60,000+
brands scored on how AI describes them
AI brand measurement and execution
AI recommends the brands it understands.
We measure whether it understands yours, and fix what it gets wrong.
Every week, we run your buyers' real questions against ChatGPT, Claude, Gemini, and Perplexity, and score the answers against what's actually true about you. Then we do something about the gaps.
No credit card required. We only read published information on your website, and never sell your data. Read our privacy policy.
How it works
Measure. Fix. Measure again.
Most tools tell you whether you were mentioned. That's a number, not an answer. Optimly runs a loop that ends with evidence about what actually changed.
-
01
Establish what's true
Your Business Profile is the owner-verified record of what you sell, who you sell to, and what makes you different. Every answer AI gives about you is scored against it. Without that, you can measure presence but not accuracy.
-
02
Run the questions your buyers actually ask
Weekly, across ChatGPT, Claude, Gemini, and Perplexity. Not branded searches — the unbranded category questions where a recommendation actually gets made.
-
03
Find where the description fails
Every answer is parsed for the sources the model used, which produces a citation graph for your brand. You see which pages AI pulls from, which competitors it substitutes for you, and where it hedges or omits you entirely.
-
04
Fix the thing underneath
Positioning, proof, entity data, review surfaces, earned placement. A senior strategist decides what to do. You approve it. We ship it, onsite and off.
-
05
Re-measure against the lag
Retrieval takes four to six weeks to move. Every action is logged with a timestamp so the re-measurement is attributable to something specific rather than to drift.
Senior brand judgment decides what to do, you approve it, and we deliver the work. Each cycle makes the next recommendation better, because we know what worked last time.
AI Brand Index
The data layer AI reads about brands.
We score 60,000+ brands on how AI systems describe them, and AI agents query the index directly when they look up companies in a category. That gives us something no monitoring tool has: a view of what AI systems actually fetch, verified through crawler data, across an entire market rather than one account.
33,900+brands scored in the AI Brand Index
across ChatGPT, Claude, Perplexity, and Gemini
4 models
retrieval measured across the frontier models
Crawler-verified
evidence of what AI systems actually fetch
Your brand is probably already in it. Claiming your profile is free.
Why Optimly
Being mentioned is not the same as being understood.
We score accuracy, not presence.
Every competitor reports whether you showed up. Almost none report whether what was said was correct, because scoring accuracy requires a verified record to score against. We built that first.
Retrieval is what makes AI accurate about you.
In our latest run, 49% of answers were grounded in retrieved sources. Grounded answers scored 0.59 on accuracy. Ungrounded scored 0.15. A model that can retrieve you is roughly four times more likely to describe you correctly — which is why the work is about being findable and verifiable rather than being loud.
We tell you what to stop doing.
Most of what's sold as AI optimization is volume. We've watched wire-service press releases fail to move citations across multiple clients and categories, and we say so rather than billing for them. A recommendation engine that only ever adds to your list isn't giving you judgment.
Citation share drifts on its own, in both directions, with no intervention at all. That's why we attribute at the page level rather than the account level — read the methodology.
The contrast
Volume is the expensive strategy.
Ship 600 pages
- Real cost in staff time and money.
- Site-level risk if a spam update lands.
- Wins one engine, temporarily.
- Nothing in it that a competitor couldn't also produce.
Ship one asset only you could make
Forum Ventures built one page we recommended. It earned 550 citations across 400 buyer questions and became the third most-cited asset they own, behind their homepage and their accelerator page.
Their AI-referred sessions converted at 12.1% against roughly 0.3% site-wide. When their total traffic fell 17.4%, the AI channel fell 4.9%. Measured in their analytics, not ours.
Results
What customers have shipped.
Forum Ventures
One recommended page earned 550 citations across 400 buyer questions.
AI House
Their FAQ page became the single most-cited page on their site.
Right Side Capital
Two articles went from zero citations to 8 and 7.
Page-level attribution, matched to the four-to-six week retrieval lag. We don't claim account-level movement as a result, and we report the accounts where nothing moved.
Pricing
Start free. Add measurement. Add execution when you want the work off your plate.
Free
$0
Claim your profile, one AI Brand Snapshot, one recommendation.
Optimly Platform
$800/month
The weekly measurement run, citation graph, and prioritized queue.
AI Brand Execution Program
From $4,000/month
Twelve actions a month, delivered onsite and off.
Find out what AI says about you.
The snapshot is free and takes a few minutes. If what comes back is wrong, we can tell you why, and we can fix it.
No credit card required
Common questions
The simple version.
Is this just an agency?+
An agency sells hours on a retainer. We sell measurement and a fixed number of delivered actions per month, with a record of what each one changed. The measurement runs whether or not you buy execution, and the execution is priced by work delivered rather than by time spent.
What's the difference between the free snapshot and the platform?+
The AI Brand Snapshot is one look at how AI describes you today. Optimly Platform is the standing instrument: the same questions run every week, scored the same way, so you can see change rather than a single reading.
What counts as an action?+
A discrete piece of work delivered, onsite or offsite. An offsite action means the asset is built and submitted. Placement decisions belong to third parties, so we commit to the work rather than to the outcome.
Do you guarantee citations?+
No. We measure whether citations moved and tell you when they didn't. Anyone guaranteeing placement in an AI answer is either guessing or gaming, and gaming is what gets demoted.
Which models do you measure?+
ChatGPT, Claude, Gemini, and Perplexity, weekly.
What is a Business Profile?+
Your owner-verified record of products, customers, positioning, and facts. Optimly drafts it from your website; once confirmed, it is the ground truth every AI answer is scored against.