Back to all articles

Searches with no results: what to fix on your website

What do people search for on your website but fail to find? Use Informio Search statistics to choose and verify useful content improvements.

Informio··6 min read

Magnifying glass over an empty wooden catalogue compartment, illustrating the search for missing information

Your website explains warranty claims, but a visitor searches for how to return a broken item. A council has published a collection schedule, yet a resident asks when branches will be picked up. If the search returns nothing, another article about your organisation will not solve the problem. First establish whether the missing piece is information, a clear name or a properly connected source.

Search history can provide a working list for the website editor. The GOV.UK Service Manual recommends search logs and support data as evidence for understanding user needs. The process below is our practical suggestion, not a promise of a particular increase in sales.

An empty result does not identify the fault

Start with one query and ask: which page or product should appear? If you cannot name it, the search engine may not be at fault. The visitor may want a service you do not offer, or information you have never published.

If the right page exists, open it. Check that the content is current and publicly accessible. Then verify that it belongs to the sources selected for Search and that its latest version has been processed successfully. Only then consider changing the wording or explaining a term. Editing AI instructions cannot repair a missing source.

What Informio Search shows

Informio search statistics offer 7, 30 and 90-day periods, popular queries and a history that can be filtered to searches with no results. You can narrow the selection by assistant, query and AI or text mode. Each search detail retains its summary and ordered results, so you do not have to guess what the system returned at that moment.

The history records submitted searches, not autocomplete suggestions or rejected requests. The number of records is therefore not a visitor count. Your organisation's statistics are available to its authorised users; they are not a public list of questions.

Two examples with different fixes

Consider a hypothetical printer shop. A search for “print from my phone” produces no suitable product, although the shop sells one. The editor discovers that the product feed contains only the technology's name. A useful fix is to add the verified capability to the public description and synchronise the source again. Phone compatibility must be checked against the manufacturer's documentation; a search for a feature does not make that feature real.

In a second hypothetical example, a resident searches for “branch collection”. The date is buried in a document called “Biodegradable waste schedule”. The council could publish a clearly named public page with the current date and a link to the valid schedule. In Search, a result card needs a public web or XML source URL. A supporting PDF alone does not become a standalone result card.

Choose one fix you can verify

At each review, choose one recurring intent. Consider similar phrasings together, but do not merge different needs just because they share a word. A warranty claim and returning an undamaged item, for example, may require different instructions.

  1. Record the original query and the expected destination page.
  2. Identify the cause: missing content, unclear wording, an unselected source or an unsuitable result.
  3. Edit the source content and assign someone to confirm its accuracy.
  4. Once the source has been processed successfully, repeat the original search and a natural variation of the question.
  5. Check the complete result, its position and destination link. Keep the date and outcome in a working note.

You maintain that note yourself; we are not claiming that Informio automatically creates tasks or approves content changes. Return to it at the next review and check whether the same problem recurs.

Do not chase zero at any cost

You can calculate the share of zero-result searches by dividing their count by all recorded searches under the same filters. Keep the period length consistent when comparing and examine the mix of queries too. Seasonal topics or a small number of searches can change the figure substantially.

No empty results does not prove that search works well. The engine might return an unrelated page and technically still have a result. Open records with results as well. Text mode does not automatically mean failure either: what matters is whether the visitor received a suitable link.

A search is not an order or a completed application

This Informio Search report does not measure clicks, conversions or revenue. If you need to track a completed form or purchase, design separate measurement. Having another analytics tool does not guarantee it captures an embedded search: Google Analytics documentation ties automatic site-search detection to configured URL query parameters. Check your particular integration.

A sensible first goal is smaller: choose one recurring problem, fix its cause and verify that the right information can be found. To apply this process to your business or council website, see Informio Search and arrange a demonstration using your own public sources.

Create an AI assistant for your website

Start without a payment card or API key. Default Informio AI is ready automatically, while your own OpenAI account or AI server remains optional.

FREE plan · 100 answers · no payment card