Customers type what they want to buy, and nobody reads it

The internal search box is where customers say, in their own words, what they want. It costs nothing, needs no setup, and almost nobody looks at it. The four things it tells you, and what to do in order of cost.

There is a place in your store where customers type, in their own words, exactly what they want to buy. It costs nothing, needs no configuration, and in most stores nobody ever looks at it: the internal search box.

Every search is a person who has already decided to buy from you, telling you what they are after. When they do not find it, they leave. And that exit shows up nowhere as a problem.

The four things internal search tells you

What you sell but call something else. The most common case and the easiest to fix. The customer searches "hooded sweatshirt", you called it a "hoodie". The product exists, search does not find it, and the outcome is identical to not stocking it.

What you do not sell but get asked for. Repeated searches with no results are a list of products your audience expects from you. It is the cheapest market research there is, and it arrives pre-segmented to people already on your site.

How they think when buying. If they search by occasion ("graduation gift"), by problem ("red wine stain"), by compatibility ("filter for model X"), they are telling you how they wish the catalogue were organised, and it rarely matches how it is organised now.

What is missing from your pages. If they search for an attribute your products have but that is not written in the fields, search cannot find it. Every no-result search on an attribute is an empty field in the catalogue.

The number to watch, and it is not how many searches happen

The metric that matters is the share of searches ending with no results, and right after it the share that return results but no click. The second is subtler and often worse: it means the engine returned something, just not what was needed.

People who use internal search typically convert far better than people who browse, because they already know what they want. Which means every failed search costs more than any other failed visit.

What to do, in order of cost

Synonyms, immediately. Take the top twenty no-result searches and see how many are products you stock under a different name. On a catalogue nobody has curated, it is usually more than half. They are fixed with a synonym list or by adding the terms to product fields.

Missing attributes, next. Searches by characteristic that find nothing point at empty fields. Filling them serves internal search, filters, the feed and the engines: the same work paying four times.

Pages for recurring questions. If a search by occasion or problem keeps coming back, that is a collection page that does not exist. Building it serves the customers searching inside and the ones searching on Google.

And for what you do not sell: decide. Either stock it or say so. A page explaining why you do not carry it and what you recommend instead converts better than an empty screen.

The "no results" screen is a sales page

In most stores it is a dead end: one line of text and nothing else. It is the last useful moment before the person leaves.

It should carry at least three things: a correction if the term was mistyped, the nearest products by category, and a way to ask a human. Somebody who searched for something specific and did not find it is exactly who is worth catching in chat.

How Uptonica handles this

Product fills attributes across the whole catalogue, and it is the same work that makes internal search, filters, the feed and chat answers work. Empty fields are why a search misses a product you stock.

See what Product does

Frequently asked questions

Where do I find this data on Shopify?

Shopify records internal searches and surfaces them in reports, with terms and no-result cases. If you use an external search app, the data lives there and is usually richer. Either way nothing needs configuring: it is already being collected, waiting for somebody to look.

How many searches do I need for this to be useful?

Fewer than you would think, because you are not looking for a statistical phenomenon, you are looking at words. Even fifty distinct searches already show the pattern of missing synonyms, and that is most of the value.

Improve the search engine or improve the data?

The data, almost always first. A smarter engine over a catalogue with empty fields still has little to read. And improved data keeps its value if you change engine one day, while the reverse is not true.

Is it worth it with few products?

With few products search gets used less, but the signal stays valuable because it is concentrated: if ten out of a hundred searches are for something you do not stock, that is a strong indication rather than noise.

Does internal search help with SEO too?

It is one of the few intent datasets you genuinely own, rather than estimated by an external tool. The words people use on your site are nearly always the ones they use on Google, and unlike estimated volumes these come from people who are already your customers, or nearly.