Snowball — AI Recommendation Index Live-web measurement · updated each cycle

AI recommendation, measured

Buyers ask AI what to buy. Be the brand it names.

AI assistants recommend a small, surprisingly stable set of brands. Most companies never learn whether they're on the list — or how to get on it. Snowball measures how often AI recommends you, shows why, and compounds your presence until you're the answer.

Get your free AI Visibility Brief See the method

Free · your brand across the assistants buyers use · 48-hour turnaround

Measured across ChatGPT Claude — Perplexity & Google next

The shift

Search is being replaced by a shortlist.

Buyers used to skim a page of links and decide for themselves. Now they ask an assistant and get three or four names — a shortlist built before a single click. If you're not on it, you're not in the running, and you never see the loss. It just looks like demand that never showed up.

Why it happens

Not random. Measurable — and driven by your sources.

The assistant doesn't invent a shortlist. It reads a handful of review sources and recommends from them. How many of those you appear in almost entirely determines how often you're named — measured across the whole field, on every major assistant.

6 of 6
The brands AI recommends sit in all six review pages it reads. The ones it never names sit in one — or none. It rewards presence in the pages, not the better product.
0→75%
How often a brand got named — before vs. after closing its source gap. Controlled test, 3 brands.
Every assistant
Same mechanism on every major AI we test. The trusted sources differ; the rule doesn't.

The honest part

Presence alone isn't enough. You have to be in the sources AI actually trusts, with a real product behind you — brands in weak sources still get named 0% of the time. That's why this rewards genuine brands, not spam, and why we target sources by how much each engine leans on them.

The method

How Snowball compounds your presence.

One loop, run continuously. Each pass adds presence; presence compounds into recommendation.

01

Measure

How often each engine recommends you — against your competitors, across the real questions your buyers ask. A rate, not a vibe.

02

Diagnose

The exact sources driving the shortlist, and precisely where you're absent. The gap, named source by source, per engine.

03

Build

Generate the assets that get you into those sources — structured identity, retrieval-ready content, review corroboration. You deploy them; it's software, not a services retainer.

04

Compound

Re-measure and prove the lift. Then again. Presence you've already built makes the next gain easier — the rate climbs on itself.

Small at first, then it rolls. That's the whole idea — a snowball.

What you get

Built for challengers fighting for the shortlist.

If you're a mid-market brand up against incumbents who own every roundup, this is the lever. In plain terms, you get:

  • 01Your recommendation rate across the major assistants, benchmarked against the brands beating you.
  • 02The source gap — the specific places AI looks that you're missing from.
  • 03The generated assets to close it, ready to deploy — then a re-measure that proves it moved.
Start here
Free brief
Your brand, measured across the major assistants, with the gaps that explain it. No call required.
Request your brief
Then, plans from$39/mo

See how AI recommends you — and change it.

Get your free AI Visibility Brief

Free · your brand, both engines · 48-hour turnaround