Right now
Traffic
Conversionsphone taps, phone copies, emails and form submits
Top pages
Where they came from
Devicesby visit, from viewport width
Browsersby visit
Citiesby visit, resolved at the CDN edge from the connection — never asked of the browser
Statesby visit
Networksthe carrier or ISP each visit came in on — a cloud provider here is almost always a bot
Landing pagesfirst page of each visit
Referrers
Addressesmore than one device at an address usually means one household — unless it is a mobile carrier
How far down they readshare of visits reaching each depth
Converting pageswhere the conversion happened
Form funnelwhich field they gave up on — field names only, never what they typed
Recent eventsclick a row for the full record
Log an inquiry or a jobphone calls, referrals, anything the website cannot see
Check before you type. Where a lead came from is worth only as much as the evidence behind it: ask the caller, or look it up — the call log, the Google Business Profile messages, the email, the text thread. "Probably Facebook" becomes a number on a chart that someone later makes a decision with. If you genuinely do not know, choose Unknown — an honest blank is worth more than a confident guess, and it can be corrected later when the customer mentions it.
From the websiteform submissions, newest first
What the visitor wrote — their name, phone and message — is in the inquiry email and is deliberately not stored here. This is the visit behind it: when it happened, the page they sent it from, and where they came from. "Log this" carries the reference into the form above so the outcome ends up on the same row.
Where they came fromas logged, not inferred
What happenedoutcomes in this window
The log
Reviews over time
What this isreviews as a dataset, not a page
Ratings and dates pulled from the business profile on a schedule, stored here as their own event stream. Bars are reviews received in each bucket, by the date the reviewer wrote them; the line is the running total a visitor would see on the profile. No review text or reviewer name is stored here — those stay on the site, which caches them under the profile API's own thirty-day rule.
Crawl is switched on for everyone right now. It costs money to run each week, so it will become a paid add-on.
How your site looks from outside
Named by an AI search
Crawlsnewest first
What this isinputs, not rankings
Once a week your site is fetched from outside, the way Google and the AI search bots fetch it, and each line is something the site controls: it loads, over HTTPS, robots.txt lets the crawlers in, the name and phone number are in the page itself, the structured data parses. When a line changes from passing to failing or back, a mark lands on the traffic chart at that moment. The AI line asks the AI models people actually use — ChatGPT and Claude, the versions their free and paid plans serve — the question a customer would ask, once each, and shows every name each one listed and where yours came. Information, not a score: it is a sample of what those models answered that day, not what every person sees.
What's new
Core Web Vitals
Event mixeverything the tracker recorded in this window
Invite someonethey click the link once and they are in
Give an email address and the invitation is sent for you. Leave it blank and you get the
link to send however you like — text, WhatsApp, in person. Either way it works once,
expires in a week, and sets up access on the device it is opened with. There is no password
to invent, forget or share.
One address, one person: an invitation sent to a shared mailbox becomes a shared login, and
the log stops being able to say who entered what.
People
Invitations not yet used
Collector version
New access token
Copy this now. It is not stored in clear and cannot be shown again.