Skip to content

Track leads and customers that arrive from AI tools via Calendly

A mention in a ChatGPT or Gemini answer is not the same as a booked call. To judge whether answer-engine work pays off, you need the referral source attached to the meeting itself, and then to the customer record. This guide shows how to route AI referral data into Calendly bookings, how to check that the values landed correctly, and which reports turn those values into a spending decision.

Why a mention is not a meeting

Citation counting tells you that an AI assistant named you. It does not tell you whether the person reading that answer clicked through, booked a slot, or eventually bought. Decisions about answer-engine optimisation get better when the unit of measurement moves from mentions to meetings, leads and closed customers, and when you can see which assistant sent each one and which page attracted them. The setup below connects those dots without changing how visitors book.

Step 1: install the snippet and rebuild the Calendly embed

Start a trial account and place the provided code snippet on your site; installation instructions are linked from the help centre. Next, take the URL of the Calendly event, personal landing page or routing form you want to use and paste it into the embed code generator in the Calendly setup guide. The generator returns a version of the standard Calendly embed that instructs the widget to populate its UTM fields from the attribution data. Replace your existing embed with that code wherever the booking widget should appear.

Step 2: how the visitor's source is decided

On arrival, the script inspects a set of signals about how the visitor reached the site. It assigns a channel — AI Referral, Organic Search, Paid Search, Paid Social and similar — and stores that classification in the visitor's browser. When the same visitor later completes a booking through the embedded widget, for example a discovery call or free consultation, the stored source data is written into that booking.

Step 3: confirm the data on the meeting record

The referral values are saved with the meeting in Calendly, next to the invitee's name, email address and any answers to your booking questions. Open the individual meeting from the Meetings page and look in the UTM details section. From there you can forward the values to a CRM through Calendly's built-in integrations such as HubSpot or Salesforce, or through an automation tool such as Zapier or Make.

What a captured booking looks like

Take a commercial real estate agency as an example. A business owner asks ChatGPT to recommend an agent for finding larger office space, clicks through to the agency site and books a call with a leasing agent using the embedded widget. The booking then carries the four values listed below. Note the last one: an empty term is the expected result for AI referrals, not a sign that tracking failed.

Step 4: report on AI leads

With the values stored in your CRM or lead database, you can report on meeting volume, lead volume and customer volume attributable to AI, split by the assistant that sent each person, and joined to revenue. Most CRMs include reporting that covers this. Dedicated options include Tableau, Google Data Studio and Microsoft Power BI. A lighter route is to connect Claude or ChatGPT to the CRM and ask a direct question such as how many AI referral leads arrived this month.

Three reports and the decision each one supports

Leads by Channel counts monthly leads per channel, so you can see the AI Referral total beside Organic Search, Paid Search and Paid Social and judge how much answer-engine work matters relative to everything else. Customers by AI tool counts new customers per month from AI referral, split by assistant, which lets you compare ChatGPT, Claude and Gemini side by side; if most AI customers arrive via one assistant, that is likely where your audience already is. Close rate by channel shows the share of leads that become customers for each channel, which is the figure that should move budget.

A verification checklist before you trust the numbers

Work through these checks on a live booking before drawing conclusions from any report:

  1. The embed on the page is the generated version, not the original Calendly embed.
  2. The meeting record shows UTM Medium as AI Referral for a visit that began in an AI assistant.
  3. UTM Source names the assistant, and UTM Campaign shows the domain the visitor came from.
  4. UTM Term is empty for AI referrals; only treat it as an error if other fields are also missing.
  5. The same values appear on the CRM record after the integration or automation runs.
  6. Your close-rate report uses your own closed-won data rather than an assumed percentage.

If the channel is missing entirely, check the snippet and embed first. If only the term is blank, the capture is behaving as designed.

Reading the close-rate example correctly

A comparison of roughly 30% for AI referral against 10% for organic search and 7% for paid search is an illustration of what the report might reveal, not a benchmark to plan against. The reasoning behind a higher AI close rate is plausible: a visitor who has already discussed their need with an assistant tends to arrive better qualified. If your own figures follow that pattern, an AI-sourced lead may be worth several times a paid-search lead, which should influence where time and budget go. Confirm it with your data first.

Getting started

Capturing AI referral data on Calendly bookings is a short setup: install the snippet, swap in the generated embed, verify one meeting record, then build the three reports. Trial length, plan pricing and any setup-time estimate are commercial terms that must be confirmed by the site owner before publication.

UTM values stored with an AI-referred Calendly booking
UTM fieldExample valueWhat it tells you
UTM MediumAI ReferralThe channel bucket the visitor was classified into
UTM SourceChatGPT (or whichever AI tool sent them)Which assistant referred the visitor
UTM Campaignchatgpt.com (the domain they came from)The domain the visit originated from
UTM TermNone (AI referrals don't have a deeper level of detail, so this shows as None)No deeper detail exists for AI referrals; an empty value is expected

Values shown are the example captured for a single illustrative booking; the source and campaign values vary with the assistant and domain that referred each visitor.

Report-to-decision map
ReportQuestion it answersDecision it informs
Leads by ChannelHow many leads arrive each month from AI Referral versus Organic Search, Paid Search and Paid Social?How much answer-engine work matters relative to other channels
Customers by AI toolHow many new customers each month come from AI referral, split by assistant?Which assistant deserves more attention
Close rate by channelWhat share of leads from each channel become customers?Where time and budget should go

Report definitions describe the intended output; actual figures depend on the reader's own CRM data and closed-won records.

Frequently asked questions

Does an empty UTM Term mean AI tracking is broken?

No. AI referrals carry no deeper level of detail, so the term field shows as None by design. Treat the setup as broken only when the medium, source or campaign values are also absent from the meeting record.

Where do I see the captured source on a Calendly booking?

Open the individual meeting from the Meetings page in your Calendly account and check the UTM details section. The referral values sit alongside the invitee's name, email and booking-question answers.

How does the data reach my CRM?

Use one of Calendly's built-in integrations, such as HubSpot or Salesforce, or an automation tool such as Zapier or Make. The values are already stored on the meeting, so the connection only needs to copy them across.

Can I skip building reports entirely?

Yes. Connecting Claude or ChatGPT to your CRM and asking a direct question, such as how many AI referral leads arrived this month, returns the figure without a dashboard.

Are the close-rate percentages in this guide benchmarks?

No. They illustrate what a close-rate-by-channel report might show. Replace them with your own closed-won data before making budget decisions.