If your conversion is a phone call rather than a checkout, almost every tracking guide online is written for someone else. There are four ways to track calls from Google Ads — call extensions, dynamic number insertion on your website, static click-to-call, and a third-party platform — and they measure different things.
The most common mistake is counting a tap on a phone link as a call. It is not one. It is an intention.
Read any article about conversion tracking and you will find yourself in an ecommerce store. Purchase events, cart values, return on ad spend. That is not much use if you are a dentist, a carpet cleaner, a plumber or a physiotherapist, where the money arrives because someone picked up a phone.
The failure modes are genuinely different, and they are worse — because a phone call leaves no trace in your website analytics at all unless you deliberately make it leave one.
The four methods, and what each one really tells you
| Method | What it measures | Keyword-level attribution | Main weakness |
|---|---|---|---|
| Call extension / call asset | Calls placed directly from the ad | Yes | Misses everyone who clicks through to the site first |
| Calls from website (dynamic number) | Calls to a swapped number on your site | Yes | Needs the swap script to load and work on mobile |
| Static click-to-call tap | Taps on a tel: link | Partial | Counts intent, not connection — inflates conversions |
| Third-party call platform | Connected calls, duration, recording, source | Yes | Monthly cost; another system to maintain |
Notice the third row. It is the one most small businesses end up with, usually by accident, because tracking a tap is the easiest thing to set up in Google Tag Manager. It is also the one that lies to you the most.
Why counting taps quietly destroys your bidding
A tap on a tel: link means someone was interested enough to open their dialler. It does not mean they pressed the green button. It does not mean you answered. It does not mean the call lasted longer than four seconds before they realised you were closed.
If you feed taps into Google Ads as conversions, Smart Bidding optimises toward people who tap. Those are not the same people as customers who book. The gap between the two is where your budget goes.
Conversion: tel_link_tap Fires on: click of a phone link Counts: opened dialler Reality: may never connect
Conversion: connected_call Fires on: answered call Counts: duration >= 60s Reality: a real enquiry
The single highest-leverage change most local advertisers can make is applying a minimum call duration before a call counts as a conversion. Sixty seconds is a sensible starting point for most service businesses: long enough to exclude wrong numbers and immediate hang-ups, short enough to include a quick “can you come Tuesday?”
What I actually set up, and what it changed
I have built this twice, for two very different businesses, and it is the part of my work with the clearest before-and-after.
Smile Line Dental — Islamabad
Before: Google Ads was running with no view of which searches produced appointments. Enquiries arrived by phone and by form, and neither was attributed to anything.
After: GTM tracking on both enquiry forms and phone calls, so specific searches could be tied to actual appointments for the first time. In the clinic’s words, they “stopped guessing.”
What I am not claiming: I do not have a published cost-per-enquiry figure for this account, so I am not going to invent one. What changed was visibility — which is the prerequisite for everything else, not the outcome itself.
The second was Mr Expert Carpet Cleaning in Slough, where the site was rebuilt from scratch with tracking designed in rather than bolted on, including a server-side GTM container. That one taught me the lesson in the next section.
The failure nobody warns you about: the form that emails
Most local service sites use a contact form that composes an email and sends it. If that send fails — a bad SMTP configuration, a spam filter, a full inbox — the lead does not go to a queue for retry. It ceases to exist. There is no record that anyone ever filled the form in.
You cannot track what was never saved. So before any conversion tracking work on a local service site, I check one thing: does the form write the lead to a database before it tries to email anyone? If the answer is no, tracking is the second problem.
This is the order that matters. Write the lead to storage, then attempt the email, the CRM push, the WhatsApp alert and the conversion event. Every one of those can fail without losing the enquiry. Reverse the order and a single failure loses a customer silently.
A sane setup, in order
- Make the lead durable firstEnsure form submissions are stored somewhere that is not an inbox. Everything downstream is optional; this is not.
- Turn on call assets with a forwarding numberThis captures the people who ring straight from the ad without ever visiting your site — usually a bigger share than owners expect.
- Add website call tracking with number swappingSo calls made after landing on the site are attributed too. Test it on a real phone, because the swap depends on a script running.
- Set a minimum call durationSixty seconds is a reasonable default. Adjust once you have listened to a sample of real calls.
- Import the conversion and let it settleDo not judge the change for at least two weeks. Bidding algorithms need time, and a fortnight of local-service call data is a small sample.
Where this advice stops
Call recording is a separate decision from call tracking, with separate legal obligations that vary by country and sometimes by region. Knowing that a call happened, how long it lasted and which keyword produced it is enough to optimise campaigns. Recording its contents is a bigger step — take advice before you do it.
And my sample here is two businesses, in two countries, in two industries. The mechanics generalise; the numbers do not. I have not published a cost-per-call figure in this article because I do not have one I can stand behind for a named account with a stated date range — and a made-up benchmark would be worth less to you than this sentence.
Do you know which searches produce your phone calls?
Most local businesses running Google Ads cannot answer that. Send me your site and I will tell you what is being tracked and what is not.
Get a free tracking audit →