A do-it-yourself UX audit

The hardest part of auditing your own product is forgetting how it works. You know where every button is, which screen to open, which field you can skip. A newcomer knows none of that, and you need to see through their eyes.

This won’t replace proper research, but it finds the obvious problems in a couple of days for almost no money.

Pick three to five flows

You don’t need to check everything. Pick the flows the product lives or dies by:

landing page to signup;
signup to first value;
the main repeated action;
payment or upgrading to a paid plan;
coming back after a week and finding what you did before.

Write each goal as one sentence from the user’s point of view: “send a client an invoice”, not “complete the invoice creation flow”.

Walk through it yourself, as a newcomer

Create a new account. Open the product in a private window, on a phone, on a slow connection. Don’t use direct links or your knowledge of where things live. Go only by what’s on the screen.

At each step, note:

where you had to think about what to do next;
where the copy didn’t say what would happen after a click;
where you had to enter something the product could have known;
where something happened with no visible feedback;
where you hit an error, an empty screen or an endless spinner.

Write it down immediately and briefly, with the screen name. By evening the impressions blur and half of them will seem not worth mentioning.

The five-person test

Your own eyes hit a ceiling quickly. The next step is watching other people use the product. For finding the big problems, five people is usually enough: by the third or fourth, the same sticking points keep coming up.

How to run it:

find people who resemble your audience but haven’t seen the product. The founder’s colleagues and friends don’t count; they’re too polite;
give a goal, not instructions: “you need to invoice a client”, not “click Create in the left menu”;
ask them to think out loud;
stay quiet. Don’t hint, explain or defend. If they ask “what do I do here?”, ask back: “what do you think?”;
record the screen and audio if they agree.

Half an hour per person is plenty. A video call with screen sharing works as well as meeting in person.

What to note

Behavior is worth more than opinions. “I like the color” tells you very little. Ten seconds of silence over a form tells you a lot.

Note:

where they stopped or went back;
what they looked for, and where;
what words they use for things, often not the ones in your interface;
where they made a mistake, and whether they realized it;
whether they reached the goal at all.

That last one matters most. If two out of five couldn’t send an invoice, everything else can wait.

Sorting it out

Put everything into one list: the problem, where it happens, how many people hit it, and how badly it gets in the way. Fix what stops people from reaching the goal first, then what slows them down, and only then what’s merely ugly.

Don’t try to fix everything at once. Fix the big things and run five more sessions. Some problems will vanish, some will turn out to be symptoms of others, and new ones will appear that the old ones were hiding.

How often

Once a quarter, or before any major change, is enough. A habit of regularly watching a stranger use your product is worth more than any one-off audit, and a lot more than team debates about what would be clearer.

SUBXCAT runs audits and user tests like these and turns the findings into a prioritized list of fixes rather than a forty-page report.