Judge marketing by the patients it produces
Most marketing reports count clicks, calls and leads. The question an executive asks is how many of those became patients, and what each one cost. This article covers how search is changing under AI answers, why click counts mislead, what the rules on tracking technology allow, what agents do each day, and how we measure marketing by new patients recorded in your own CRM.
Search is changing under AI summaries
People click fewer links when an AI summary appears. In Pew Research Center's study of Google searches, users clicked a traditional result on 8% of visits that showed an AI summary, against 15% of visits that did not. They also ended their browsing session more often after a page with a summary, 26% against 16% (Pew Research Center, 2025).
Patients also ask chatbots directly. In a KFF poll fielded from 24 February to 2 March 2026, 32% of US adults said they had used AI chatbots for health information in the past year (KFF, 2026). If an assistant names two other providers when someone asks for care near your city, that patient may never reach your website or your phone line.
Why click and call counts mislead
A monthly agency report can show calls up while new patients stay flat. Clicks and calls sit early in the path. Between them and a first visit come an answered phone, a coverage check, a booked slot and a patient who shows up. Each step can lose people for reasons the ad platform never sees.
Your CRM already records new patients and, with tracked numbers and form sources, where they came from. Counting new patients by source there gives a cost per new patient by channel. That figure tells you where the next dollar should go, and it can rank campaigns differently from cost per click.
Tracking technology and health data
In June 2024 a federal court in American Hospital Association v. Becerra vacated part of the Office for Civil Rights bulletin on online tracking technologies. The vacated part covered tracking that connects a person's IP address with a visit to an unauthenticated public web page about specific health conditions or providers. The rest of the bulletin stands, including its guidance on authenticated pages such as patient portals (Holland & Knight, 2024). HHS later dropped its appeal (Fierce Healthcare, 2024).
Lawsuits over pixels have continued. BakerHostetler describes a rise in class actions and settlements tied to hospital systems' use of web tracking technology in 2023 and 2024 (BakerHostetler, 2024).
So we treat tracking as a privacy decision first. What, if anything, goes back to ad platforms is set by your privacy officer and counsel. Agents check every page for tags that were added without review. The compliance page covers that daily scan.
Ad platform rules
Ad platforms set their own limits on healthcare. Google's healthcare and medicines policy says some healthcare content cannot be advertised at all, some is allowed only in certain locations by advertisers Google has approved, and several categories, such as prescription drug services and, in the United States, health insurance, need Google certification (Google Ads Help, accessed October 2026). Agents check new ads and pages against the policies that apply to your account before they go to your compliance reviewer.
What agents do
- A morning report on the searches, map listings, reviews and AI-assistant answers you gained or lost overnight, and the new pages and ads competitors launched.
- Recurring technical checks of every page: broken links, missing markup, page speed, and whether AI agents can read the page.
- Weekly budget recommendations tied to the new patients counted in your CRM.
- A claims check on every new ad and page before it reaches compliance.
A person approves every change. Agents prepare the recommendation and the evidence, and your marketing lead decides whether budget moves.
Pages agents can read
AI assistants and agentic browsers read your pages the way a search engine does, then act on what they find. Services, accepted insurance, locations, hours and how to book should be stated in plain text and structured markup, and they should match your map listings and phone scripts. When facts disagree across your site, listings and directories, an assistant may skip you or give a patient the wrong hours. Pages that state these facts clearly also help the intake line answer an assistant's call, which the patient intake page describes.
What we measure
- Cost per new patient by channel
- Inquiry to first visit, by source
- How often AI assistants name you when asked a fixed panel of typical patient questions
Each figure starts from a dated baseline. Our free outside-in audit already looks at search, maps, reviews, AI-assistant answers and tracking from public information, so part of that picture exists before you share any data. See what the audit covers.