✦ Task type · Website testing
Findability tests, and the words you must not use
If you name the thing the way your menu names it, you have not run a test. You have read your own navigation aloud to somebody.
A findability test measures the distance between your information architecture and your customer's mental model. It is the cheapest test on this list and the easiest to invalidate by phrasing.
What to put in the public brief
- That it is a short set of hunting tasks on a website
- Desktop or mobile
- That a screen recording is required
- Roughly how long
- Whether any prior familiarity disqualifies them
Keep the URL and the task wording private, because the wording is the instrument.
What to put in the private instructions
Write each task as a situation, never as a label
"You want to know if it ships to Ireland before you buy" rather than "find the shipping information page".
Say whether site search is allowed
Both are valid tests and they answer different questions. Not saying is how you end up with a mixed result you cannot interpret.
Ask them to say where they expected it to be
Before they start hunting. The guess is more useful than the search, because it tells you the label they would have looked for.
Cap it at four tasks
And put the one you care most about second. First is warm-up, last is fatigue.
Proof worth requiring
Ask for
The recording, whether each task succeeded, the route taken, and their expected location before starting.
Do not ask for
A satisfaction rating. Someone who eventually found it will rate it fine, and the three wrong turns on the way are the entire finding.
When this task type is the wrong tool
If people cannot tell what your site is for at all, findability is the wrong layer — a first-impression test comes first.
If the content genuinely does not exist, no navigation test will say so kindly. Check that the answer is on the site before paying six people to look for it.
Common questions
What is the most common mistake?
Wording the task in your own vocabulary. Ask someone to "find the returns policy" and they will look for the word returns. Ask them "you bought the wrong size, what would you do" and you find out whether your information architecture matches how a customer thinks about the problem, which is the actual question.
How many testers?
Six to ten per task, and no more than four tasks per tester. Findability degrades sharply with fatigue, and a tester on their sixth hunt has learned your site, which makes them useless for the seventh.
What reward is fair?
Eight to sixteen dollars for a short set of tasks with a recording. Short work, but concentrated, and it is worth paying enough that people do not rush the hardest one.
Should I count success or time?
Both, and neither alone. Somebody who found it in four minutes via search and two wrong menus succeeded on paper. The route is what tells you the menu is wrong.
Thinking of taking this task?
The same job, from the other side. Rewards are set by the poster, so treat these as the range this task type usually sits in rather than a promise.
- Usually pays
- $8 to $16
- You need
- Any device with a browser
- Proof expected
- A recording, and where you expected each thing to be
Say where you would look before you start hunting. That guess is worth more than whether you eventually found it.
Create a worker accountTest your navigation
Set the reward and how many testers you need. Workers reserve a spot; the URL and the task list stay private until they do.
Post this task