Wayfinding

Your site is built for someone who already knows what they want.

If your analytics show people arriving, searching once, and leaving, and the conclusion in the room was that the search needs work, it's worth knowing that the research points somewhere else entirely. The visitors you're losing usually aren't failing to find an answer. They're failing to work out what their question is, and almost nothing on a typical site is built to help with that.

This has been documented since 1989, mostly in library science, almost never in the rooms where commercial sites get designed.

The model your site was built on, and the one people actually use

The standard account of how someone finds information runs in a straight line. A person has an information need, turns that need into a query, the system returns objects matching it, and the person picks the relevant ones. Need, query, results, selection.

Every site is built for that person: a search bar, filters, specifications, a clear call to action, so they can get in and out fast. When someone genuinely works that way, it's an excellent way to serve them.

In 1989 Marcia Bates, at UCLA, published a paper in Online Review arguing that this isn't a realistic account of how people actually search. She called what happens instead berrypicking. In real searching the need itself changes as you go. You start with a vague question, find something half-useful, and that thing reshapes the question. You search again with a different query, pick up another fragment, follow a citation sideways, and abandon your original framing entirely. The search is satisfied not by one final result set but by bits gathered along a wandering path, the way you'd pick berries one at a time off scattered bushes.

Peter Pirolli and Stuart Card added the mechanism ten years later with information foraging theory: people follow information scent, the cues on a page suggesting a path leads somewhere useful, and they abandon a trail when the scent weakens. That reframes a bounce. It isn't a person who failed to find the answer. It's a person who stopped detecting a reason to keep going.

Put those together and the diagnosis inverts. A visitor who leaves after one search may not need a better search engine. They needed a page that taught them enough to ask a second, better question.

Two mindsets, one visit

Both models are correct. They describe different people, and often the same person on different days.

Task-oriented, running the classic model. This person knows what they want and knows your vocabulary. They want out of your way quickly. Serve them with search, calls to action, utility navigation, saved profiles, order history, specifications. Anything that shortens the path.

Exploratory, running berrypicking. This person is working out what the question is. They don't know your nomenclature and half the time they don't know your product exists. Serve them with main navigation that teaches the shape of the site, landing pages with orientation content, crosslinking, related recommendations, clear framing. Anything that produces the next berry.

Your job isn't to choose between them. It's to know which affordance serves which mindset and stop putting them in each other's way.

The main navigation is the piece most often got wrong. It gets treated as a menu, a list of destinations. It isn't. It's the primary instrument by which someone builds a mental model of what you have and how it's organized. It's a teaching surface. Treat it as a menu and the exploratory visitor never learns enough to form a real question, so they never become the task-oriented visitor the rest of the site was built for.

What this looked like at Ethicon

Ethicon makes surgical products. Sutures, staplers, energy devices, the things that end up in a surgeon's hands during an operation. When I worked on their customer engagement strategy, the company was shifting away from selling products and toward selling procedures.

That sounds like a marketing distinction. It's a claim about how the customer thinks. A product-focused company organizes everything by what it makes: catalog, sales team, navigation, org chart. A procedure-focused company organizes by what the customer is trying to do. And a surgeon isn't trying to buy a stapler. A surgeon is planning a bariatric case, and somewhere inside that plan sits a decision about a stapler, arriving late, in context, alongside twenty decisions that aren't about products at all.

The site was organized by product, so it was answering a question nobody had asked yet.

The design work followed from the model rather than from opinion. We restructured the main navigation so specialty and procedure content sat at the top level, where a surgeon's mental model actually starts. We rebuilt the faceted and hierarchical taxonomies around intent and the implicit relationships between things, rather than around the internal product hierarchy. We expanded site search to return faceted results serving both mindsets from one query. We used the navigation deliberately to drive traffic into procedural content, because the procedure is the context in which a product decision eventually gets made.

Page-level hierarchy became the reciprocal of the navigation: broad orientation at the top for the person still forming the question, specific reference material further down for the person who already has one. One page, both mindsets, in the order the mindsets arrive.

The clinical audience learned what Ethicon knew about a procedure before being asked to consider what Ethicon sold for it, which is also the order a surgeon would prefer to be spoken to.

Why it travels

I've used this framework on a bank, a hospitality operator, and a non-profit since. It travels because the failure it corrects isn't a healthcare failure.

The failure is the same every time. An organization knows its own products, so it organizes information the way it organizes itself. That structure is legible to everyone inside the building and to exactly one kind of visitor outside it: the one who already knows the vocabulary. Everyone else gets a site that assumes they've done the work the site was supposed to help them do.

Then, because the exploratory visitors leave without converting, the organization concludes it needs better search, a better call to action, or a redesign, all of which serve the visitor who was already fine.

We navigate using mental models built from prior experience. The work here is to take something complex and make it navigable by borrowing a structure the person already has, rather than imposing the one you happen to have.

Before you touch your navigation

Ask which visitor your site is currently built for. If the honest answer is "someone who already knows our product names," you have a site that works beautifully for people who didn't need it.

Then go and find the exploratory path. Not the one in the analytics, because analytics only show you the paths the site permits. Find someone who doesn't know your vocabulary, sit beside them, and watch what they do with the first screen.

They'll show you your org chart. They're the only ones who can see it.

---