Almost every other problem with an online shop announces itself. Baskets get abandoned and you see it. Deliveries go wrong and somebody rings up. Payments fail and the numbers tell you.
Not this one. A shopper who cannot find what they came for does not send an email asking where it is. They go back to Google, or to a competitor, and you never find out they were there at all. It is the quietest way a shop loses money, and it is usually happening on a scale that surprises people the first time they look.
Worth saying at the outset: this is about finding things once somebody is on your site. Being found in the first place is a different job, and the search side of it is a subject of its own.
Your menu is organised for you, not for them
Most shop menus are a picture of how the business thinks about its stock. That is completely natural, because whoever built it knew the catalogue inside out, and the categories that felt obvious were the ones already in use in the stockroom or the accounting system.
The trouble is that customers do not think in stock categories. They think about the job they are trying to do. A hardware shop might file something under Fixings and Fasteners, while the customer is looking for how to hang a heavy mirror on a plasterboard wall. Both are reasonable. Only one of them is what somebody typed.
There is a simple test that costs nothing. Ask five customers, or five people who are not you, to describe what they last bought from you without using any of your category names. What they say is the vocabulary your menu should be using.
Two other things go wrong here regularly. Top-level menus grow until nobody can take them in, when six or seven clear choices work far better than fifteen. And products end up buried three or four levels down, where the number of people still following drops sharply at every step. If something sells well, it wants to be reachable in one or two clicks, not four.
The search box that only finds what you already know
Somebody using your search box has usually decided they want to buy. They know roughly what they want and they are trying to skip the browsing. That makes search failures expensive in a way that browsing failures are not.
The default search on many platforms matches words in product titles and little else, which fails in entirely predictable ways.
Synonyms. Sofa, settee and couch are the same object to a customer and three different searches to a shop. So are trainers and running shoes, and any pair where the trade word differs from the ordinary one. Most platforms let you define synonym lists, and almost nobody fills them in.
Typos and plurals. If searching for “chizel” or for “chisels” when the product is “chisel” returns nothing, the search is being needlessly strict.
Codes and names. Trade customers search by part number and everybody else searches by description. Both need to work.
Descriptions, not just titles. If your search only looks at product names, everything you carefully wrote in the description is invisible to the person most likely to buy.
Test yours honestly. Take twenty things you sell, search for each the way a customer would say it out loud rather than the way it is catalogued, and count the failures. It is rarely zero.
Filters are the questions a customer would ask in a shop
Once you pass a few dozen products, filters do more work than the menu does.
The right filters are the questions somebody would ask a member of staff. What size, what colour, will it fit mine, is it suitable for outdoors, how much. That list is rarely the same as the attribute fields sitting in your product data, which tend to reflect what the supplier sent you rather than what a customer wants to narrow by.
A few things that quietly spoil good filters. Options that return nothing should not be offered, because a filter that leads to an empty page teaches people to stop using filters. A filter with only one option is just clutter. Showing how many products sit behind each option helps enormously, since it tells somebody whether narrowing is worth the tap. And the most useful filters should come first rather than in alphabetical or database order.
On a phone, filters usually live behind a button, and whether anybody presses that button depends entirely on how obvious it is. This is worth checking on your own phone rather than assuming.
What happens when there is nothing to show
A no results page is the most abrupt thing a shop can say to somebody who was ready to spend money, and on most sites it says nothing else at all.
It should always offer a way onward. What you searched for, spelled out, so they can see whether the fault was theirs. Suggestions or close matches. A few popular products. A line inviting them to ask, because if they were looking for something you can order in, you would rather know.
That last part matters more than it sounds. Searches that return nothing are a free list of things your customers want and you do not sell, compiled by the people who wanted them.
Your site search report is the most under-read data in most shops. Your analytics will tell you what people typed into your search box, how often, and which of those searches returned nothing. It takes ten minutes to read and it is the closest thing you will get to your customers telling you what is missing.
Find out where people are getting stuck
Two methods, and you want both.
The first is the search report above. Read last month’s terms, sort them by how often they came up, and mark every one that failed or returned something unhelpful. That is your fix list, in priority order, written by your customers.
The second is watching a real person. Ask somebody who does not know your site to find a specific product while you sit quietly and do not help. It is uncomfortable and it is the most useful ten minutes you will spend on your shop. You will watch them ignore the menu item you were proudest of, and you will see exactly where the hesitation happens.
Where to start
Start with search, because it is the cheapest to fix and the failures are the most expensive. Add the synonyms. Make sure descriptions are being searched. Give the no results page something useful to say.
Then look at the menu, using customer language rather than catalogue language, and check that your best sellers are not buried.
Filters come third, not because they matter least, but because getting them right depends on having tidy product data, and that is a longer job than the other two.
Getting a catalogue structured so it is genuinely easy to browse and search is part of how we set up and look after an online shop, and it is often the first thing we look at when a shop is busy but not converting. If you would like a second opinion on where people are getting lost, get in touch.
Once somebody has found the thing, the page has to do the rest of the work, which is where photographs and descriptions come in.