In a lot of categories the most valuable customer does not fill out a form. She calls. Healthcare runs on the phone. So do legal, home services, senior living, insurance, and most of what a multi-location business sells. And in account after account I audit in these categories, the phone is invisible: the dashboard counts form fills, the algorithm optimizes to form fills, and the calls that produce the actual revenue register as silence.
The scale of the blindness surprises people who have never measured it. When I have installed call tracking into previously untracked accounts in call-heavy categories, calls have come in anywhere from forty percent to double the volume of form submissions, and the callers convert to customers at multiples of the form-fill rate, because dialing a phone number is a higher-commitment act than typing an email address into a box. Which means the accounts were not missing a slice of their conversion data. They were missing the majority of it, and the better-qualified majority at that.
The compounding damage
Undercounting would be survivable if it were only a reporting problem. It is a steering problem, and the damage compounds quietly. If half your real conversions are calls and none of them reach the ad platforms, smart bidding is learning from the wrong half of your customers. It will dutifully find you more form-fillers and fewer callers, which in healthcare often means more browsers and fewer patients: the person comparison-shopping procedures fills out forms, the person in pain calls. Month over month the algorithm walks the account away from your best buyers, each optimization cycle a little more confidently wrong, the same self-inflicted wound as optimizing to raw CPL, executed through omission. I have watched accounts where reported performance improved for two consecutive quarters while actual booked appointments fell, and nobody could see the divergence because the falling number was never on a dashboard.
Run this test on your own account before you dismiss the problem: pull last month's inbound call log from the front desk or the phone system, count the calls that were genuinely new inquiries, and set that number next to last month's tracked conversions. If the calls are more than a fifth of the total and your ad platforms received none of them, your bidding algorithms are optimizing a business that does not exist.
The fix is mature technology
None of this requires invention. I have run CallTrackingMetrics and its peers across hundreds of campaigns: dynamic number insertion ties each call to its source, keyword, and landing page, calls flow back into Google and Meta as conversions, and suddenly the bidding systems can see the customers who dial. Go one step further and score the calls, by duration at minimum, by disposition if the front desk can manage it, so a ninety-second wrong number stops counting the same as a booked appointment. That scoring step is where the real money is, and it is the step most implementations skip; feeding the algorithm every call teaches it to find dialers, feeding it qualified calls teaches it to find patients, clients, and policyholders.
Here is the operational truth nobody puts in the software demo: the hard part of call tracking is not the tracking. It is the twenty seconds of front-desk behavior after the phone rings. The receptionist who answers, tags the call, and books the appointment is the last mile of your conversion infrastructure, and no marketing team controls that person. The implementations that work treat the front desk as part of the media program, with training, simple disposition buttons, and someone actually reviewing call recordings monthly, because the recordings will also tell you why calls do not convert, which is intelligence no pixel will ever produce.
Healthcare adds a compliance layer, and it is not optional. Call recordings and tracking data in a medical context can carry protected health information, so the configuration decisions are compliance decisions before they are marketing decisions: recording policies, data retention, what gets passed to ad platforms, all of it reviewed like the regulated data it is. I built these frameworks for healthcare clients under HIPAA scrutiny, and the rule of thumb is simple: decide what the ad platform is allowed to know before the tracking vendor decides for you.
If your category rings, measure the ringing. The algorithm can only want what it can see, and right now it cannot see the customers who matter most.