The customer who types into your search bar has already done the hard part. They've arrived on your site, formed an intent, and asked for something specific. They are, by any measure, the most qualified visitors you have, and yet most enterprise commerce teams have no idea who owns the experience that follows.
On-site search optimization sits in a quiet gap between merchandising, marketing, and IT, and the absence of clear ownership produces a slow, expensive form of decay across the user experience layer of the entire site. Across the industry, the search bar converts at two to five times the rate of general browse traffic. Searchers spend more per order, return more often, and represent a disproportionate share of total revenue.
Walk into a typical enterprise team and ask who is responsible for the quality of that experience, and you'll get a shrug or a name buried in the engineering organization who hasn't reviewed search analytics in a year. The single most valuable surface on the site quietly goes unowned.
Why search is the most qualified traffic on your site
A shopper landing on a homepage is browsing. A shopper landing on a category page is interested. A shopper typing into the search bar is shopping.
The act of typing a query is a customer telling you exactly what they want, in their own language, with no friction between intent and signal. There is no more honest data in the entire customer journey. Search intent is captured at its purest in the moments before a shopper has decided which products and services to evaluate, and that signal is harder to fabricate than any cohort or segment built from behavioural inference.
Searchers convert at multiples of non-searchers across virtually every category — apparel, electronics, home goods, and B2B distribution — because they've already done the hardest part of any sale, which is deciding what they're looking for. The search bar is, in effect, the customer pre-qualifying themselves and then handing you a brief.
For enterprise retailers, search traffic often accounts for 20 to 30% of revenue, despite representing a smaller share of sessions. The conversion math is that lopsided. Yet because search lives within a technical surface, it tends to fall outside the natural ownership boundaries of merchandising, marketing, and CRO teams. The result is that the most targeted traffic on the site is treated as the responsibility of nobody in particular.
The cost of a "good enough" search engine
Most enterprise teams stand up search at launch, configure a baseline of synonyms and filters, and never revisit it. The catalogue grows. Customer language drifts. New product attributes are added. Old ones go dark. Merchandising priorities shift weekly. None of it makes its way into the search layer.
The downstream effects are quietly enormous. A "no results" page is, in the data, a customer who told you exactly what they wanted and walked away empty-handed. A relevant but poorly merchandised result page is almost as bad — it shows the right products in the wrong order, and conversion can collapse 20 to 40% against a properly ranked result set. Filter taxonomies that don't match how customers shop force people to scan rather than refine, and most don't bother. Personalization signals built into the search engine go unused because nobody has tuned the ranking weights to align with the brand's actual customer behaviour.
None of these failures look dramatic in analytics. They look like flat conversions. They look like a slow leak.
The cost of unrecovered no-results queries
A no-results page deserves to be treated as its own discipline within search merchandising, not as a fallback state. We've framed it as a customer walking away empty-handed; in practice, no-results pages are also some of the highest-density data the business has access to. Every null search is a documented gap between what customers want and what the catalogue, the synonym dictionary, or the product taxonomy is currently able to deliver.
The teams that take no-results seriously do three things on a recurring basis. They review the top null queries weekly, treating them as the most actionable signal in their analytics. They distinguish between three failure modes: catalogue gaps where the product genuinely doesn't exist on the site, language gaps where the product exists but is indexed under different terminology, and ranking failures where the product exists, is indexed correctly, but ranks below the displayed result threshold. Each failure mode points to a different intervention, and conflating them is one of the most common mistakes we see at audit.
They also instrument the page itself. A well-designed no-results experience offers recovery paths: spelling correction, suggested categories, recently viewed products, and a clear secondary call to action. Doing nothing here leaves targeted traffic to bounce silently. Done correctly, no-results pages reclaim a meaningful share of otherwise lost sessions and, over time, become the most reliable input into the synonym and merchandising backlogs that improve search for everyone else.
The four levers of search merchandising
Modern on-site search is governed by four interlocking disciplines, none of which is purely technical.
Synonyms
They are the dictionary that the search engine uses to interpret queries. "Joggers" and "sweatpants" mean the same thing to a customer but not to a default search engine. New product names, new category names, slang, abbreviations, and seasonal vocabulary all need to be mapped on an ongoing basis. A synonym dictionary that hasn't been updated in a year is producing null searches every day.
Guided navigation and filters
Determines whether customers can refine their search or have to manually scan. Filter design is a merchandising decision, not an engineering decision: which attributes matter most to shoppers, in which categories, in which order. Filter sets that made sense for last year's catalogue often don't for this year's.
Personalization
Uses prior behaviour, such as past searches, past purchases, and past clicks, to reorder results for individual customers. Done well, it materially lifts conversion. Done poorly, or left at default settings, it does nothing.
Merchandising rules
These rules let teams promote, demote, or pin specific products in response to campaigns, inventory, or strategic priorities. Without active rule management, the search results page reflects the algorithm's interpretation of relevance rather than the business's intent.
These four levers interact. Tuning one without the others produces marginal results. Tuning all four together, continuously, is what turns search from a feature into a channel.
