Site search and merchandising: helping shoppers find the product
On-site search matters because the people who use it have told you what they want to buy. If your search box returns nothing, returns the wrong thing, or buries the product on page three, you lose a shopper who was ready to spend. The built-in search on Shopify, Adobe Commerce and the other main platforms is good enough for a small, tidy catalogue. It starts to cost sales once the catalogue is large, technical or badly described, and that is when a dedicated tool such as Klevu, Nosto or Luigi's Box can pay for itself.
The decision should come from your own numbers, not from a vendor demo. Your analytics already hold the evidence: how many visitors search, which searches return nothing, and how many searchers go on to buy. This guide covers what to look at, what the built-in tools do, what the paid ones add, and when to leave well alone.
The short version
- Search and merchandising are the same job: putting the right product in front of the shopper, whether they typed a query or opened a category.
- Fix your product data and use the free tools on your platform before you pay for anything.
- Three reports tell you most of what you need: searches with no results, searches with no clicks, and conversion rate for visitors who search against those who don't.
- Dedicated tools earn their fee on large or complex catalogues, where typo tolerance, better ranking and merchandising rules change what shoppers see.
- No search tool fixes thin product titles, missing attributes or products filed in the wrong category.
- A search tool is a monthly cost and a dependency on every page, so set a target before you sign and measure against it.
Why search and merchandising matter
A shopper who types "blue waterproof jacket size 12" has done your segmentation for you. They are not browsing. In most stores we look at, visitors who search convert at a noticeably higher rate than those who don't, although the gap varies a great deal from store to store. Check your own figure before you assume it.
Merchandising is the other half. It is the order products appear in on a category page, which filters are offered, and what is pinned to the top during a promotion. Most category pages are still sorted by whatever the platform defaults to, often newest first or alphabetical. That order has nothing to do with margin, stock depth or what sells.
What does built-in search do, and where does it run out?
Shopify
Shopify's own Search & Discovery app is free. It lets you create synonym groups, boost chosen products for chosen search terms, set up filters on collection and search pages, and choose related and complementary products for product pages. It also has its own analytics. Shopify lists semantic search (matching on meaning as well as exact words) as available on the Grow, Advanced and Plus plans for stores under 200,000 products, at the time of writing (October 2026).
For a catalogue of a few hundred well-described products, that is enough. We'd configure it properly before considering anything else, and many Shopify stores never do.
Adobe Commerce and Magento Open Source
Magento Open Source relies on a search engine you host, OpenSearch or Elasticsearch, and what you get from it depends on how well it has been set up. Adobe Commerce adds Live Search, Adobe's hosted search service, which replaces the default storefront search rather than running alongside it. If you are on Adobe Commerce and not using it, start there.
Where built-in search runs out
- Language. Misspellings, plurals, part numbers typed with or without spaces, and trade terms your product titles don't use.
- Ranking. Built-in search mostly ranks by text match. It does not learn that one of your forty "black dress" results is the one people buy.
- Merchandising control. Pinning, burying, boosting by margin or stock, and scheduling those rules around a campaign are limited or manual.
- Scale. Thousands of products with many attributes need filters that adapt to the query. A filter list designed for the whole catalogue is clumsy for a narrow search.
- Insight. Basic reports show what was searched. They rarely show which queries lose money.
What Klevu, Nosto and Luigi's Box add
Dedicated tools replace the search box, the results page and often the category pages. They index your catalogue, learn from clicks and purchases, and give your team a console for rules. The three we work with most overlap heavily, with different centres of gravity.
| Tool | What it is strongest at | Worth knowing |
|---|---|---|
| Built-in (Shopify Search & Discovery, Adobe Live Search) | Synonyms, basic boosts and filters at no extra cost | Right first step for small or simple catalogues |
| Klevu | Search, category merchandising and recommendations | Since 2025 part of Athos Commerce, alongside Searchspring and Intelligent Reach |
| Nosto | Personalisation across search, categories, recommendations and content | Also includes A/B testing and pop-ups, so it can replace more than one tool |
| Luigi's Box | Search, product listing, recommender and search analytics | Offers a 30-day free trial at the time of writing (October 2026) |
None of the three publishes a simple price list that we could confirm, and quotes generally depend on catalogue size and traffic. Ask each for a price based on your own numbers, and ask what happens to the price when traffic grows.
