See how much traffic ChatGPT sends you
Visits from ChatGPT, Perplexity, Claude, Gemini and Copilot are counted as their own channel, next to search and social.
Learn about AI traffic →Use casesA screen for every question.
Every case below is a screen you have on day one. The clips show Vesper, a demo store, never anyone’s real numbers.
01Where customers come from
Which AI assistants send you visitors, and what those visitors do next.
Find which pages assistants send people to
The AI referrals card lists the pages assistants link to, and whether those visitors convert better than everyone else. It is rarely the homepage.
Learn about AI traffic →See which source brings paying customers
Send a value with the event that means someone paid, and each payment is credited to the source that brought the visitor. A source with few visitors and most of the revenue stands out.
Set up revenue by source →Compare ChatGPT visitors with Google visitors
Click a source and every view narrows to those visitors: their pages, their conversion rate, their devices. Click another and compare.
Learn about filters →See whether the newsletter drove signups
Campaign tracking joins the UTM tags on your links to conversions, so you can see how many signups the last send produced.
Learn about campaigns →See what happened on launch day
After a Product Hunt or X launch, the chart marks the day and the briefing says where the visitors came from and what they did. Or ask the question in your own words.
Learn about signals →Keep bots out of your numbers
Crawlers and bots are detected and left out of your counts and your bill. Overview shows how many were blocked.
Learn about bot filtering →02What they do on the site
Where visitors go, where they convert, and where they stop.
See where visitors leave before signing up
User flow traces sessions page to page and shows the page where most of them stop. The briefing points at it before you go looking.
Learn about user flow →Track signups, trials and purchases with one line
Call trackEvent with a name, or put one attribute on a button. Every event gets its completions, conversion rate and value, and Clerion draws the funnel to the one you pick as the win.
Set up custom events →Ask leaving visitors why
List the pages you care about in the snippet and visitors about to leave one are asked once, from five plain answers. Too expensive, or not sure what it does, says more than a bounce rate.
Set up the exit question →Find the element people keep clicking
Three or more fast clicks on one element are recorded on anything, not only links and buttons, so the thing that looks clickable and does nothing shows up by name.
Read about rage clicks →See how far people read
Scroll depth, active time and clicks per session show whether a page is read or skimmed, with no setup.
Learn about user flow →Fix mobile conversion with one list
Ask for a mobile conversion fix list. Clerion joins the funnel, load time and bounce data into one list, worst leak first.
Learn about Ask Clerion →03Whether the site is working
Pages, speed, errors and deploys, from the same script tag.
Check whether AI can read your site
Clerion reads each page's titles, headings, meta tags and schema the way an assistant does, and scores every page. The weak ones come with a note on what to fix.
Run the free check →Find the pages with traffic and no description
The site briefing lists the pages visitors reach that have no meta description or no schema, so you fix the ones that matter first.
Learn about AI-readiness →Find the slow pages
Load time and Core Web Vitals per page, with a note on what to improve. Slow pages also show up as signals when they cost you visitors.
Learn about performance →Catch a broken signup form before anyone tells you
Script errors are caught by the same script tag, grouped by page and counted by sessions affected, with the visitor's last actions and a note on the likely fix.
Learn about error intelligence →See whether a deploy broke something
Add one line to your deploy job and every deploy is compared with the day before it: error rate and conversion, with changes for the worse in red. The briefing names the deploy behind a drop.
Set up deploy markers →04What to do about it
Clerion finds the patterns in your data and tells you what to do, in writing.
Read the one fix to make first
Open the Overview and the briefing is already written: what changed, why, and the fix to make first, ranked by the goal you set. Every number comes from your data.
Learn about the briefing →Know what changed without looking
Signals list every change on your site with the biggest first: a spike, a drop, a page that stopped converting. Each one links to the sessions behind it.
Learn about signals →Plan ads for the regions where you are weakest
Ask for a Facebook ads plan and Clerion names your weak countries, their device mix and the landing pages that convert there.
Learn about Ask Clerion →Write the traffic section of your investor update
Ask for it and Clerion writes it with exact numbers: sessions, growth, signups, and the one thing that changed.
Learn about written work →Have the missing meta descriptions written
Clerion finds the pages with traffic and no description, then drafts the descriptions for your top pages, ready to paste.
Learn about written work →Have the bug ticket written for your developer
Ask which bug to fix first and get a ticket to paste: steps, the likely cause on your stack, sessions affected and the breadcrumbs.
Learn about written work →Ask from Claude or Cursor instead
Connect your own model over MCP and it reads the same figures as tools, where you already work. Change the model and nothing about your analytics changes.
Connect your model →Do them on your own site
One script taglive in a minute