What is ecommerce site search?
Site search is the search box on your own store and the software behind it that decides what it returns.
A shopper types something. The software interprets the query, retrieves matching products, ranks them, and gives the shopper ways to narrow the result. That ranking is the part people underestimate: the tool is not only matching words, it is choosing an order — and deciding what never surfaces at all.
Work out what they meant. Misspellings, synonyms, and phrasing your catalogue does not literally contain. Rain jacket has to reach a product called a shell.
Decide what comes first. Every query returns an order, and something has to produce it — relevance scoring, popularity, margin, stock, or a rule someone wrote.
Let the shopper narrow. Size, colour, price, availability. Only as good as the attributes your product data actually carries.
Let a human intervene. Boost, bury, pin or curate results for particular searches — a recurring reason teams want a different tool.
Four jobs behind a mature search experience. These interviews make clear that filtering and merchandising matter alongside interpretation and ranking.
The short version. Site search is a ranking system with a merchandising surface attached. That second half is why it is so often bought by a merchandising team rather than an engineering one — and why our search is bad and we cannot control our search are different complaints with different fixes.
Where most stores start: native search
In these interviews, most ecommerce teams begin with whatever search the commerce platform includes.
This is what makes site search different from most categories on this channel. For most stores some form of search arrives with the commerce platform, so the first decision is usually not do we need this? but is the one we already have still sufficient?
That shows in how much more these buyers talk about built-in search than anyone else does. Among buyers discussing site search, mentions of native or platform-built-in search run several times higher than among buyers discussing ecommerce platforms, personalization or customer data.
And they describe it in the language of adequacy rather than enthusiasm. One says their setup does what it needs to do. Another: for now, we're satisfied enough where we don't think it's a hot button issue. A third had the budget conversation and lost it — that was one of the biggest pushes I tried to make that leadership just didn't really get behind.
When do buyers upgrade off native?
When they can name the specific thing they cannot do.
The upgrade conversation is much more common in these interviews than the reverse. What is striking is how concrete the triggers are — buyers rarely describe leaving native search because results were bad in general.
A rule the native tool will not support. One buyer wants to drive more into our consumer base when it comes to apparel and gear and needs merchandising rules beyond what they have — noting that their platform's search is not really a merchandising tool.
The ability to intervene at all. A buyer on a third-party tool describes what they gained: the power to build filter trees, block certain searches and give relevant results or hide relevant results — a bit more control over search than Shopify's native search.
A capability the platform does not offer. One buyer relying only on Shopify's search capabilities is evaluating dedicated vendors for AI relevance, conversion optimization, and personalization of search — three requirements at once, only one of which is about AI.
The pattern underneath is that in these interviews native search usually stops being sufficient gradually rather than catastrophically. It keeps working while becoming the reason someone cannot do their job — which is why the trigger is usually a specific blocked task rather than a moment of breakage.
The counter-current: some go back
A few buyers move the other way, and it is worth knowing about without overstating it.
One describes re-platforming and reconsidering the whole stack: we don't really need a third-party search tool. Shopify's native search tool does just as well. So we made that switch. Another, unhappy with their specialist vendor, is strongly considering Shopify's native site search because I believe they've had some updates to it recently.
This is a small number of buyers next to the group going the other way, so it is not a trend. What it is good for is a question: if you are paying for a dedicated tool, can you still name what it does that the built-in one would not? Some buyers describe native search as having improved enough to change that calculation — and the suggestion to re-ask the question is our reading, not a claim buyers make in aggregate.
What do buyers rate the tools?
A wide spread, and the most-rated specialist is not near the top of it.