Search on mobile devices is a different problem
So far we've treated search as a single discipline, but anyone running search merchandising at scale knows the mobile experience behaves differently enough to warrant its own attention.
Search queries on mobile devices are shorter on average — often two to three words versus four to six on desktop — because typing on a phone is friction. That changes everything downstream. Autocomplete carries more weight on mobile because customers will tap a suggestion rather than finish typing. Synonyms have to do more interpretive work because the input is sparser. Ranking has to handle ambiguity more confidently because users won't refine as patiently. Filters need to be redesigned around tap targets and scroll patterns rather than the desktop sidebar default that most platforms ship.
Voice search compounds this. A growing share of mobile sessions begin with a spoken query, which tends to be longer, more conversational, and phrased as a question rather than a keyword string. The search engine has to handle "do you have running shoes for wide feet in size 11" with the same competence as "wide running shoes 11." Most don't, out of the box.
The right way to handle this is not to maintain two separate search configurations but to recognize that the same four levers — synonyms, guided navigation, personalization, merchandising rules — need to be tuned with mobile behaviour as a first-class input, not an afterthought. Accessibility considerations follow the same logic: voice input, screen-reader compatibility, and high-contrast result displays affect more users than most teams assume, and the same instrumentation that improves the mobile user experience usually improves accessibility as a byproduct.
Search as a measurable revenue channel
The brands extracting full value from their search investment treat it the way they treat paid search: with weekly attention and clear KPIs.
Null search reports get reviewed. Top queries get audited against the result quality. Click-through and conversion rates get tracked by query type. Synonyms get added for trending terms. Merchandising rules get adjusted for promotional priorities. Filters get pruned and reshaped based on actual usage. Personalization weights get tuned as the customer base evolves.
This isn't search engineering. It's search merchandising — and it sits in a discipline gap between marketing, merch, and IT, which is exactly why it goes neglected at most enterprise brands.
Instrumenting search performance
Treating search as a channel is the principle. The execution depends on instrumentation that most enterprise commerce teams haven't actually put in place.
The minimum set of KPIs for a search program is narrower than people think. Search-driven conversion rate, separated from site-wide conversion. Click-through rate on the first three positions of the results page. Null search rate, ideally trended weekly. Search-to-purchase revenue, attributed correctly. Filter usage rate by category. Synonym hit rate against trending queries. Personalization lift, measured against a non-personalized control where the platform allows.
The teams that operate search well don't necessarily track more than this. They track these reliably, weekly, with someone whose job it is to read the report and act on it. Dashboards without owners produce no improvements. Reports without rituals produce no improvements. The instrumentation matters only to the extent that it feeds a decision loop.
Most enterprise teams we audit have access to all of this data inside their search platform. They just haven't connected it to a meeting, a person, or a set of standing actions. That's the real gap, not the data.
What ongoing search maintenance looks like
A maintained search experience is built on a simple operating rhythm.
Weekly: review null searches and top queries, add high-priority synonyms, and adjust merchandising rules for active campaigns. Monthly: audit filter usage by category, review search-driven conversion against site baseline, refresh personalization weighting against the latest behavioural data. Quarterly: rethink filter taxonomies against catalogue evolution, run structured A/B tests on ranking logic, align with merchandising on seasonal and inventory priorities.
The work isn't complicated. It's relentless. Which is why most teams, without dedicated ownership, don't sustain it.
Who owns search at well-run companies
The article opened on a problem: in most enterprise commerce organizations, no one owns search. The closing question is what good ownership actually looks like.
The pattern at companies running search well is a single named owner — usually inside the merchandising or ecommerce team rather than engineering — with a defined operating cadence, access to the search platform's configuration layer, and a regular reporting line into the broader commerce leadership. The role is part merchandiser, part analyst, part product manager. It does not require deep technical skill, because the underlying platforms have matured to the point where most of the daily work is configuration and judgment rather than code.
Where the ownership sits matters less than that it sits somewhere. At some companies it lives with the head of ecommerce, at others with a dedicated search merchandiser, at others with a digital product manager whose remit includes discovery surfaces broadly. What's consistent is that the role exists, has a name, and is held accountable to specific search performance KPIs alongside the rest of the commerce P&L.
The companies that haven't named this owner are not failing because search is hard. They're failing because search is sitting in the discipline gap between three teams, none of whom believe it's theirs. The reframe is straightforward: assign it, fund it, and measure it. Most of the work that follows is operational rather than strategic, which is exactly why it tends to be undervalued and exactly why it pays off.
The reframe or search results
Search isn't a feature you launch. It's a channel you operate.
If 20 to 30% of revenue passes through search, a 10% lift in search conversion translates to a 2 to 3% lift on the entire business. There are very few projects with that kind of leverage. There are even fewer that compound, but search does, because every improvement to synonyms, ranking, and merchandising rules continues paying out on every future query.
The brands winning here don't have better search technology than their competitors. They have someone whose job it is to make sure search keeps working.
