An open AEO research lab

AI decides which businesses get promoted.
We make sure you can see how.

AI search is a black box that decides which businesses get discovered. It changes often, and an industry of expensive tools is happy to keep it confusing. Quoted does the research instead: what makes AI keep citing you, in your market and for your offer. No statistics degree required.

Free — see what your buyers ask AI

We read your site, infer your ideal customer, and generate the questions they ask AI. Takes about 30 seconds.

8,700+ URLs re-measured weekly 3,250+ domains tracked 4 AI engines findings retired honestly when they stop holding

How you start

Free, free, then paid. In that order, on purpose.

You shouldn’t have to pay anyone to find out whether AI search matters for your business. So the first two steps cost nothing and prove everything — and the third only makes sense once you’ve seen the first two.

Free
1

See the questions

Enter your site above. We infer your buyer personas and map the questions they ask AI across the whole buying journey — yours to keep, share, and download.

Free
2

See who’s winning them

We run a live citation check on a sample of your highest-intent questions and show you exactly which sources AI is sending your buyers to today — including whether you’re one of them.

Paid
3

Start your study

Not a dashboard. Practical, easy-to-implement recommendations — with receipts. All of your questions, tracked weekly across four engines and turned into validated plays and content recipes for your market. See the study →

What a study delivers

Not a dashboard. Recommendations with receipts.

Every week your market’s citation data runs through statistical validation, and what survives becomes deliverables like these — real examples from the live corpus, shown the way they arrive in your playbook:

Play Validated Perplexity

Dense internal linking

Pages that stayed cited averaged 115 internal links vs 55 on pages that churned · d=0.50 · n=518 · p<0.0001

In plain words: heavily interlinked pages keep their citations. Ranked target pages included.

Recipe Cross-validated

FAQ schema + off-page authority

Pages with both stayed cited 74% vs 38% without · 1.9× lift · beats the best single ingredient by 24%

In plain words: the combination outperforms either tactic alone — and combinations can’t be copied from a blog post.

Play Avoid ChatGPT

Long, winding sentences

Pages with longer average sentences lost their ChatGPT citations markedly faster · d=−0.72 · held across 6 consecutive weekly re-analyses

In plain words: an avoid-play — we tell you what NOT to spend effort on, which no best-practices list ever will.

Recipe Cross-validated

Careful, qualified claims + shallow URLs

Pages that hedge like a researcher and live near the site root stayed cited 55% vs 30% · 1.8× lift · n=64 · p<0.0001

In plain words: a pattern nobody would have guessed, and no snapshot tool could find — only weekly measurement surfaces it.

Every recommendation is re-verified weekly

Recommendations don’t age in a drawer. Each one is re-tested against fresh data every week — and if its pattern stops holding, we retire it and tell you. No stale advice, ever. No other tool will say that.

And explained monthly, in plain words

The monthly brief covers what changed in your market’s citations, what to do about it, and the evidence behind both — written to be forwarded to your CEO, board, or client without a translator.

The Problem

AI citations shift on a weekly basis, but businesses are still investing like it’s the SEO days.

Most AEO conversations get stuck on which tool to buy. The actual problem is that nobody has run rigorous research on what makes AI search engines cite some content and ignore others—and without that research, every recommendation is improvised.

If you run an agency
Your clients are asking. You don’t have a defensible answer yet.
  • Competing agencies are quoting $8–15K/month AEO retainers. You’re defending a $4K SEO retainer.
  • Quarterly reviews are getting harder. Your client is asking about AI search and your monthly report doesn’t address it.
  • Profound or Stacker showed up in a stakeholder’s inbox, and now you’re answering uncomfortable questions.
  • You added “AEO” to your scope of work, but the tactics in your decks are still the SEO playbook with a coat of paint.
If you’re in-house
You can’t answer five questions your board could ask.
  • What’s actually driving citations to your pages, and what isn’t?
  • Which content investments are working and which are wasted budget?
  • How do you stay on top of engine updates that change weekly?
  • How are you showing up when actual buyers search—not when an analyst tests generic prompts?

The research that backs AEO recommendations is being actively accounted for by AI platforms, and those who aren’t staying on top of the changes will be left behind.

51%
of B2B software buyers now start research in an AI chatbot, not Google. (G2, 2026)
94%
of citations shown by generic AEO trackers don’t match what buyers actually see in persona-aware sessions. (The Glassbox Report, Issue 1)
43.4%
CAGR on the AEO services market through 2035. Brands will hire someone for this. The only question is who. (Dimension MR, 2026)

Decay Beats Dashboards

The tools tell you who won today. We tell you why winners keep winning.

