Thorough does not mean exhaustive
A thorough evaluation is one in which you can say what you checked, how you checked it and what you could not check. It is not one in which you looked at every page, which is rarely possible. The habits below help you cover the ground that matters, keep your findings reliable and avoid the quiet gaps that cause people to miss obvious problems.
Sample pages deliberately
When you cannot look at every page, the pages you pick determine what you will find. Left to instinct, most people choose the home page, the pages they like and the ones they already worry about. That produces a flattering and incomplete picture.
Instead, pick samples on purpose:
- Some of your most important pages.
- Some typical, middle-of-the-road pages.
- Some of your oldest pages.
- Some of your newest pages.
- A few chosen at random, such as every tenth item in a list.
- Pages of each type: articles, product or service pages, lists, forms.
Record how you chose them. If a pattern appears in the sample, widen it a little to check whether it holds.
Take notes as you go
Memory is unreliable, particularly after hours of checking. Write each observation when you see it, with three things: where it was (the page), what you saw and how sure you are. Include screenshots for visual problems. Notes should be specific enough that someone else could find the same issue without asking you.
It also helps to separate observations from interpretations. “The page takes several seconds before the text appears on my phone” is an observation. “The images are too large” is an interpretation that might be right or wrong. Keeping them apart lets you revisit your reasoning later.
Repeat important checks
A single look is a single sample. Pages behave differently at different times, on different connections and on different devices. For anything that matters, check it more than once:
- Test page speed on a few pages at different times of day, and note the range.
- View key pages on more than one phone and more than one browser, since device testing often reveals layout issues that one device hides.
- Look at search results more than once, since they can differ from one search to the next.
If two checks disagree, that is information. It may mean the problem is intermittent, or that your measurement is noisy.
Look at what visitors see, not what you see
You are logged in, you know where everything is, and your browser has saved information about your site. All of these make your experience unlike a newcomer’s. To reduce the gap:
- View the site in a private browsing window so you are not signed in.
- Try tasks a first-time visitor would try, such as finding contact details or a price.
- Ask someone unfamiliar with the site to attempt a task while you watch silently.
Watching a real person stumble over something you thought was obvious is one of the most useful forms of evidence you can get.
Use more than one kind of evidence
Any one source has blind spots. Reading pages shows quality but not scale. Tools show scale but miss judgment. Performance reports show what people do but not why. A thorough evaluation combines at least two or three types: your own reading, some counts or lists from a crawl of your site, and whatever visit information you have. When they agree, you can be more confident. When they conflict, investigate before concluding anything.
Check the corners
Blind spots tend to be in the places no one visits regularly. Make a short habit of checking:
- Old sections of the site that are no longer in the menu.
- Orphan pages, which are reachable only through search results or old links.
- Legal, contact and policy pages, where broken links often go unnoticed.
- Any subdomains or separate areas, such as a blog or shop hosted elsewhere.
- Staging or test pages that may have been left public by accident.
- The site as it appears with forms submitted and pages in different states, such as an empty cart or a no-results site search.
Use a fixed checklist, then go beyond it
A written list keeps you from skipping the dull items. A short evaluation checklist is a good base. But a checklist only finds what it asks about. After finishing it, spend some time browsing without a list, following whatever looks odd. Many real discoveries come from this kind of exploration.
Guard against your own assumptions
A few mental habits distort evaluations:
- Confirmation. Looking only for evidence that supports your first guess about what is wrong.
- Anchoring. Fixing on the first number or issue you see.
- Tool trust. Assuming a report is complete because it is long.
- Recency. Over-weighting what you looked at last.
Counter these by writing down at least one alternative explanation for each major finding, and by asking what evidence would show you were wrong.
Finish by reviewing your own work
Before you call the evaluation done, reread your notes and ask: did I cover each area I planned to? Which areas got only a light look, and is that stated? Is there any finding I cannot back up with a note or screenshot? Are there findings that need a second check? Then write down the gaps and the date, so the next round can begin where this one left off. Common audit mistakes are often not dramatic errors, but small omissions that went unnoticed because no one wrote down what was and was not covered.