Search measurement guide
Turn Search Console Performance Data Into a WordPress Work Queue
Export the same Search Console view each month, group pages by purpose, flag changes worth investigating, and write each queue item with evidence, an action, and a review date.
On this page
Once a month, export the same Search Console Performance view, group your landing pages by what each page is for, and flag the pages whose clicks or impressions changed or whose results are seen but rarely clicked. Turn each flag into a queue item that states the observation, the evidence and filters behind it, a proposed action, and a date to check the result. Then pick no more than three actions to do that month.
A report you look at is not a work queue. The queue is the short list of decisions you made from it, each one traceable to the data.
Export the same view every month
Changing the view between months makes the comparison meaningless. Fix these settings and record them at the top of each month’s sheet:
| Setting | Suggested value | Why record it |
|---|---|---|
| Property | The domain or URL-prefix property you use | Different properties can show different totals |
| Search type | Web, with the same text-based or multimodal choice each month | Image, video, and news are separate search types |
| Date range | Last 28 days, compared to the previous period | Keeps periods equal in length |
| Grouping | Pages, then Queries for flagged pages | Page grouping and query grouping count differently |
| Filters | Country or device, if the business is local or device-specific | Filters change totals |
The search type filter covers web (text-based or multimodal), image, video, and news, per Google’s Performance report overview, checked September 27, 2026. A Domain property includes all subdomains and protocols, while a URL-prefix property includes only URLs under that prefix, per Google’s help on adding a property, checked September 27, 2026.
Know the limits of what you export. The Performance report table shows at most 1,000 rows, and anonymized (rare) queries are omitted from the query table but included in chart totals unless a query filter is applied. Google’s help on troubleshooting Performance report data, checked September 27, 2026. Daily data is labeled in Pacific Time in all views except the 24-hour view, per the same page.
Aggregation also differs by grouping. Data grouped by queries, countries, devices, or dates is aggregated by property, while data grouped by pages or search appearance is aggregated by page, so chart and table totals can differ. Google’s Performance report overview, checked September 27, 2026. Compare page rows with page rows, and do not expect them to sum to the chart.
Segment pages by their actual purpose
A sitewide total mixes pages with different jobs. A seasonal article losing impressions in autumn and a service page losing clicks are different problems, and averaging them hides both.
Add a purpose column to the exported pages:
| Purpose | Examples | What “working” means |
|---|---|---|
| Service or product | Service pages, product categories | Clicks from queries describing the offer |
| Location | Area or branch pages | Clicks from searches in that area |
| Guide or article | How-to posts | Impressions and clicks for the questions it answers |
| Brand and navigation | Home, about, contact | Branded queries find the site |
| Utility | Cart, account, thank-you pages | Usually nothing; they rarely need search traffic |
Search Console’s branded and non-branded query filter can help separate the brand row, where it is available. It is not offered for sub-properties or sites with a low number of impressions. Google’s Performance report common tasks, checked September 27, 2026.
Flag changes that warrant investigation
With pages segmented, look for three patterns. Each is a reason to look closer, not a diagnosis.
- Significant change from the previous period. Google suggests sorting by the Difference column to find queries or pages with significant change. Google’s Performance report common tasks, checked September 27, 2026. Start with service, product, and location pages.
- Seen but rarely clicked. Google’s help says a low click-through rate can mean people see the page in results but don’t think it answers their query, and suggests reviewing the title, description, or content. Same source. Only flag pages with enough impressions to make the rate meaningful for your site.
- Mismatched queries. Open a page’s queries. If most impressions come from searches the page was not written for, the page may be answering the wrong question, or another page may be missing.
Ignore small movements on pages that are performing. Google’s traffic-drop guidance says small position fluctuations can happen at any time and recommends avoiding radical changes if a page is already performing well. Google’s guide to debugging Search traffic drops, checked September 27, 2026.
Write each item with evidence and an action
A queue item that says “improve the catering page” is not actionable. Use a fixed template:
| Field | Content |
|---|---|
| Page | URL and purpose |
| Observation | What the data shows, in numbers, with the period |
| Evidence | The view: property, search type, dates, filters |
| Working explanation | What you think is happening, labeled as untested |
| Action | The smallest change that tests the explanation |
| Owner and date | Who does it, and when |
| Review date | When you will compare equal periods again |
The rows below are a synthetic dataset for a fictional site, garden.example, with invented figures. They show how a few items might be chosen from a longer list.
| Page | Purpose | Observation (28 days vs previous, illustrative) | Working explanation | Action |
|---|---|---|---|---|
| /raised-beds/ | Product category | Impressions 4,100 → 4,300; clicks 210 → 120 | Title still advertises a spring sale that ended | Correct title and description; recheck in 28 days |
| /delivery/ | Service | Top queries mention “same day”; page does not say whether it is offered | Page does not answer the main question | Client confirms policy; add it to the page |
| /blog/winter-pruning/ | Guide | Impressions down 60% | Seasonal; same drop last year | No action; note it |
| /compost-guide/ | Guide | Impressions up, clicks flat, position around 35 for new queries | New queries where the page ranks low | No action this month; watch |
Only the first two become work this month. The other two are recorded so nobody reopens them next month without new evidence. For pages seen but rarely clicked, diagnosing low click-through rates covers the investigation in detail, and impressions rose but clicks fell covers the mixed signal in the compost row.
Carry the work into WordPress
Search Console does not edit your site, and WP Visibility does not import Search Console data. The link between them is you.
For metadata items, edit the Search title and Search description fields in the WP Visibility sidebar of the post editor, or from the command line with wp visibility set <id> --description="...". If an assistant drafts the changes through WP Visibility’s post SEO abilities, with the Assistant connection and Autopilot modules on and Hold for my review selected, the drafts wait in the review queue for approval; see approving or rejecting a proposal.
Record the date each change went live in the queue sheet. When the review date arrives, compare an equal period after the change with the period before, using the same filters. Google notes that it can be difficult to attribute an improvement to a specific change, because other events also affect performance, and suggests checking whether the timing maps closely to your change compared with similar periods in the past. Google’s Performance report common tasks, checked September 27, 2026.
Monthly checklist
- Same property, search type, date range, and filters as last month, recorded.
- Pages tagged by purpose.
- Flags from differences, low click-through on meaningful impressions, and mismatched queries.
- Each item has evidence, a labeled explanation, an action, and a review date.
- No more than three actions chosen.
- Change dates logged for the next comparison.
