How Referenced works.
From prompt to position.
Give it your domain. It works out what buyers ask, asks every answer engine daily, and tells you whether the answer named you — and what to change if it didn't.
No card · first scan free · 4 minutes to set up
Set up your tracking in minutes.
Paste your domain. Referenced reads your site, works out what you sell and who you sell against, then proposes the prompts your buyers actually type — in their wording, not marketing's.
Keep the ones that matter, rewrite any of them, add your own. Nothing runs until you say so.
See exactly where you stand.
Visibility, share of voice, position and sentiment — measured the same way every morning. Every answer stored verbatim, every citation kept. Filter by prompt, engine, topic or day.
See why the answer says what it says.
A score you can't interrogate is a rumour. Every run keeps the answer word for word, the brands it named in order, and the pages it was built from — yours, a competitor's, or a review site you've never heard of.
That's how a drop stops being a mystery: you can read the sentence that replaced you.
Know what to fix, in order.
Every scan is checked for the gaps that cost you answers: a prompt you never appear on, a competitor's page doing the work yours should, a claim an engine keeps getting wrong.
Each one comes with the answer that triggered it and a draft of the fix. You publish it — nothing is written to your site.
Answer the £150 price question on your own page
Nine prompts about price under £150 name Merrell and inov-8. Your product pages never state a price band.
Write the Salomon vs Hoka comparison
Both engines cite Hoka's own comparison page. There is no equivalent on your domain to cite instead.
Correct the outdated Speedcross review
Two engines still quote a 2022 review describing the old outsole. The review site accepts updates.
Add sizing detail to the Mudclaw comparison
Answers that mention fit skip you entirely — no engine has a sentence about width to quote.
Set it once and it keeps working.
Scans run on a daily schedule, without you asking. You get told when something moves enough to matter — not every day it doesn't.
Everyone who needs the number has it.
Seats come with your plan — invite your marketing team or your agency. On Pro and Agency, give a client read-only access to just the sites you choose.
It watches every engine that answers.
ChatGPT, Google AI Overviews and Perplexity on every plan. Claude, Gemini, AI Mode, Grok and Meta AI as add-ons — the same prompts, the same day, so the numbers stay comparable.
Frequently asked questions
The practical detail behind setup, the daily scan, and what the numbers do and don't claim.
How long does setup actually take? + −
Four minutes in most cases. You paste your domain, Referenced reads the site and proposes a business profile, 5 competitors and 50 prompts. You confirm or edit them and the first scan starts immediately — results are in the dashboard within the hour.
Where do the prompts come from? + −
Referenced reads your site and proposes prompts in the wording real buyers use, not marketing copy, grounded in your confirmed category, market and competitors. You keep, edit or add to the list before anything runs, and only the prompts you keep count towards your plan or ever get scanned.
How often does it run? + −
Once a day, automatically, once you've confirmed your prompts — there's nothing to trigger by hand. Each run asks every engine you've enabled the same prompts and stores the answers, so trends build scan over scan rather than resting on one lucky result.
What exactly is the visibility score? + −
It's the share of your tracked prompts where an engine named your brand in its answer, tracked alongside share of voice — how much of the conversation is yours rather than a rival's — average position and sentiment. All four are measured the same way on every scan.
Does it write or publish anything for me? + −
No. Referenced drafts a fix for each gap it finds — a reply, a pitch, a content brief — and shows you the answer that triggered it. You review and publish it yourself; nothing is written to your site or posted on your behalf automatically.
Can I see the actual answers? + −
Yes. Every engine's answer is stored verbatim, word for word, along with the brands it named and the pages it cited, so you can read the exact sentence behind any score rather than taking a number on trust.
Which engines are included? + −
ChatGPT, Google AI Overviews and Perplexity are included on every plan, including the free trial. Claude, Gemini, Grok, Meta AI and Google AI Mode are available as premium add-ons from $29 a month on Brand, added whenever you want them from the Models page.
Who on my team can use it? + −
Every plan includes at least one seat: Brand has one, Pro has 10 and Agency has 25, at no extra cost, so you can invite your team or an agency partner. Anyone without a seat can still see the numbers as a read-only viewer on just the sites you choose: Pro includes 3, and Agency includes more.
How visible is your brand in AI answers?
Tell us your domain and we'll send you a visibility report — free, no card.