Most websites are a list of what the owner wants to say. Every visitor arrives with something they want to do. The whole craft is in the distance between those two lists.
October 2026
Ask an organisation what belongs on its website and you get everything: the history, the values, the press releases, the new logo, the department that felt overlooked last time. Ask a visitor and you get one thing. What does it cost. Are you open on Monday. How do I cancel. Where do I send this form. Does it fit my machine. Who do I talk to.
The long neck
Gerry McGovern gave that gap a method and a name. You collect every plausible thing someone might come to your site to do — a long, flat list, a hundred or more, in the visitor’s words, with no grouping and no hierarchy to nudge the answer. Then you ask a lot of real people to pick the few that matter to them.
The small tasks each have an owner in the room. The top tasks belong to nobody — which is why they end up three clicks down.
The result is the same shape almost everywhere. A handful of tasks take an enormous share of the votes; the rest trail off into almost nothing. Not a long tail — a long neck. Five or six things carry the site, and a hundred others share the crumbs.
What makes this hard is not the measuring. It is that the tiny tasks have owners and the top tasks do not. Every one of those hundred items is somebody’s project, somebody’s budget, somebody’s annual objective, and that somebody is in the room. The five things the public actually came for belong to no one in particular, which is precisely why they are buried three clicks down behind a menu written in the language of the organisation chart.
I worked with McGovern on this at the European Commission, while its central website was being rebuilt. If you can identify the top tasks of an institution with tens of thousands of pages and dozens of directorates, you can do it for a business with twelve. The method is the same, the politics are simply louder.
I have run the same method for a number of Belgian municipalities, and the answer was almost comic in its consistency. What people wanted from the town hall’s website, above everything else, was the opening hours of the recycling centre and of the municipal swimming pool. Not the policy plan. Not the council minutes. Not the mayor’s column. The bins and the pool.
Run it again today and new ones would push their way to the front — how to register a property for short-term rental, say, or what a permit costs and how long it takes. These are questions people need answered on a Tuesday evening, with a deadline in their head. They are rarely what the organisation itself is busy with that week, and that is the whole difficulty: the list does not care what is on your agenda.
And then it gets harder
Suppose you have your five tasks. You still have to survive the way people actually use a page, which Steve Krug described a quarter of a century ago in a book whose title is the whole argument: Don’t Make Me Think. Three of his observations have not aged a day.
People don’t read, they scan. Nobody arrives to appreciate your page. They sweep it for something that looks like what they want, and the rest of your careful prose is scenery passed at speed.
People don’t choose the best option, they choose the first plausible one. Weighing up six menu items costs effort; clicking the first one that might work costs nothing, and you can always press Back. So the label that reads clearly wins over the label that is technically correct.
People muddle through. Most visitors never understand how a site is organised, and they don’t want to. They form a rough idea that mostly works, and they keep it. If your structure only makes sense once it has been explained, it does not make sense.
Put those together with the long neck and you get an uncomfortable conclusion: a page has a second or two to show, without being read, that the thing the visitor came for is here and is one click away. Not to explain it. To show it.
What it looks like when you take it seriously
The home page stops being a table of contents. It becomes the shortest route to the five things, in the visitor’s words, above everything the organisation is proud of.
The menu shrinks. Everything that is not a top task still exists — it is findable, it is in the archive, it is one search away — but it stops competing for attention with the things people came for.
Labels are written by visitors, not by departments. “Prices” beats “Our offering”. “Cancel a booking” beats “Customer self-service”. The test is whether someone can say the label out loud and mean it.
Success is measured as tasks completed, not visits. Time on page is not a compliment; often it means somebody could not find the thing. The honest question is whether a stranger got what they came for, and how long it took.
The method is older than most of the web and it has not changed much, because the part that does the work is the counting. Here it is as a procedure, for anyone who has to do this on a Tuesday afternoon.
The method, start to finish
- Collect every candidate — every plausible thing somebody might come to the site to do, written in the visitor’s words, not the organisation’s. A hundred is a normal number. Take them from support mail, from the phone, from the search box, from whoever answers the questions.
- Cut each one to a single line and remove the duplicates that are only duplicates once you say them out loud. Keep the list flat: no groups, no headings, no order that hints at an answer.
- Put the whole list in front of real visitors and ask each person to pick the five that matter most to them. Not rank, not rate — pick five. Fifty people is enough to see the shape; a hundred makes it hard to argue with.
- Count the votes and let the neck appear. You do not decide the top tasks; the count does. The top five usually take a quarter of all the votes, and the bottom half take almost none.
- Check the result against the evidence you already have — site search logs, the questions that arrive by mail, the pages people actually open. If the count and the logs disagree, find out why before you act.
- Spend the page where the votes are. The top five get the home page, the menu and the first screen. Everything else stays reachable and stops competing.
- Run it again every year or two. Tasks move — a new service, a changed season, a law. The list is a measurement, and a measurement has a date on it.
The boring part is the point
Top tasks are almost always dull. Opening hours. Prices. An address. A form. The phone number of a person who answers. Meanwhile the page everyone in the organisation is proudest of — the story, the vision, the film — is usually somewhere around number sixty on the list.
That story is worth telling. It is simply not worth putting in the way. Serve the dull thing instantly, and you have earned the two minutes in which somebody might want to know who you are.
Read the originals
Both of these are short, plain and cheap, and either one will change how you look at your own site.
- Gerry McGovern on top tasks — gerrymcgovern.com
- Steve Krug, Don’t Make Me Think — sensible.com