Use cases

Use these use cases to review search performance, investigate traffic changes, choose SEO work, and prepare reports. Each use case explains what to ask, which tools the assistant uses, and what the data can show.

Choose a use case

Choose the use case that matches the work you need to do:

1. Read a site's search performance.
2. Prepare a search-performance report.
3. Investigate a traffic drop.
4. Find page-two keywords worth working on.
5. Find pages that need a title or snippet review.
6. Build a content-refresh shortlist.
7. Compare performance across site sections.
8. Separate branded and non-branded performance.
9. Find queries with growing impressions.
10. Choose which SEO opportunities to work on first.

Review performance and prepare reports

Read a site's search performance

Check clicks, impressions, click-through rate, and average position for each connected search engine. Compare the current period with the previous period, or read a trend to find when performance changed.

Start here if you want an overview before choosing a specific investigation.

Prepare a search-performance report

Prepare a report for one site or several client sites. Summarize results for each engine, explain important changes, and list the next actions.

Choose the dates and report sections before running the report. Use the same dates across sites if you need a combined total. Request each calendar month separately if you want to compare months with different numbers of days.

Investigate losses and find improvements

Investigate a traffic drop

Find the queries that lost clicks and check the available evidence for ranking changes, page changes, lower click-through rates, or lower impressions. Use the result to choose what to investigate before changing a page.

The data does not always identify a cause. Report an inconclusive result if the tool cannot separate the possible causes. Search-console data alone cannot prove that an AI Overview or another search-result feature caused the loss.

Find page-two keywords worth working on

Find non-branded queries with impressions and average positions between 11 and 20. Identify the pages that appear for those queries and check whether several pages compete for the same query.

Use the shortlist to choose existing pages to improve. Review business relevance before treating a query as a priority.

Find pages that need a title or snippet review

Find pages whose click-through rate falls below the site's expected rate for their position. Review the queries each page appears for before proposing changes to its title or snippet.

Low click-through rate is a reason to investigate. It does not prove that the title needs changing.

Build a content-refresh shortlist

Find pages that lose traffic across several complete periods or show a drop that lasts. Review each page's query data to decide what to investigate.

Separate content pages from login pages and other utility pages. A traffic loss does not always call for a content refresh.

Review sections, brands, and query growth

Compare performance across site sections

Compare clicks and impressions across URL directories, such as /blog/ and /tools/. Find the sections that changed and review individual pages within them.

Use supported text filters for other URL groups. Directory analysis does not support custom wildcard or regular-expression patterns.

Separate branded and non-branded performance

Check the brand names and aliases used to classify queries. Compare branded and non-branded clicks, impressions, and click-through rates across periods.

Review example queries to check the classification. The reported shares describe the returned query data, not necessarily all traffic recorded for the property.

Find queries with growing impressions

Find queries whose impressions rise across several complete periods. Group related queries and check which existing pages appear for them.

Use the results to choose pages to review or expand. Rising impressions can reflect better search visibility, so they do not prove that overall search demand grew.

Choose the next work

Choose which SEO opportunities to work on first

Build a shortlist of ranking opportunities, click-through-rate improvements, declining pages, and underperforming queries. Review the evidence and the page involved before choosing an action.

Start with ranking and click-through-rate opportunities if a full queue takes too long. Treat estimated clicks as projections, not promised results. Do not add overlapping estimates for the same opportunity.

Before you start

Check the site, data coverage, and dates before drawing conclusions:

1. Check the server status. Call get_server_status at the start of the session so the assistant can check whether its tool list is current.
2. Find the site. Call list_sites to see accessible Google properties and Bing sites. A listed property is not always attached to a Composway workspace.
3. Check the connection. Call get_connection_status to confirm which property or site is attached and when its data last synced. Connected account credentials do not mean both engines are attached to that site.
4. Set the period. Use explicit dates for comparisons and reports. Read the dates returned by each tool because tools that analyze complete weeks or months can exclude a partial period.
5. Read the limits. Check unsupportedInputs, coverage information, and source notes before interpreting an empty result or a missing value.

What the data can show

These use cases use search analytics, not AI-answer visibility or competitor monitoring. They also do not require paid topic discovery or live search-result checks.

Keep these limits in mind when reading a result:

• A missing query or page does not prove it received zero traffic. Google and Bing can limit the rows available to a tool, and Google omits some queries for privacy.
• Use site snapshots and site trends for property totals. Query and page breakdowns can cover fewer rows than those totals.
• Keep average position separate for Google and Bing. Composway does not calculate a combined average position because the engines' position measures are not confirmed comparable.
• Country and device breakdowns are not available through these tools. Image and video analytics use Google data only.
• Several pages appearing for one query do not prove that consolidation is the right action. Review the pages, search intent, and inventory limits first.
• Use full page URLs for page-query lookups.
• Report missing engines, failed sections, and incomplete coverage. Do not present missing data as zero or a partial report as complete.