Two tools, two questions
Search performance is usually measured with two different kinds of tool, and mixing them up causes most reporting mistakes.
- A search engine's own console, such as Google Search Console or Bing Webmaster Tools, answers: how did my pages appear in search, and how often were they clicked? It sees the results page, not your site.
- A web analytics tool answers: what did visitors do after they arrived? It sees your site, not the results page.
The two will never show identical numbers. They count different events, at different moments, with different rules about privacy and blocked scripts. That is expected, not a fault.
The four Search Console numbers
Google's Performance report for search results centres on four measures. In plain terms:
| Measure | What it counts | Read it as |
|---|---|---|
| Impressions | Times a link to your site was shown in results | Visibility, not attention |
| Clicks | Times a searcher clicked through to your site | Visits that search sent you |
| CTR | Clicks divided by impressions | How often being shown led to a visit |
| Average position | The average topmost position of your result | A rough guide only (see below) |
The help documentation also notes that very rare queries are hidden for privacy reasons and that data is aggregated, so the totals for a chart and the sum of a table may differ.
Why average position misleads
Average position is the most quoted figure and the easiest to misread. It is averaged across every query and every time your page was shown. A page that appears first for a rare query and fortieth for a common one can report a middling average that describes neither situation.
It can also move in the "wrong" direction for good reasons. If a page starts appearing for many new, broader queries at lower positions, its average position gets worse while its clicks go up. Look at clicks and impressions for specific queries and pages before drawing conclusions from the average.
Worked example
A tutorial page's average position drops from 8 to 14 over a month, and the owner worries. Filtering by query shows the page kept its place for its main topic and began appearing, lower down, for twenty related questions. Impressions doubled and clicks rose. Nothing was wrong.
Comparing periods fairly
- Compare like with like. Many topics are seasonal. Compare a month with the same month last year, not only with the month before.
- Use whole weeks. Weekday and weekend searching differ. A period that starts on a Monday and one that starts on a Saturday are not equal.
- Note what changed. Keep a dated log of site changes, new pages and campaigns. Without one, it is impossible to tell your own effects from outside ones.
- Be wary of single days. Search data is noisy. A trend over several weeks is more trustworthy than one spike or dip.
Tagging the other channels
Analytics tools can usually recognise visits from search engines on their own. Visits from emails, some social apps and printed links are harder to identify, and often appear as "direct" traffic. The common fix is to add campaign parameters, known as UTM parameters, to the links you share. Google Analytics Help describes them and offers a URL builder.
A tagged link adds a few labelled values to the address, for example utm_source=newsletter, utm_medium=email and utm_campaign=autumn-guide. Agree a short naming list and use it consistently; inconsistent spellings split one campaign into several rows. Do not add these parameters to links between your own pages, because that overwrites the original source of the visit.
A monthly routine
- Check the Performance report for the last 28 days against the same period last year.
- List the five pages with the biggest changes in clicks, up or down, and look for a cause in your change log.
- Check the indexing reports for new errors or pages unexpectedly excluded.
- Compare tagged campaign traffic in analytics with what each campaign was meant to achieve.
- Write down one change to make next month, and record it in the log.
Next: Shortcuts that backfire: a plain reading of spam policies