ranking and the search result before revising the title or description. |
| A question query appears for an existing buying guide. | Read the guide. Add a useful answer if it is missing, rather than creating a competing URL automatically. |
| A matched page loses clicks. | Separate changes in impressions, CTR and query mix before recommending a rewrite. |
| A promising question has only one or two impressions. | Keep it as an idea to validate; do not present it as proven demand. |
Maple Orbit supplies Search Console performance data. Page content, inventory, conversion rates, competitor research and market-wide keyword volumes need separate evidence. The assistant should say when that evidence is missing.
Ask for a complete review.
Use my selected Search Console property. State the property and exact dates. Use the most recent complete 28-day period with final Web data and the immediately preceding 28 days.
1. Compare property-level clicks, impressions, CTR and average position. Show absolute changes, click and impression percentage changes, and CTR changes in percentage points.
2. Retrieve query-and-page rows separately for both periods. Compare only matching keys for increases and declines. State the row counts and retrieval limits; do not treat a missing row as zero.
3. Shortlist relevant question-shaped queries and searches with impressions but few clicks. Include the supporting metrics. Distinguish stronger signals from sparse evidence.
4. Show country and device breakdowns separately. Do not infer country-filtered queries or country-by-device results from separate tables.
5. Suggest five content reviews in priority order. For each, give the evidence, the existing URL if returned, what to inspect, and what additional information is needed. Do not claim to have read a page unless you actually fetched its content.
Never invent missing metrics or market-wide volumes. If a period has no final data, explain that and ask for an earlier range. Use tables and charts where the returned figures support them.
Select and copy the prompt into your connected Maple Orbit conversation. Your own account data will determine the answer; it will not reproduce these sample results.
What a good answer makes clear
The selected property, exact periods, Web search scope, returned row counts, missing evidence and calculation method should travel with the recommendation. Useful analysis shows why an action is worth considering and what still needs checking.