What you are paying for, in practice:
- Search that copes with typos, natural phrasing and product codes.
- Results ranked by what shoppers click and buy, not only by matching words.
- Merchandising rules your trading team can change without a developer.
- Autocomplete that shows products, categories and suggestions as the shopper types.
- Reporting that ties a search term to revenue.
We implement Klevu, Nosto and Luigi's Box, and we tell a fair number of retailers they don't need any of them yet.
How do I tell if search is costing me sales?
Use data you already have. On Shopify the reports are in the Search & Discovery app and under Analytics. In Google Analytics 4, enhanced measurement records a view_search_results event with the search term when your results page URL uses a standard query parameter such as "q" or "search". If yours uses something else, it needs adding in the settings, and it is worth checking that the event is firing at all.
- Measure how many sessions include a search. If it is a tiny share, check whether the search box is easy to find on mobile before concluding nobody wants it.
- Compare conversion rate and revenue per session for sessions with a search against sessions without. Searchers should convert better. If they convert the same or worse, search is failing them.
- Pull the searches with no results. Shopify has a report for exactly this. Sort by volume. Each line is either a product you don't sell (useful buying information), a product you do sell under another name (a synonym to add), or a spelling the engine could not handle.
- Pull the searches with no clicks. Results appeared and nobody chose one. That is a ranking or relevance problem.
- Search your own store for your top 20 terms on a phone. Note where the best-selling product appears. This takes half an hour and is often the most persuasive evidence in the room.
- Put a value on it. Take the sessions that hit a no-results page, apply your normal searcher conversion rate and average order value, and you have a rough ceiling on what better search could recover.
That last number is your business case. If the recoverable revenue is a small multiple of a tool's annual cost, the free fixes are the better use of time. If it is many times the cost, a dedicated tool deserves a proper trial. Our analytics team can set this reporting up if it is not there.
Common mistakes
- Buying a tool to fix a data problem. If a product is not tagged with its colour, no engine can filter by colour. Clean titles, attributes and categories first.
- Installing and walking away. The gains come from someone reviewing the no-results list every month and adjusting synonyms and rules. Without an owner, a paid tool performs like the free one.
- Ignoring speed. Third-party search adds scripts to every page. A slow autocomplete can cost more than poor relevance. Test page speed before and after.
- Judging by demo. Every demo catalogue searches well. Ask for a trial on your own products and your own awkward queries.
- Buying overlapping tools. Search, recommendations, personalisation and testing are often sold separately by vendors whose products now cover all four. Check what you already pay for.
When a dedicated tool is the wrong choice
If you sell fewer than a few hundred products, if only a small share of visitors search, or if most of your revenue comes from a handful of hero products reached from ads and email, better search will not move the total much. Spend the money on product pages, delivery messaging or the work described in our guide to conversion rate optimisation.
How we'd approach it
- Baseline. We make sure search is tracked properly, then pull the reports above for the last 90 days.
- Free fixes. Synonyms for the top no-results terms, redirects for common queries such as "returns" or "delivery", product data gaps closed, and the search box made obvious on mobile.
- Re-measure after a month. Often this is enough for a small catalogue.
- Shortlist, if the numbers justify it. One or two tools matched to your platform, catalogue and team, trialled against a target agreed in advance.
- Run it as a programme. A monthly review of search terms and merchandising rules, as part of wider optimisation work, with someone named as the owner.
Questions we get asked
Is Shopify's built-in search good enough?
How much does a site search tool cost?
Klevu, Nosto or Luigi's Box: which should we choose?
Will a search tool slow the site down?
Want a second opinion on your own setup? Thirty minutes with a strategist, and you leave with a clear next step.