| Tool | Rating | What it means here |
|---|---|---|
| Coveo | around 7.5 | Too few buyers for a decimal. An enterprise option that appears in evaluations where AI relevance at scale is the requirement. |
| Constructor.io | around 7.5 | Too few buyers for a decimal. Appears in evaluations by buyers leaving native search specifically for relevance. |
| Bloomreach | 7.3 | Spans search, merchandising and messaging, so buyers are rating more than the search box. |
| Algolia | 6.9 | The most-rated site-search tool in our corpus by a wide margin, and it rates below smaller names here. Cost is a recurring complaint in these interviews. |
| Nosto | 6.7 | Sits across search, recommendations and personalization; several buyers run it alongside native search rather than instead of it. |
| Searchspring | 6.3 | A mid-market specialist. Buyers name filters, merchandising and recommendations as where it falls short of what they wanted. |
| Fast Simon | around 5 | Too few buyers for a decimal, and the lowest figure here. |
| Klevu | not rated | Too few buyers rated it to publish a number. |
| Platform-native search | not rated | Too few buyers rate their platform's built-in search as a product for a number to mean anything — which is itself telling about how it is regarded: as part of the platform rather than a tool you chose. |
The spread here is genuinely wide — among the widest of any category we can measure this way — and the most-adopted tool sits in the lower half of it. That is worth holding lightly — the tools serve different sizes of catalogue and different teams — but it does mean the shortlist should not start from adoption alone. Where buyers go when they leave, and what they wish they had checked, are covered on why ecommerce teams switch site-search vendors and what site-search buyers wish they had known.
Want this read against your own stack?
Get my read →Common questions
What is ecommerce site search?
Ecommerce site search is the search box on your own store and the software behind it that decides what comes back. It takes what a shopper typed, works out what they probably meant, and returns a ranked set of products with filters to narrow them. That last part matters more than it sounds: the software is not only matching words, it is choosing an order. Which products appear first, which are filtered out, and which never surface at all are decisions the tool makes on every query.
What does site search software actually do?
Four jobs. It interprets the query, handling misspellings, synonyms and phrasing a catalogue does not literally contain. It ranks the matches, deciding what comes first. It provides filters and facets, so a shopper can narrow by size, colour, price or whatever your catalogue supports. And it gives merchandisers control — rules to boost, bury, pin or curate results for particular searches. Buyer interviews show that filtering and merchandising matter alongside interpretation and ranking, which is why site search is often bought by a merchandising team rather than an engineering one.
Is your ecommerce platform's built-in search good enough?
For many stores, for a while. Native search is where most catalogues start, and buyers describe it in terms of adequacy rather than enthusiasm — one says theirs does what it needs to do, another that they are satisfied enough that it is not a hot-button issue. The honest answer is that native search often becomes insufficient gradually rather than failing outright: it keeps working while slowly becoming the reason a merchandiser cannot do something they want to do.
When should you upgrade from native site search to a dedicated tool?
When you can name the specific thing you cannot do. In these interviews the upgrade conversation is far more common than the reverse, and the triggers buyers describe are concrete rather than general: merchandising rules the native tool will not support, filters and facets a large catalogue needs, or an appetite for AI relevance the platform does not offer. One buyer relying only on their platform's search describes evaluating dedicated vendors specifically for relevance, conversion optimisation and personalisation. Where the complaint is vaguer than that, it is worth checking whether the problem is the tool, the merchandising process, or the product data underneath it.
Do companies ever move back to native site search?
A few do, and it is worth knowing about even though it is the minority direction. One buyer describes re-platforming and concluding they did not really need a third-party search tool because the native one did just as well. Another, unhappy with their specialist vendor, says they are strongly considering the platform's own search because it has improved recently. The counter-current is small next to the number of buyers going the other way, so it is a reason to check whether you still need what you are paying for rather than evidence of a trend.
What is the difference between site search and product discovery?
Site search is the query-driven part of product discovery: a shopper types what they want, and the system interprets, ranks and filters the results. Product discovery is broader and also includes category browsing, recommendations, merchandising and the other ways shoppers find products without typing anything. Many vendors span both, which is why buyers often evaluate them together — and why a tool bought for search can end up judged on its recommendations too.
Does a better site search tool fix bad search results?
Not on its own. Search quality depends heavily on the product data underneath it — titles, attributes, categorisation and images the software has to work from. A tool can rank and interpret, but it cannot invent an attribute your catalogue does not carry. Buyers also describe product-data quality as a major constraint on what a new search tool can do, which is covered in detail on the pages about switching and about what buyers wish they had known.
Get this research made for your stack
Whether you have outgrown your platform’s built-in search depends on your catalogue, your merchandising ambitions and who has to operate it. Do a 15-minute interview and get the read for your store — what peers your size run, what pushed them to switch, and what it cost.
Get my personalized read — 15-min interviewNo password · your interview is anonymized before it ever informs a page like this one.