This post is for clinic and hospital owners, and for the marketers who report to them, who count phone calls from Google Ads as conversions. Most accounts we see in this space use one number for every campaign, usually 30 or 60 seconds, because it was the default someone typed in during setup. The post explains why one number misfires and how to calibrate it department by department using your own call data.
It does not give you benchmark thresholds. Call length depends on how your front desk answers, whether callers are put on hold, how the line is routed and what the patient is asking. Any universal "right" number would be invented. What it gives you is a method you can run in an afternoon with the data Google Ads and your phone system already hold.
If you are still deciding which ad formats carry the phone number, settle that first; Google is retiring call-only ads in favour of call assets on responsive search ads. This post assumes the call is already being tracked and asks what should count.
What the setting does, and what it does not
When you create a conversion action for calls from ads, or for calls to a number on your website, Google asks for a minimum call length. A call at least that long is counted as a conversion. Anything shorter is not. Google's guide to tracking calls from ads describes this, and its page on phone call conversion tracking notes that if calls are not being recorded, the minimum may be set too high.
That second point matters more than it sounds. The threshold is not only a reporting choice. If you bid with Target CPA or another automated strategy, the conversions you count are the signal the bidding system learns from. Set the line too low and the system learns to chase calls that were never leads: wrong numbers, switchboard queries, people asking about visiting hours. Set it too high and it sees almost no conversions, loses signal, and may bid conservatively.
So the threshold is a training label. It tells the bidder what a good outcome looks like. A label that is wrong for half your departments teaches the wrong lesson on half your budget.
Why a flat 60 seconds fails
Think about what a short call means in different departments.
- Emergency or urgent-care lines. A caller may need only to confirm the address and whether the department is open. The call is brief, the intent is high and the outcome (the patient arrives) is exactly what you want. A 60-second rule throws these away.
- Elective, high-value procedures. A person asking about a knee replacement, IVF or a cosmetic procedure usually wants to talk through options, cost and timing. A 20-second call is almost certainly a wrong number or an abandoned call. A 60-second rule may still let through a minute of hold music and a hang-up.
- Diagnostics and health check-ups. Calls are transactional: slot, price, preparation. They are short and are still bookings.
- Follow-up and existing-patient calls. These land on the same number as new-patient enquiries and are not leads at all, regardless of length.
Emergency numbers also raise a separate question. Whether to advertise an emergency line in paid search is a decision for your clinical and legal teams, not a bidding choice, so keep emergency calls out of any optimisation goal unless they have approved it.
Step 1: Pull the call-length distribution
You need the length of every call from ads for a period that is long enough to be representative. Four to eight weeks is usually enough for a single hospital department. In Google Ads, the call details report (available in the report editor, as Google's help page mentions) lists each call with its duration, the campaign and the ad group. Export it. If you run your own call tracking, export the same fields from there, plus whether the call was answered.
Then, for each department, sort calls by length and look at the shape:
- How many calls end in under ten seconds? These are almost always abandoned or misdials.
- Is there a visible dip or plateau between short calls and long ones? A dip often marks the boundary between "not a conversation" and "a conversation".
- Where do the median and the upper quartile sit?
You are looking for a natural break, not a percentile target. Here is a hypothetical example, using invented numbers to show the reading. Suppose an elective orthopaedic line shows a cluster of calls ending at 5 to 12 seconds (hang-ups), a thin middle, and a large cluster from 90 seconds upward. A threshold near 60 to 90 seconds would separate the two groups cleanly. Now suppose an urgent-care line shows a cluster at 15 to 40 seconds that your team confirms are genuine "are you open" calls. A 60-second rule would count none of them.
Step 2: Listen before you decide
Numbers tell you where the break is. They cannot tell you what the calls were. Ask the front desk, or your call-recording process if you have lawful consent in place, to sample calls on both sides of your candidate threshold. Check that calls above the line are mostly real enquiries, and that calls below it are mostly noise. If you record calls, make sure the recording and storage follow your consent notice and data protection obligations; our article on healthcare marketing compliance and the DPDP Act covers the principles.
Write down what you find in plain language. For example: "Calls over 45 seconds on the dental line are almost all booking or price enquiries. Calls under 15 seconds are almost all hang-ups." That sentence, not the percentile, is your justification for the setting.
Step 3: Decide whether one conversion action is enough
Google lets you create multiple call conversion actions. If departments differ sharply, create a separate action for each group and attach each to the relevant campaigns, rather than forcing one setting on everybody. A practical split for a multi-specialty hospital might be:
| Line type | Threshold approach | Notes |
|---|---|---|
| Elective surgery and procedures | Longer minimum, set at the break in the distribution | Short calls are rarely leads |
| Diagnostics and check-ups | Shorter minimum | Transactional bookings are brief |
| Emergency and urgent care | Usually not an optimisation goal | Confirm with clinical and legal teams |
| Existing-patient and admin numbers | Do not advertise these | Keep them off paid assets |
If a campaign mixes departments, it is a structural problem, not a threshold problem. Splitting campaigns by department is easier to measure and is the pattern described in the guide to multi-department campaign architecture, which sits close to this topic. See also the common errors collected in ten Google Ads mistakes healthcare practices make, which is a good place to check whether your conversion goals are differentiated.




