A small team making search understand humans.
searchHub is built in Germany by search engineers who spent 20 years fixing other people's site search by hand — and then coded the fixes into software.
First we fixed search by hand. Then we built the software.
For more than twenty years we optimized other people's product discovery technologies — synonyms, redirects, rules, one shop at a time. Eventually we wrote down everything we kept doing by hand and turned the list into software.
That software became searchHub: the optimization layer that makes any product discovery technology better understand humans, whether it is in-house or off-the-shelf. Optimized, not replaced.
Five convictions that decide what we build, and what we leave out.
We held these convictions long before searchHub existed, back when we were fixing discovery journeys by hand, one shop at a time. If you want to understand why the product works the way it does, this is the place to start.
Optimized, not replaced.
The product discovery technology a shop already runs is rarely what holds it back. What holds it back is the distance between how people look for things and how the shop presents them, and that distance follows you onto any new platform. searchHub closes it on top of whatever you already run, in-house or off-the-shelf, so the discovery journey improves without a migration.
Prevent, don't explain.
A report that explains why last month's journeys went nowhere is useful, but it arrives too late for the shoppers who took them. We built searchHub as an agent that looks at the whole discovery journey, from the first keystroke in the suggest box to the results and the basket, finds where conversion potential lies untapped, and improves those journeys before the next shopper sets out on them.
Neural only where it provides value.
We reach for a neural model when it does something a plain rule cannot, and we stay with the rule when the rule already does the job. Both are measured the same way, against the same baseline. That is also why, wherever this site says AI, it tells you what that AI does.
Automatic, with a veto.
Nobody should spend their days maintaining synonym lists and redirects by hand, so searchHub runs continuously and keeps its own shortlist of improvements current. What it never does is act beyond your veto. Every decision stays yours to override, and the people who know the shop keep the last word.
Proof before promise.
An improvement is an improvement once the numbers say so, not once someone says so. Whether searchHub found a change, we made it by hand, or your team did, it is measured against the baseline. First A/B results within weeks.
Six values you can hold us to.
This is how we work, with our customers and with each other.
When you write to us, you reach the people who build searchHub, and every question lands with someone who can change the code rather than just log it.
We run the business case on your own data, not a sample, before we ask for a signature. If the numbers do not add up for your shop, we tell you so.
Every number on this site comes with its source and every claim with the mechanism behind it, because one A/B result tells you more than ten adjectives.
A journey that ends without a purchase is a puzzle to us before it is a ticket. Twenty years in, we still want to know what the shopper was looking for and where the shop lost them along the way.
searchHub has been in production since 2017, and the product has grown at the pace of proof: we add a capability once it has shown its value, not because a trend is asking for it.
Small enough that no one walks alone.
The same people who build searchHub answer when you write to hello@searchhub.io.
Sometimes all your search needs is to be found.
Drop us a line. We enjoy light-hearted conversations about search, too.