Why a short list beats a long one
Most reporting tools can produce dozens of numbers. Very few of them change what you would do next. A useful set of evaluation metrics is small, stays the same from month to month, and each item answers a question you actually have: Can people find the site? Do they use it? Does it work properly? Does it achieve its purpose?
This article groups the measures worth tracking under those four questions. For each one it says what the number can tell you and where it misleads. None of them is a score of overall quality, and none guarantees a ranking.
Visibility: can people find the site?
Impressions count how often pages from your site appeared in search results. A rising trend usually means search engines are showing your pages for more queries, or showing them more often. It does not mean anyone saw them in a meaningful way; a result far down the page may be counted without being noticed.
Clicks count visits that came from those results. They are closer to real interest than impressions, but they depend on how your result looks and on what else appears on the results page, which changes over time.
Average position summarizes where your pages tended to appear. Treat it as a rough direction, not a precise rank. Results are often personalized or vary by location, and one average can hide a page ranking well for one phrase and poorly for another.
Indexed pages is a count of pages search engines have stored. If important pages are missing, people cannot find them, which is the topic of indexing gaps. A large number is not good in itself; a site full of thin or duplicate pages may show a high count and little value.
Engagement: do visitors use it?
Visits (sessions) show how much activity the site gets. Compare like with like: the same month last year is more informative than last week, because many topics are seasonal.
Engagement signals such as time on page, pages per visit and scroll depth hint at whether content holds attention. They are easy to misread. A short visit can be a success if the page answered the question immediately, and a long one can mean confusion. The same caution applies to bounce rate, which is useful only when you know what a good visit looks like for that page.
Returning visitors can show whether people find the site worth coming back to, which matters more for guides and publications than for a one-off service page.
Technical health: does it work properly?
Crawl errors and broken links tell you where the site fails visitors and search engines. Counts of broken links and error pages are among the easiest numbers to improve, and fixing them is a routine part of audits.
Page speed and core web vitals describe loading, responsiveness and visual stability. Treat the measurements as signs of where users may be struggling. Test results from a lab setting and results from real visitors can differ, so it helps to understand field data before reacting to a single test.
Mobile usability results, covering things like tap targets and readable text, matter because many visitors use phones. A clean report does not prove the experience is pleasant, so open the pages on a real device occasionally.
Security status, such as whether every page loads over a secure connection, is a simple pass or fail check that is worth confirming regularly.
Outcomes: does it achieve its purpose?
This is the group that most often gets neglected. A site exists to do something: sell a product, collect inquiries, build a subscriber list, inform readers. The matching metric is a conversion, meaning a visitor completing the action you care about. Tracking it needs a little setup, which conversion tracking covers for small sites.
Outcome numbers are the best check on all the others. If clicks rise but sign-ups do not, the extra visitors may not be the right audience. Choosing which outcomes to follow is the subject of performance indicators, and deciding what counts as a good result is part of measuring success.
What no metric can tell you
- Why something changed. A metric shows that a number moved, not the cause. Updates to search engines, seasonality, a competitor’s move and your own edits can all produce similar changes.
- Quality. Content that is accurate and helpful can still have unremarkable numbers, and the reverse can also happen.
- The future. A good month does not promise another one. Search results change, and outcomes are uncertain.
For these reasons it helps to read numbers over time rather than in isolation, a skill covered under metric interpretation, and to write down changes you make so you can connect cause and effect later.
Putting a starter set together
A sensible starter set for a small site might include:
- Clicks and impressions from search, reviewed monthly.
- Count of indexed pages compared with the number of pages you intend to have.
- Number of broken links and error pages.
- A speed and mobile check on your most important templates.
- One or two conversions tied to the site’s purpose.
Record the starting values before changing anything, because without a baseline you cannot tell whether a change helped. After that, add a metric only when you can name a decision it would inform. If a number would never change what you do, leave it out of the routine.
Common mistakes with metrics
- Tracking everything and acting on nothing.
- Comparing your numbers with other sites that have different goals, audiences and sizes.
- Reacting to a single day’s movement.
- Treating a tool’s overall “score” as a verdict instead of a prompt to investigate the details behind it.
- Optimizing the metric instead of the experience, for example shortening pages purely to lower an engagement number.
Keep the list short, keep the definitions consistent, and treat each number as a question to investigate, not a grade.