Your site search knows your roadmap
People type exactly what they want into that little box, in their own words, every day. Almost nobody reads the log.
Companies pay real money to find out what customers want. Surveys, panels, session recordings, the occasional focus group behind glass. Meanwhile there’s a small box on their own site where people type exactly what they’re looking for, in their own words, unprompted, every day, for free.
Almost nobody reads it.
We get why. A search log doesn’t look like research. It has no deck, no personas, no journey map with tasteful arrows. It’s a plain list of phrases in a dusty corner of the analytics, and because it costs nothing, it gets valued at exactly that. Budgets have a way of trusting whatever came with an invoice.
Two messages, one box
Every query in that log is one of two messages. The first is “I can’t find it”: the thing exists, your visitor couldn’t locate it, and your navigation just failed an exam it didn’t know it was taking. Message two is “you don’t have it.” Those are feature requests, filed voluntarily, ranked by frequency, from people motivated enough to go looking. Whole research programs exist to approximate what the zero-results report already says.
Both messages arrive carrying something rarer than data: honesty. A survey answer is a small performance, because people answer surveys the way they answer a waiter asking how everything was, but nobody performs for a search box, nobody is polite to it or tries to guess what you’re hoping to hear, they just type the thing they want and hit enter.
It’s the most candid sentence a user will ever write you.
The log is also a vocabulary lesson. Your nav says one thing; your users type another. When the menu reads “Solutions” and the search box keeps receiving “pricing,” that’s not a search problem. It’s a translation problem, and the users have already handed you the dictionary. We’ve watched teams argue for an hour about what to call a section while the answer sat in the search log, spelled the way customers spell it, in quantity.
An afternoon a month
Reading it takes an afternoon a month.
- Pull the top fifty queries and check that each one leads somewhere sensible.
- Read the zero-results list, which is the closest thing to a customer-written roadmap you’ll ever own: some entries are typos to forgive, some are products to consider, and a few are the thing your next quarter should have been about.
- Look at searches that end in an exit, because that’s someone asking a question, receiving your answer, and leaving anyway.
- Notice people searching for things that sit in plain sight on your homepage, since that means the sight isn’t plain.
The usual objection is volume. “Hardly anyone uses our search.” Fair, but even thirty queries a month is thirty verbatim sentences of customer language, which is more than most teams collect in a quarter of meetings about what customers probably think. Small samples of the truth beat large samples of guessing.
The other objection is that search is a symptom of failure, that a perfect site would need no search at all. Maybe so. But your site isn’t perfect, and until it is, the symptom log is the diagnosis, free of charge.
Somewhere in your analytics is a list of features ranked by demand and a glossary of how your customers actually talk, compiled nightly, at no cost. Research this good usually arrives with an invoice attached. This one just sits there, waiting for somebody to log in.
