ClinicGrower knowledge base
Marketing Attribution for Clinics: Read the Evidence Carefully
By ClinicGrower · Updated · Guidance for clinic owners and their teams
In this guide
Start with the outcome and its source
Choose the outcome you need to understand: an accepted enquiry, confirmed consultation or another approved business event. A contact click and a booked consultation answer different questions. Record the date range, cost scope, observation window and system supplying each outcome.
Use booking and payment systems to reconcile operational totals. Advertising platforms and analytics apply different attribution rules, identifiers and reporting dates, so their totals need not match. Keep unattributed outcomes visible. Do not add two platforms' claimed conversions together and call the result unique new patients.
Understand what the model is doing
A last-click model assigns credit according to its final eligible interaction rules; this is not always the literal last visit. Google's paid-and-organic last-click model normally excludes direct traffic when an eligible non-direct interaction exists. Data-driven attribution distributes credit using the observed patterns available to the platform.
Neither model supplies a complete causal history. Missing consent, browser restrictions, offline conversations and unconnected systems can limit the evidence. GA4 no longer offers first-click, linear, time-decay or position-based reporting attribution models. Do not follow an old instruction to select position-based as a fallback.
Review the current attribution reports
In the relevant GA4 property's Advertising area, locate the key-event attribution paths report and select the specific outcome being reviewed. Check the date range and source, medium or campaign breakdown. Compare available models using the same outcome and period; record the chosen settings in the report.
Look at observed paths and time to the event as context. A channel appearing early in a path may deserve investigation, but its presence does not prove incremental value. People who never complete the selected event are not represented by that event's completed paths. Small or incomplete samples need cautious interpretation.
Use consistent campaign labels
For appropriate external campaign links, agree a short naming convention for source, medium and campaign. For example, a ClinicGrower business webinar link could use ?utm_source=newsletter&utm_medium=email&utm_campaign=clinic_reporting_webinar. Use non-sensitive labels and keep a shared register of their meaning.
Do not put names, email addresses, patient identifiers or clinical information in URLs. Avoid using campaign tags on ordinary internal navigation. For Google Ads, verify the account's auto-tagging and analytics integration before adding manual values that could confuse reporting. Tags label observable visits; they do not capture every exposure or restore withheld consent.
Connect business outcomes proportionately
Maintain agreed enquiry-source and outcome fields in the clinic's authorised operational system. Record unknown values honestly and avoid retrospectively guessing a source. Check duplicate enquiries, cancellations and the lag between enquiry and treatment before comparing periods.
Only consider offline imports or server-side advertising integrations after reviewing purpose, consent where required, data minimisation and current provider restrictions. These connections do not authorise uploading clinical records or remove healthcare-related limits. Use aggregated management reporting when a person-level advertising connection is unsuitable.
Turn the findings into a bounded decision
Write down a specific hypothesis, such as whether one campaign generates more suitable enquiries at a sustainable cost. Review matched cohorts and the clinic's capacity before adjusting a budget. Change one material factor at a time and define the review period in advance.
Cost per accepted enquiry = included campaign cost ÷ valid unique accepted enquiries under the agreed definition. ROAS = attributed revenue ÷ advertising spend; it is not profit or proof of incremental revenue. Assess margin and delivery costs before concluding that a channel is commercially successful.
Use controlled experiments where practical to test whether outcomes change because of an intervention. Attribution supports those decisions; it cannot replace the comparison needed to estimate causation. Private account and revenue investigation belongs in an agreed diagnostic scope, beyond the public Free Clinic Growth Audit.
A clinic example: distinguish the report columns
The Wellness clinic aesthetics and wellness case study reports 169 treatment-form enquiries over a dated 90-day window. Its campaign-level conversions and conversion-action inventory use different totals. The case therefore keeps the action definitions visible and does not manufacture a blended cost per lead or a patient count.
When comparing your own reports, check the date window, selected campaigns, conversion action, reporting column and counting rules before combining figures. A precise description of the evidence is more useful than a larger but ambiguous total.
A practical method for organic search and AI enquiries
Use three separate measures: visibility in a search or AI answer, visits carrying an observable referral, and accepted enquiries that retain that source. A citation can occur without a click. An AI assistant may open a visit without usable referral information. Neither should be converted into an invented organic lead.
For each accepted enquiry, retain the permitted landing page and source alongside a separate outcome record. Recognised paid campaign evidence takes precedence over a search-engine referral. Match actual referral hostnames rather than accepting a hostname merely because it contains a familiar brand. Keep unknown sources and missing consent visible.
ClinicGrower tested this path on 12 September 2026 using eight synthetic scenarios: Google and Bing organic search, ChatGPT and Perplexity referrals, paid traffic, a misleading hostname, withheld analytics consent and an unknown source. The checks covered the website source collector, submission mapping and database persistence using isolated temporary records. They did not send a new live enquiry or email and do not prove that every real visitor can be attributed.
Review a weekly cohort of valid enquiries by source, landing page, qualification and booking outcome. Exclude agreed tests and duplicates. Record incomplete outcomes separately and compare equivalent periods. The Clinic KPI guide explains the denominator rules; the Performance OS overview explains connection limits.
- Visibility: record the provider, question, date and cited URL.
- Visits: count observable organic and AI referrals under the reporting system and consent rules in use.
- Leads: count valid accepted enquiries, then qualified and booked outcomes from the same cohort.
- Unknown: show missing source and outcome evidence explicitly.
Sources and further reading
Continue exploring
Questions clinic owners ask
What can marketing attribution tell a clinic, and what can it miss?
Attribution assigns recorded outcomes to recorded marketing interactions under a chosen model. Consent gaps, offline conversations, device changes and incomplete records limit the picture, so compare it with operational outcomes instead of treating it as proof of causation.
How can a clinic make attribution reports more useful?
Use consistent campaign labels, agreed outcome definitions and comparable reporting periods. Show unattributed activity and missing connections explicitly, then reconcile the report with enquiries, bookings and verified commercial records.