The sales pitch for Consent Mode often includes one line: Google will fill in the conversions you lose when visitors decline cookies. That is true only when certain volume conditions are met, and a single-clinic website in a mid-sized city will often not meet them. If you assume modelling will repair your numbers and it does not, you will under-count enquiries and misjudge which campaigns work.
This article is for clinic owners, practice managers and marketing leads at small and mid-sized providers in India and the GCC. It states the thresholds Google publishes, shows with hypothetical arithmetic whether a typical small clinic would reach them, and sets out what to do when it does not. The thresholds are Google's, and they can change, so check the linked help pages before you rely on them.
Two different kinds of modelling
People say "modelled conversions" as if there were one thing. There are two, in two products.
GA4 behavioural modelling estimates user and session metrics for traffic where analytics storage is denied. It improves what you see in GA4 reports.
Google Ads conversion modelling estimates conversions from ad clicks where consent was not given. It affects the conversion counts that Ads uses for reporting and for bidding.
They have separate thresholds and separate effects. A clinic can qualify for one and not the other.
The thresholds, as Google states them
For GA4, the behavioural modelling help page says the property needs:
- At least 1,000 events per day with
analytics_storage denied, for at least 7 days. - At least 1,000 daily users sending events with
analytics_storage granted, on at least 7 of the previous 28 days. - Consent Mode implemented across all pages, with tags loading before the consent dialog appears.
Google adds that meeting the thresholds does not guarantee eligibility, because the model also considers factors such as the ratio of new to returning users. It also says that training may take more than seven days of meeting the data threshold in a 28-day window, and the data may still not be sufficient.
For Google Ads, the consent mode modelling help page lists two checks: consent mode (or the IAB TCF v2.0 framework) must be correctly implemented, and there is "a daily ad click threshold of 700 ad clicks over a 7 day period, per country and domain grouping".
Google's wider Ads description of consent mode adds a point that is relevant for small advertisers: in basic mode, where tags stay blocked until the visitor chooses and nothing is sent if they decline, modelling relies on a general model. In advanced mode, cookieless pings allow an advertiser-specific model. Whether cookieless pings are acceptable for your organisation is a question for your counsel; our article on server-side tracking and consent mode for healthcare covers the consent side.
Would a small clinic reach them? Hypothetical arithmetic
These numbers are invented to show the method. Replace them with your own.
Clinic A, a dental clinic in a tier-2 city, receives about 250 visitors a day to its website, each generating about 6 events. Suppose 80 visitors a day decline analytics cookies.
- Denied events per day: 80 visitors × 6 events = 480. That is below 1,000.
- Granted daily users: 170, far below 1,000.
Clinic A does not meet the GA4 thresholds, even if all of its consent handling is perfect.
Clinic B, a multi-branch fertility chain, receives 4,000 visitors a day with 8 events each. Suppose 1,200 decline.
- Denied events: 1,200 × 8 = 9,600, above 1,000.
- Granted users: 2,800, above 1,000.
Clinic B plausibly qualifies for GA4 modelling, subject to Google's other factors.
For Google Ads, the arithmetic is about clicks, not visits. At 700 ad clicks over seven days, a clinic needs an average of 100 ad clicks a day to the same domain from the same country. A clinic spending a modest daily budget at a cost per click typical of local searches may be at 40 or 50 clicks a day, and would not reach the number. Note that Google groups by country and domain, so splitting traffic across several domains makes it harder, not easier.
The binding constraint for small sites is almost always the count of consenting users or ad clicks, not the number of denied events.
Why the lack of modelling matters less than it sounds
A small clinic is unlikely to buy on thousands of conversions a day anyway. If the clinic gets 30 enquiries a month from search ads, three or four of them hidden by consent choices will not change a sensible decision. What changes decisions is a large and growing gap between enquiries in your own records and conversions in the platform.
So measure the gap yourself.
A method for measuring the gap
- Pick a 28-day window with stable campaigns.
