Colin Lee MacLeod — More Better Good
MENU ☰

Test Your Website Search With Questions Readers Actually Bring

A search box is a starting point. The useful test is whether someone can move from a half-remembered phrase to the right page.

A search test moves from the reader’s task to their query and the resulting page.
Original illustrative workflow; not a screenshot or tested site interface.

As a website grows, its owner remembers categories, working titles and where everything was filed. A new reader has none of that context. They may remember a character, an ingredient or one word from an article. Test the search journey with those people in mind before adding another menu full of links.

Start with five reader tasks

Write five things someone might want to find. For a mixed magazine, they might be a particular story, a beginner lesson, a cookbook, information about a service and a practical guide. Choose material that actually exists. Keep the expected destination beside each task so the test has a clear target.

Now write the search words someone might use without knowing your title. “English practice” is a different test from pasting the exact title of a lesson. Ask someone unfamiliar with the site what they would type. Record their words rather than correcting them before the test starts.

Check what your actual search does

WordPress’s Search block provides a search interface for a post or page. Its presence alone does not tell you how a particular site will rank results or whether additional plugins change its behaviour. Check the installed implementation. WordPress: Search block.

For every test query, record the first useful result, whether the intended page appears, and what the reader does next. If the results are confusing, inspect their titles and excerpts. A repeated “Welcome” or an unexplained working title gives the reader little reason to choose the correct item.

Use a small test sheet

  • Reader task: what the person is trying to find.
  • Words entered: the exact query.
  • Expected page: a verified destination.
  • Observed result: what actually appeared.
  • Next edit: a specific content or interface change.

A fictional example: a reader enters “beginner conversation,” but the useful lesson appears under an internal course code. The first edit might be a clearer title and opening description. It is not automatically a reason to buy a new search plugin.

Give an empty result a way forward

Deliberately search for something the site does not contain. Does the page explain that no matching results were found? Can the person change the query without returning to the homepage? Are there a few relevant browse routes, rather than a wall of unrelated promotions?

Proposed copy could read: “No matching pages found. Try a shorter phrase, or browse Lessons, Stories and Cookbooks.” Use only section names that exist and link each one correctly. Do not tell people you have found a close match unless the system actually has.

Make the choice understandable

W3C’s supplemental cognitive-accessibility guidance describes search as another way to find content when menus do not work well for a person. It also discusses grouping related results under meaningful headings. That is a useful design direction, not a claim that adding search makes an entire site accessible. W3C: Provide Search.

Use result labels that help distinguish a lesson from a personal story or a service page. Keep essential information as readable text. On a small screen, check that the field, button, query and result titles remain visible and usable. Test the actual journey with a keyboard too; a decorative magnifying-glass icon should not be the only clue about what a control does.

Retest after the smallest useful change

Improve one confusing title, excerpt or empty-state route, then repeat the same query. Keep before-and-after notes. If a technical change is needed, describe the observed failure to the maintainer rather than requesting “better search” without an example.

Retain the five-task sheet when you add a new content section. It becomes a modest regression check: can people still find the useful pages after the latest redesign? The goal is not more results. It is a clearer path to the right result.

The Four Finger Rule. Real life. Real conversations. Explore the Four Finger Rule.