Marketing attribution decides which of your channels gets credit when a new patient books. Most guides assume an online checkout and unlimited tracking. A practice has neither: many new patients phone, and patient data can't go to ad platforms. This page explains the attribution models with one example patient, shows where phone calls break them, and recommends a model a small practice can run every month.
In short: Marketing attribution is the rule you use to assign credit for a new patient to the channels that patient touched before booking, such as a Google search, an ad or your Google listing. A practice should pick one simple rule, apply it the same way every month, and check it against what new patients say at intake.
What marketing attribution is
A patient rarely books after one contact. They search, read, leave, see an ad, come back and call. Each of those touches was a channel you paid for in money or time. Attribution answers: which of them earned this patient?
The answer drives budget. If you credit only the last step, your Google listing looks like a hero and your blog looks useless. Credit only the first step and the reverse is true. Neither is the whole story, which is why the choice of model matters more than the software.
Practice owners and marketers search for this regularly:
| Search (US, monthly) | Searches |
|---|---|
| marketing attribution | 720 |
| multi touch attribution | 720 |
| marketing attribution models | 480 |
| last click attribution | 170 |
Google Ads Keyword Planner data via DataForSEO, US, 12-month average to Sep 2026.
Attribution is one input to measuring marketing ROI; the other is an honest count of new patients.
The models: first-touch, last-click and multi-touch
Illustrative example. A fictional patient, Dana, has low back pain. On day 1 she searches "physical therapy for back pain" and reads a blog post on a fictional clinic's site. On day 6 she taps a Meta ad from the same clinic. On day 9 she searches the clinic's name, opens its Google listing and taps Call. She books and completes a first visit.
Here is how each model credits Dana:
| Model | Organic blog post (day 1) | Meta ad (day 6) | Google listing call (day 9) |
|---|---|---|---|
| First-touch | 100% | 0% | 0% |
| Last-click | 0% | 0% | 100% |
| Linear multi-touch | 33% | 33% | 33% |
| Position-based (40/20/40) | 40% | 20% | 40% |
- First-touch rewards whatever introduced the patient. Good for judging awareness channels; blind to what closed the deal.
- Last-click rewards the final step. Easy to measure; it over-credits branded searches and your listing, which mostly catch people other channels already persuaded.
- Multi-touch spreads credit across steps. Linear splits evenly; position-based favors the first and last; data-driven models weight each step by its observed effect.
What Google Analytics 4 offers today is narrower. Its Help Center lists three attribution models (opens in a new tab): data-driven, paid and organic last click, and Google paid channels last click. First click, linear, time decay and position-based were removed in November 2023. In GA4's attribution settings (opens in a new tab), the lookback window for most key events defaults to 90 days, with 30 or 60 as options.
Why phone calls break most attribution
Look again at Dana. Her decisive action was a call from the Google listing. Your website analytics never saw it: she didn't visit the site on day 9. In GA4, the blog visit and the ad click exist, but neither led to an on-site action, so neither gets credit. The new patient simply doesn't appear.
That is the normal case for practices, not an edge case. Calls escape attribution when patients:
- dial from your Google listing or a map app;
- dial from memory, a sign or a referral card;
- switch devices between seeing an ad and calling;
- call the number they saved last year.
Three partial fixes exist. Google's Business Profile performance report counts calls from your listing (opens in a new tab), though not where those people first heard of you. Counting taps on the phone number on your site, by visit source, catches calls from the site. And call tracking assigns different numbers to different channels, at the cost of a vendor that needs a BAA if it records calls.
What you can track without patient data
You don't need patient identities to attribute marketing. You need counts by source.
HHS's guidance on online tracking technologies (opens in a new tab) says tracking on login-protected pages generally has access to PHI, and that public pages where people schedule appointments or use a symptom checker can involve PHI too. A vendor that receives PHI is a business associate needing a BAA. In June 2024 a federal court vacated the part covering an IP address plus a visit to a public health-condition page; the rest stands. HIPAA's Privacy Rule (opens in a new tab) places no restrictions on properly de-identified information.
What that leaves you, safely:
- Visits and actions by source from analytics on public pages: sessions, phone-number taps, booking-button clicks and direction taps, attributed to the referring channel.
- Tagged links. Tag links with UTM parameters on your listing, ads and emails, so the source is explicit.
- Listing reports. Calls, website clicks and direction requests from Google.
- Intake counts. "How did you hear about us?" recorded in your practice-management system and exported as monthly totals by answer.
- Aggregate conversion counts to ad platforms, if you share anything at all. Never names, emails, phone numbers or conditions.
This is general information, not legal advice. Confirm how it applies to your practice with a qualified advisor.
Medfy's first-party analytics attributes calls, forms, bookings and directions to their sources on your own site, and the AI CMO dashboard shows leads by source in one view. Ad platforms never receive patient data.
Choosing a model for a small practice
Use the simplest model that you will run every month without fail:
- Default to last-click from your own analytics for on-site actions. If you use GA4 for practices, "paid and organic last click" is the closest built-in option.
- Add the intake question as a second, independent view. It catches word of mouth, referrals and listing calls.
- Compare the two monthly. Where they agree, act. Where they disagree, the gap tells you which channel analytics can't see.
- Look at first-touch quarterly for content and social, which rarely win the last click but often start the journey.
- Turn attribution into money. Divide each channel's cost by its credited new patients to get cost per new patient.
Data-driven attribution learns from the conversion paths in your own account. With a handful of tracked conversions a month, it has little to learn from, and the result is hard to explain to the person approving your budget.
Common attribution mistakes
- Counting leads as patients. A form fill isn't a patient. Attribute completed first visits.
- Crediting your brand name. Branded searches mostly harvest demand other channels created. Report them separately.
- Trusting each platform's own numbers. Google Ads and Meta each claim the same patient. Use one source of truth for the total.
- Switching models mid-year. You lose the trend. Change only with a clear reason, and restate the old months.
- Ignoring offline sources. Referrals and signage never show in analytics. The intake count is the only place they appear.
- Putting analytics on patient forms. It creates HIPAA risk and adds little. Read is Google Analytics HIPAA compliant before tagging any form page.
A quick way to see what you're tracking today, and what you're missing, is a free marketing audit. Its ads check flags privacy risks such as pixels on patient forms.
Frequently asked questions
What is the best attribution model for a small business?
For a small practice, last-click from your own website analytics, checked every month against a how-did-you-hear question at intake, is the most useful starting point. It is simple, everyone can understand it, and the intake answer corrects its blind spots. Multi-touch and data-driven models need more conversions and cleaner tracking than most small practices have.
How do you attribute phone calls to marketing?
Count taps on the phone number on your website by the source of the visit, and read calls from your Google listing in its performance report. Ask every new caller how they heard about you and record it as a count. If you need more detail, a call tracking vendor can assign numbers per channel, but it needs a Business Associate Agreement if it records or transcribes calls.
What is multi-touch attribution?
Multi-touch attribution splits the credit for a conversion across several of the steps that led to it, instead of giving it all to the first or last one. A linear rule splits it evenly; a data-driven model weights steps by how much they appear to matter. It shows the supporting role of channels like blog posts and social ads that rarely get the last click.
Does GA4 attribution work for healthcare websites?
Partly. GA4 can attribute on-site actions such as phone taps and booking-button clicks to the source of the visit, on pages where it is safe to run. It cannot see calls dialed from your Google listing or from memory, and Google says it does not sign BAAs for Analytics, so keep it off forms, portals and booking pages and treat its numbers as a partial view.