- Count real enquiries from your own systems: form submissions logged in the booking system or CRM, phone calls from your call tracking, WhatsApp chats opened. Use sources that do not depend on cookies.
- Count conversions reported in GA4 and in Google Ads for the same window and the same definition.
- Divide the platform count by your own count for each channel. Call this the capture ratio.
- Repeat monthly. The ratio is more informative than any single month.
- Segment by consent state. If your consent platform reports the share of visitors who accept, compare the capture ratio to that share. A big difference suggests tagging problems rather than consent.
If the capture ratio falls after you introduce a consent banner, the fall reflects consent choices and tagging order. Fix tagging order first: defaults set before measurement, the update sent promptly, tags firing on the right trigger.
What to do if you do not qualify
- Use first-party records as the source of truth. Count enquiries in the CRM and compare channels by source. The post on offline conversion tracking for clinics shows how to bring appointment outcomes back into reporting.
- Add call tracking. Phone calls are often the largest enquiry channel for small clinics, and call tracking does not rely on browser cookies. See call tracking for healthcare.
- Use fewer, clearer conversion actions. Give Google Ads one primary conversion that reflects a real enquiry. Fewer conversions with clear definitions help bidding more than a long list of micro-events.
- Work with longer windows. Evaluate campaigns monthly, not daily, when volumes are low.
- Do not weaken consent to raise numbers. Defaulting to granted or hiding the reject option to lift conversions makes your measurement look better and your compliance worse.
- Consider server-side tagging only for control, not for recovery. It improves cookie durability and filtering, but it does not create consent where a visitor refused it.
Reading Google's own conditions carefully
Three details in the published conditions catch small advertisers out.
First, the GA4 requirement that tags load before the consent dialog appears. A banner that blocks the Google tag until the visitor clicks cannot feed the model, because no consent-denied signal ever reaches Google. That is basic mode behaviour, and it is a legitimate choice, but then GA4 behavioural modelling is not available to you.
Second, GA4 modelling covers user and session metrics. Google's page says it does not apply to event counts such as page views in standard reports, and that several features are unsupported, including audiences, user explorer, retention reports, predictive metrics and data export. A clinic that relies on exports to a data warehouse will not see modelled data there. To see modelled numbers in reports, the reporting identity must be set to Blended.
Third, losing eligibility is silent. Google states that if a property falls below eligibility, estimated data stops and returns only from the date the property qualifies again. A clinic that qualifies in a campaign month may drop out in a quiet month, and its trend lines will contain a break. Annotate your reports when eligibility changes, so nobody interprets the break as a performance change.
India and GCC context
For visitors in India, the DPDP Act and its phased Rules are the legal frame, and your counsel decides what consent your analytics and advertising tags require. For the Gulf, the UAE and Saudi Arabia have their own personal data laws and, for health data, sector rules. Google's modelling thresholds are the same everywhere; what differs is your obligation to ask. A GCC-facing clinic with one domain and low click volume should assume it will not receive Ads modelling and plan for first-party measurement.
Where to start
- Pull the last 28 days: ad clicks per country, GA4 events with consent denied, GA4 users with consent granted.
- Compare with the published thresholds above.
- Build the capture-ratio table from your own enquiry records.
- If you do not qualify, set up call tracking and CRM source reporting before you spend more on ads.
- Review quarterly, since volumes and Google's rules both change.
For help setting up measurement that does not depend on modelling, see our analytics and tracking service and the Google Ads guide for healthcare.
Filed underconsent mode modelled conversionsconsent mode low trafficGA4 behavioral modeling thresholdGoogle Ads conversion modelling clinicconsent mode small clinicmodelled conversions India

About the author
Arush Thapar
Co-Founder & CTO · Gurugram, India
Arush builds the systems the rest of the company runs on — the patient-acquisition pipelines, the AI chatbots, the analytics that tie a single click to a booked appointment. If a campaign reports a number, it's because something his team built is tracking it.
Free guide · 15 pages
The Patient Acquisition Blueprint
A 90-day plan from first search to booked appointment.