Search Isn’t Broken
It’s solving yesterday’s problem.
Every few months, another headline declares that ecommerce search is broken. Site search conversion is too low. Customers abandon after a bad query. The fix, we’re told, is always more search: better autocomplete, smarter synonym matching, AI-powered natural language queries that finally understand what the customer “really” meant.
I don’t think ecommerce search is broken. I think we’ve been asking it to do a job it was never built for.
Search has always had one core assumption baked into it: the customer already knows what they want, and they’re just trying to locate it faster. Type “trail running shoes size 9,” get trail running shoes in size 9. That’s not a discovery tool — it’s a retrieval tool. And retrieval tools are only useful to people who’ve already done the discovering somewhere else.
The problem is, fewer and fewer people are doing that somewhere else. And the “somewhere else” that remains isn’t a search bar at all.
The Old Model: Ecommerce Search as the Front Door
For most of ecommerce’s history, site search sat at the center of the experience because it mirrored how people searched everywhere else — Google first, then the store. The mental model was linear: need forms, need gets typed into a box, box returns the answer. Every investment in commerce search — from basic keyword matching to today’s AI-assisted search — has been an investment in making that box smarter.
That investment made sense when the box was genuinely the starting point of the journey. But it stopped being the starting point a while ago, and the shift happened faster than most commerce teams have adjusted for. Increasingly, the moment a customer forms the idea “I want this” doesn’t happen on your site or on Google at all. It happens in a feed, mid-scroll, before they were looking for anything.
The Shift: Ecommerce Search as a Late-Stage Tool
Here’s the pattern I keep coming back to: the customers who actually search on your site aren’t undecided. They’re often the most decided people in your funnel — they’ve already been inspired somewhere else, formed an idea, and are now trying to locate it inside your store. Search isn’t losing relevance because it’s a bad tool. It’s losing relevance as the place where the customer’s journey begins, because that job has moved somewhere else entirely.
Just over half of Gen Z shoppers now say they discover products most often on Instagram and TikTok rather than Google, according to eMarketer research — a reversal that would have sounded absurd a decade ago, when “search engine optimization” was practically synonymous with “how customers find you.” That’s not a generational quirk that will age out. It’s a preview of where discovery lives for everyone: in feeds built around interest and momentum, not in boxes built around typed intent.
What that means for commerce teams is uncomfortable but clarifying: ecommerce search doesn’t need to be reinvented as a discovery engine. It needs to be demoted to what it’s actually good at — fast, accurate retrieval for customers who already know what they’re looking for — while the real discovery investment moves upstream, into the content and merchandising that shape intent before anyone opens a search bar at all.
"Search isn't the problem. It's just no longer the whole story — and treating it like the whole story is what makes it look broken."
Why This Matters More Than It Looks Like It Does
I’ve watched brands sink real budget into “fixing” search — new NLP models, semantic matching, AI query understanding — chasing a conversion lift that never fully materializes, because the underlying problem was never search quality. It was that too few customers were arriving at the search bar with a formed idea in the first place. You can build the smartest search engine on earth and it won’t manufacture intent that doesn’t exist yet.
This is why I keep coming back to what we covered in Chapter 1: The Future of Product Discovery — the real growth lever isn’t a smarter box, it’s a stronger front door. Search still matters. It’s just no longer the place where the relationship with a customer starts. It’s the place where a relationship that started somewhere else gets completed.
I think of it as two very different jobs wearing the same interface. Job one: help someone who has no idea what they want yet find a reason to want something. Job two: help someone who already has a reason locate the specific thing that satisfies it. A search bar can only ever do the second job well, because it requires the customer to already have language for what they’re looking for.
Treating search as a late-stage confirmation tool rather than an early-stage discovery tool also changes what “good” ecommerce search actually looks like. A search bar built for confirmation should be fast, forgiving, and precise. A search bar mistakenly built as a discovery tool tends to do the opposite: it tries to guess intent that isn’t there yet, and ends up slower and more frustrating for the exact customers it should be serving best.
What To Do Differently
The fix isn’t a bigger search budget — it’s a different allocation entirely. Audit where your actual discovery is happening. If it’s on social platforms, content, or word of mouth, that’s where the inspiration investment belongs, not in a fourth iteration of your search algorithm. Then let search do the job it’s actually suited for: get already-decided customers to the right product with as little friction as possible.
For endurance and wellness brands especially, this reframe matters. Someone doesn’t search “electrolyte powder for POTS” or “recovery drink for skaters” out of nowhere — they arrive at that query because a piece of content, a story, or a community moment planted the idea first. If that upstream moment doesn’t exist, no amount of search sophistication downstream will manufacture it.
Once you stop asking ecommerce search to do a discovery job it was never built for, it gets a lot easier to see what search is actually good at — and to invest accordingly.