Daily visibility dashboards tell you who AI named today. Then the answer re-rolls, and yesterday’s “win” quietly disappears—no alert, no explanation. Citation decay is a weeks-scale phenomenon: you can’t see it in a daily snapshot, only by re-asking the same questions week after week and tracking which sources keep getting cited. That’s why Quoted measures weekly and models persistence—the half-life of a citation—instead of celebrating mentions that don’t survive the next re-roll. The daily number tells you where you stand. The weekly curve tells you what to fix.

cited 50% ~4.5-week half-life week 1 week 8
The average citation’s trajectory in our corpus. Navy dots are weekly re-measurements — the only way this curve is visible at all.
Pages with comparison tables stay cited twice as often over time. Only persistence tracking can surface a pattern like this. (n=4,329 cybersecurity URLs, p<0.001)
0%
average citation overlap between generic and persona-rewritten phrasings of the same question—a daily rank on one phrasing tells you almost nothing. (The Glassbox Report, Issue 1)
Weekly
the cadence citation decay actually happens on—and the cadence every Quoted persistence model is built from.
Read the decay research →

The Engines We Track

Measured where your buyers actually ask.

Every engine answers the same buyer question differently. Our research shows slight variations in questions result in answers that share as low as 0% of their citations, and engines disagree with each other on a regular basis. So we re-ask your market’s questions on each engine in different ways, every week, and track what causes citations to last on a per-engine basis:

ChatGPT
The largest AI answer surface—a majority of AI-assisted buyer research starts here. Live search over Bing’s index.
Gemini
Google’s assistant, grounded in Google’s own search index—the second-largest consumer AI surface and the closest signal to where Google search itself is heading.
Claude
The fastest-growing assistant and the enterprise favorite—live web search at answer time. Its citation behavior differs measurably from the others, which is exactly why we track it separately.
Perplexity
The research-buyer’s engine—citation-first by design, over its own live crawl. Punches far above its traffic share in B2B evaluations.

Why these four: together they cover the engines B2B buyers actually research on, and each retrieves from a different live index—four independent measurements of the same market, not one measurement four times. When a pattern holds across engines, that’s corroboration; when it doesn’t, that’s a per-engine playbook. Google AI Overviews tracking is next on the roadmap.

Share of consumer AI-assistant usage Worldwide web visits, May 2026 (Similarweb) · navy = tracked by Quoted ChatGPT 53.9% Gemini 27.9% Claude 9.2% Perplexity 1.3% DeepSeek 4.1% Grok 2.4% Copilot 1.3% ~92% of consumer AI-assistant usage is covered by the engines we track

Link Building, Reimagined

Link building works differently in AI search.

The engines don’t weigh domain authority. They cite from a small, stable set of source sites—review platforms, analysts, the trade press—and return to them week after week. We measure exactly which sources drive citations in your space, so your outreach targets what AI actually reads.

~50
source domains drive over half of every AI citation in a vertical—and the top ones are cited every single week. (Quoted research, 8-week window.)
reviewplatform analysts tradepress forums comparisonsites yourplacement the AI answer
The sources the engines actually read in a vertical—and where a placement earns you a seat in the answer.
See how it works for agencies →

The Glassbox Pledge

Built to be trusted, not just bought.

Six commitments, kept in the product itself — not in a values deck. The full pledge lives on our mission page.

1

Our research is free, forever

Everything we learn outside client engagements is published openly in The Glassbox Report.

2

No naked claims

Every finding ships with its evidence: sample size, effect size, confidence tier, and what we don’t know yet.

3

Honest expiry

Findings are re-tested weekly. When one stops holding, we retire it publicly — no stale advice, ever.

4

The data does the talking

Whether or not it’s in your favor — or ours. Anything less, and you’d fall behind.

5

Plain words

If you need a statistics degree to understand us, we’ve failed.

6

Cancel anytime

The study earns its renewal each month with evidence — or it doesn’t deserve one.

And the three questions we actually get

How is this different from Profound, Searchable, or Semrush’s AI tools?

They monitor: who got cited today. We explain: why citations last, measured weekly in your market, validated statistically, and turned into recommendations with the evidence attached. Monitoring tells you the score; we coach the next play. And you can test us for free — the prompt set and live citation check cost nothing.

Is this just SEO with new branding?

No — and we can prove it with data: generic and buyer-realistic phrasings of the same question share as little as 0% of their citations, and AI engines don’t weigh domain authority the way Google does. Different index, different rules. The SEO playbook isn’t wrong; it’s answering a different exam.

What if the research finds nothing useful for us?

Then you’ll know that — with evidence, before you’ve paid us anything. The free steps show you your market’s questions and today’s citation winners. And if you start a study and a recommendation later stops holding, we retire it and tell you. We’d rather lose a line item than ship stale advice.

Your study can’t start retroactively.

wk 1 ✓

The citation data your competitors will wish they had in a year is being generated in your market right now — and every unmeasured week is gone for good. Whoever starts measuring first in a vertical owns a data moat the rest are permanently behind. Start with the free steps; the study is there when the evidence convinces you.