Watching Ten People Use Your Site

What analytics cannot tell you

Your numbers report that people leave the pricing page. They do not report why, and every explanation you can think of is plausible.

The price is too high. The page is confusing. They were comparing and came back later. They could not find what a particular option included. They were on a phone and the table did not fit.

All five produce the same figure, and you cannot distinguish between them from behind a dashboard.

Watching one person try to use the page distinguishes them in about ninety seconds.

Why ten is enough

The finding that makes this practical for a small business.

The serious problems on a site are the ones almost everybody hits, so the first few people reveal most of them, and each additional person finds progressively less that is new.

Five people will surface the majority of your significant problems. Ten gives you confidence that a repeated problem is real rather than one person's habit.

Beyond that you are mostly confirming what you already know. This is not a study, it is a search for faults, and faults that only one person in fifty hits are not where your attention belongs.

Who to ask

Not colleagues, and not anybody who has seen the site.

The ideal is somebody resembling a customer who has never visited. Failing that, almost any competent adult outside your business will find the structural problems, because those are not subtle.

Friends, family, a neighbour, somebody at a networking event, customers you are on good terms with. Offer a coffee or twenty dollars.

What you must avoid is a room of people who already know what your services are called, because knowing the vocabulary removes exactly the obstacle you are trying to observe.

Give them a task, not a tour

The single most important part of the method.

Asking somebody what they think of your website produces politeness and opinions about colour.

Asking them to accomplish something produces behaviour, and behaviour is the data.

Good tasks are realistic and specific: find out whether they service your area, find the price of the mid-range option, book an appointment for next Tuesday, find out what happens if the part arrives damaged.

Give one task at a time, in the words a customer would use rather than the words on your site, and then be quiet.

Say nothing

The hardest instruction and the one that decides whether the exercise works.

When somebody hesitates, the urge to help is close to overwhelming, and helping destroys the finding.

Sit slightly behind them. If they ask a question, say that you want to see what they would do on their own. If they get stuck for a long time, let them stay stuck, because the length of time is itself the measurement.

The only prompt worth using is asking what they are thinking, which produces the reasoning behind the hesitation without directing it.

What to write down

• Every hesitation, and how long it lasted.
• Anywhere they scrolled back up looking for something.
• The words they used for things, versus your words.
• What they clicked that was not clickable.
• Where they gave up, and what they said at that moment.
• Anything they did not see that was on the screen.

The third item alone frequently changes a site's navigation, and it is free.

A worked example

A heating company watched seven people attempt one task: find out whether the firm covered their village and roughly what a system replacement would cost.

Six of the seven went looking for a service area page. The site had that information in a sentence at the bottom of the about page.

Five used the word boiler. The site consistently said heating systems, which was more accurate and was not what anybody typed or looked for.

Four tapped the header photograph expecting it to be a link.

Three gave up on the price question entirely, since prices were given as from figures with no indication of what a typical job cost.

They added a service area page with a list of villages, changed the vocabulary throughout to match what people said, and published a typical price range with an explanation of what moved it.

The whole exercise took one afternoon and cost seven coffees.

Do it on a phone

Hand them a phone rather than sitting them at a computer, unless you know your customers are overwhelmingly on desktop.

Most problems are worse on a phone, and some exist only there.

Let them use their own device if they will, since their font size settings, their browser, and their habits are all part of the real conditions.

Watching somebody use their own phone also removes the awkwardness of an unfamiliar machine, which otherwise contaminates the first minute of every session.

The three-in-a-row rule

A simple filter for deciding what to act on.

If three people in a row hit the same obstacle, it is real and it is costing you money. Fix it before running anybody else.

If one person struggles with something nobody else notices, note it and move on. It may be a habit, a preference, or an unfamiliarity with the internet generally rather than with your site.

This keeps the exercise from producing an unmanageable list, which is the usual reason such findings are never acted on.

The counter-case

Watching people has limits and they are worth stating.

It tells you nothing about whether people can find you in the first place, since you brought them to the site. Discoverability is a separate problem entirely.

It also does not measure magnitude. Ten people cannot tell you whether a fault costs you two percent or twenty, only that it exists.

And observed behaviour under mild scrutiny is not identical to behaviour alone, though the difference is usually small compared with the size of the problems you will find.

The method finds faults. Deciding which faults are worth the most is a separate judgement, informed by how many people your numbers say reach that page.

How to run it

1. Write three realistic tasks in customer words.
2. Find ten people outside your business.
3. Give them a phone and one task at a time.
4. Say nothing and take notes.
5. Record the hesitations, not the opinions.
6. Fix anything three people hit in a row.
7. Run three more afterwards to confirm.

An afternoon, and it is the highest return available to almost any small business site.

The version you can run alone is covered in testing your own site in ten minutes.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.