Browsy queries are the SEO concept that needed a name

Liz Reid went on Odd Lots this week and dropped a phrase Google has apparently been using internally for a while: *browsy queries*. The context was AI search behaviour — why some users go straight to AI Mode, why others stay on the classic SERP, and why Gemini gets a different traffic mix entirely.

The phrase landed without much fanfare. It should have landed harder.

Because "browsy queries" is the missing vocabulary for a behaviour pattern SEOs have been tripping over for two years without being able to name it. And the moment you have the name, a lot of recent strategy advice starts looking wrong.

What Reid actually said

The full quote is worth reading carefully. Reid was distinguishing between three surfaces — classic Search, AI Mode, and Gemini — and explaining that user behaviour isn't monolithic across them. Informational queries skew toward Search and AI Mode. Creative and productivity queries skew toward Gemini. AI Mode pulls complex, multi-step questions where users expect to do follow-ups.

And then this: *"if you're doing a very browsy query, you might choose to prefer all of the SERP."*

A former DeepMind engineer's LinkedIn fills in the gap. She built a machine learning model at Google specifically to identify "browse intention" queries — things like *"best places to visit in Orlando"* — and surface engaging content for them on the SERP. The model lifted global CTR by 5%.

So this isn't a casual phrase. It's a classification Google has been working on internally, with its own model, and apparently its own SERP treatment.

The three-mode model nobody quite drew

Most SEO content over the last 18 months has treated AI search as a binary: either the user gets an AI Overview and bounces, or they get a traditional SERP and click through. Reid is describing something more useful — a three-mode model based on what the user is actually trying to do.

Resolve. The user has a specific need. They want one answer. *"Restaurant in Shoreditch for five, one vegan, kids welcome, under £40 a head."* This is what AI Mode is built for. The user articulates the problem, gets a synthesised answer, and either acts on it or asks a follow-up.

Browse. The user is exploring. They don't have a fixed destination in mind. *"Best places to visit in Orlando."* They want options, images, comparisons, the lateral richness of a full SERP. An AI Overview that picks for them is actually the wrong format — it collapses the choice they came to make.

Create. The user wants the AI to do work. *"Rewrite this email to sound more formal."* This goes to Gemini, ChatGPT, Claude. It's not really search at all.

Most of the GEO advice circulating right now treats every query as if it's in mode one. Optimise for citations. Structure your content for extraction. Get into the AI Overview. That's coherent strategy for resolve queries. It's actively counterproductive for browse queries — where the SERP is what the user wants, and the AI Overview is the format Google itself is choosing not to serve.

Why the browse mode is the one SEOs should actually care about

Resolve queries are the ones generating all the apocalyptic GEO commentary. They're the ones where AI Mode answers without a click. They're the ones where the publisher economy is genuinely under pressure.

The browse mode is where SEO still does most of its work, and almost nobody is naming it as a category.

But they're not where most commercial intent lives.

Commercial discovery is overwhelmingly browsy. *"Best running shoes for flat feet."* *"Marketing agencies in Bristol."* *"Things to do in Edinburgh in November."* *"WordPress hosting comparison."* Users in those moments are not trying to get an answer and leave. They're trying to weigh options, see what's out there, build a shortlist. Reid is telling us — explicitly — that Google knows this and is preferring to show those users the full SERP.

Which means the click economy for commercial queries is not collapsing the way the AI-search-killed-SEO crowd has been claiming. It's bifurcating. Resolve queries are getting compressed. Browse queries are still mostly SERP-shaped, and the SERP is where 18 years of SEO fundamentals still apply.

The browse mode is where SEO still does most of its work, and almost nobody is naming it as a category.

What this changes about query research

Keyword research has always implicitly mixed these modes together. You'd export a list of phrases from a tool, sort by volume, and pick the ones with commercial intent. The mode of the query — resolve versus browse — was never a column.

It needs to be now. Because the optimisation playbook is genuinely different for each.

For browse queries, the old game still works. Comprehensive content, comparison tables, images, structured data that helps Google build a rich SERP entry, internal linking that signals you cover the category. You're competing for SERP real estate, and the SERP is what's being shown.

For resolve queries, the game is different. You're competing to be one of two or three sources cited inside an AI answer. Brand authority matters more than article depth. Schema matters more than word count. Being the source the model has seen ten thousand times matters more than being the source with the best one-pager. The fundamentals are the same — domain authority, useful content, technical hygiene — but the surface is doing different work.

If you treat them as the same problem, you'll over-invest in resolve-mode tactics for browse-mode queries and wonder why your traffic is fine but your conversions are flat.

The diagnostic the industry has been missing

Reid's framing also suggests a diagnostic that the SEJ piece on "why AI search skips your content" gestures at without quite landing.

Most agencies right now are running AI visibility audits that ask one question: *is your content being cited in ChatGPT and Perplexity?* That's a useful question, but it's incomplete without a second one: *what mode are the queries in your category?*

If your category is dominated by resolve queries, citation tracking is the right metric and you have a real problem if you're invisible. If your category is dominated by browse queries, citation tracking is a vanity metric — your users aren't using AI Mode to find you in the first place, and the absence of AI citations is mostly noise.

The honest audit looks at query mix first. What proportion of your priority queries are resolve, browse, or creative? Then it looks at performance per mode. Then it tells you where the actual leak is. Most audits skip step one and start measuring downstream of a question they haven't asked.

What I'd do with this on Monday morning

If I was sitting down with a client this week, the exercise would be straightforward. Take your top 50 commercial keywords. For each, do the search yourself and note what Google actually shows — full SERP, AI Overview, AI Mode prompt, mixed. Tag each query as resolve or browse based on the user need behind it. Compare the tag to what Google chose to serve.

You'll find three buckets. Queries where Google's mode-detection matches the user need and the SERP looks how you'd expect. Queries where Google is showing a richer, browsy SERP and your job is classical SEO. And queries where Google is collapsing the answer into AI and your job is to be one of the two or three brands cited.

Each bucket needs different work. Treating them as one undifferentiated optimisation problem is why so much current GEO advice is producing so little measurable lift.

The honest limit

Reid is one source, and "browsy query" is not yet a public taxonomy. We don't have Google's actual classifier, we don't have query-level data on how the modes split, and the proportions in any given vertical will vary wildly. Anyone telling you that 60% of queries are browse and 40% are resolve is making it up.

What we do have is Google's own product lead naming a category, an internal Google ML model built around it, and a coherent behavioural model that explains a lot of recent observations better than the binary AI-vs-SERP frame.

That's enough to start using it as a lens. It's not enough to build a measurement framework on yet. The measurement will come — and the agencies that get there first will be the ones who started thinking in three modes instead of two while everyone else was still arguing about whether SEO is dead.

The industry needed a vocabulary for this. Now it has one. Use it.