vitrivis All writing
Websites

Why most healthcare websites fail the five-minute doctor test

7 min read

Hand a surgeon your phone in the corridor outside theatre. Open your product page. Give them five minutes.

Then ask four questions. What is it for. How does it work. Does it work. What happens next.

Most sites lose them on the first one.

We have run this test on device pages, clinic sites and hospital service lines. The failure is almost never the design. The design is usually expensive and usually fine. The failure is that the page was built in the order the company thinks, not the order the reader asks.

The five minutes are real

Nobody sits down to evaluate your website.

They are between cases. They are standing up. They are on a phone, on hospital wifi, three floors below ground, behind a firewall that blocks half the internet. A rep mentioned something. A colleague sent a link. They have five minutes before the next patient is on the table.

That is the whole window. It does not extend because your content is good.

A survey of 2,200 office-based physicians found that what mattered most to them was the credibility of the source, being able to get at it quickly at any hour, and how easy it was to search. Not depth. Not completeness. Access and trust, fast.

Design for the five minutes and the long evaluation takes care of itself. Design for the long evaluation and you get neither.

Failure one: the page runs in your org chart’s order

Home. About Us. Our Technology. Our Products. Contact.

That is the order your company was built in. It is the order the approval chain thinks in. It is not the order anyone asks in.

A surgeon asks what it is for. Only then how it works. Only then whether it works. A referring dentist asks something else entirely: what happens to my patient, and when do I get them back. A patient asks whether it will hurt, how long it takes, and what could go wrong.

Three readers. Three different first questions. One page.

Most device pages answer the company’s question first, which is who we are. Nobody arrived wanting to know who you are. They will want to know later, when they are deciding whether to trust the claim. Putting it first spends the only five minutes you get on the least urgent thing on the page.

The fix is not a redesign. It is a reorder. Take the four questions, put them in the order a reader actually asks them, and build the page as a sequence of answers. Everything that does not answer one of them moves down or comes off.

Failure two: it does not load

This is the unglamorous one and it kills more pages than bad copy.

Google’s research, across more than ten thousand mobile domains, found that 53% of mobile visits are abandoned when a page takes longer than three seconds. The same study found the average mobile site took 19 seconds to load on 3G and 14 on 4G. Bounce probability climbs by about a third as load time goes from one second to three.

Those numbers are from consumer browsing on ordinary networks. Now move the reader into a hospital.

Hospital wifi is not your office fibre. It is congested, filtered, and frequently worse than the cellular signal, which is itself worse than it should be because the building is concrete and the reader is underground. The page that renders in 1.2 seconds on the marketing team’s laptop takes fifteen in the corridor where it is actually opened.

Look at what you are shipping. A hero video that autoplays. A font stack pulling from three domains. A chat widget, a consent banner, an analytics tag, a heatmap tag, and a tag manager to load the other tags. Each was approved by someone who never opened the site anywhere but the office.

Test on the network your reader is on, not the one you are on. If nobody in the approval chain has opened the page on cellular data with the signal at one bar, the page has not been tested.

Failure three: the evidence is a wall of PDFs

Doctors weight credibility above almost everything. So companies put up evidence pages. Then they fill them with links to PDFs.

A link to a forty-page paper is not evidence. It is a deferral. It says: the proof exists, go and find it. In five minutes, on a phone, nobody is finding it.

What works is putting the chain on the page. The claim, then the number the claim rests on, then the study the number came from, then the link. One sentence, traced back to the table, figure and clause it sits on. A reader who can see the chain will believe the claim without opening the paper. A reader who cannot will not believe it even if they do.

This is also the part that gets you through medical and legal review faster, because the reviewer is looking for exactly the same chain. If your page already shows its working, the approval is a check rather than an argument.

Failure four: the mechanism is a mood film

They asked how it works. You gave them ninety seconds of slow push-ins, a string section, and a device rotating on a black background.

That film has a job. It is not this one.

Mechanism means sequence. What happens first, what happens next, what the device is doing at each step, and what changes in the tissue because of it. A surgeon watching it should be able to narrate it back afterwards. If they cannot, it was not a mechanism film. It was a brand film with anatomy in it.

The test is simple and slightly brutal. Play it to someone who does not already know the product, then ask them to describe the steps in order. Where they hesitate is where the film is decorating rather than explaining.

Failure five: the next step is a contact form

The reader is convinced. They scroll to the bottom. There is a form with six fields and a dropdown, and it goes to a shared inbox.

That is not what a doctor’s next step looks like.

Their real next steps are specific. Who do I call. Can I watch one being done. Is there a proctor near me. Does my hospital already have this. Who trains my scrub team, and how long does it take. What does it cost the department, roughly, before I take it to purchasing.

None of those is a form. Several of them are a phone number and a name. One of them is a calendar. If your page ends in a generic enquiry box, you have converted genuine intent into an admin task, and the admin task will lose to the next case.

How to run the test

You can do this in an afternoon. It costs nothing and it is uncomfortable, which is how you know it is working.

Pick three real readers. Not colleagues, not the medical advisor who already knows the product. One surgeon, one referrer, one patient or carer.

Use a phone, on cellular. Not the office network. If you can, do it in the building where they work.

Set a timer for five minutes. Then leave them alone with it.

Ask the four questions afterwards, out loud, without prompting. What is it for. How does it work. Does it work. What happens next. Write down their answers in their words, not yours.

Note where they scrolled past and where they stopped. The scroll is more honest than the answers.

Do not explain anything. The moment you explain, the test is over and you have learned nothing. If you feel the urge to jump in, that urge is the finding.

What passing looks like

A page passes when a surgeon can answer all four questions in their own words after five minutes, having never spoken to you.

Not when it wins an award. Not when the board likes it. Not when it looks like the competitor’s, but better.

Most of the work in getting there is subtraction. The order changes, the evidence chain comes onto the page, the mood film moves to where mood belongs, and about a third of the content goes, because it was answering a question nobody asked in the five minutes they had.

The page gets shorter. It also gets faster, which fixes the second failure for free.

Sources