Click to call is a phone number or button on a web page, ad or listing that starts a phone call when someone taps it, usually built with a tel: link.
It tends to help when your buyers already prefer to phone (urgent home services, insurance, loans, high-ticket B2B), and it can hurt when it pulls people away from a form you can track and follow up on.
Our database holds no clean test that adds only a call button and measures booked calls, so treat any single "calls lift leads by X%" claim with caution and test it on your own traffic.
What the database does show is the pattern around it: making the contact path visible on mobile, cutting the ask down to a phone number, and offering a short conversation instead of a download all won in real tests.
This page covers how to build and place a call button, then how to count calls as conversions so the test result means something.
The click-to-call button: when it helps and how to build it
A call button is a call to action like any other.
The question is not "should we show our number" but "which contact path converts more visitors into revenue for this audience, on this device."
What our tests show
The closest test we have is from a drain-repair company.
Digital NRG tested three mobile contact-page layouts for Trent's Drains; the winner moved the form up and put a "Call us on [number] or use the form below" line right before it.
The case study names the winner for mobile form submissions but gives no uplift figure, and the call line was tested together with the layout change.
That result matters because the phone line did not replace the form.
It sat next to it, so urgent visitors could call and everyone else could still submit.
A Brazilian bank went further toward the phone.
Mercantil Bank, with Croct, replaced a multi-field form with a popup asking only for a phone number, plus copy explaining the loan simulation.
It reported +5.5% form submissions, +15.8% qualified lead rate and -23.6% CAC.
The field cut and the new copy were tested together, and the vendor published no sample size.
In B2B, a "talk to us" offer can beat content.
For AvidXchange, Method Savvy found that a PPC landing page inviting prospects to a five-minute introductory conversation beat a white paper download on landing page conversion rate (no figure given).
Speed after the contact also counts. EDP, an energy supplier, passed solar simulator contacts to sales right away instead of waiting for users to finish the flow, and reported a call response rate above 20%.
That is mostly a process change, but it is a reminder that a phone lead goes cold fast.
Placement: keep the contact path in reach
Two placement tests carry over well to call buttons, even though they tested other buttons.
Booking.com added a sticky CTA bar to a long mobile partner sign-up page and got +36% clicks to start registration.
Insighten added a persistent "Get a Quote" button to Westfund's floating navigation and reported +10.82% quotes started, +10.24% quotes generated and +28.84% new member joins (desktop only).
The shared lesson: if visitors have to scroll back up to find the way to contact you, some of them won't.
A sticky call bar on mobile follows the same logic, but it should be tested, not assumed.
When a call button is the wrong choice
- Nobody answers. A call that rings out after hours or goes to a full voicemail box is worse than a form. Show the button only during staffed hours, or label it with the hours.
- Your team can't handle the volume. Calls need a person on the line right away; forms can wait in a queue.
- You need details before you can quote. If sales has to collect ten facts anyway, a short lead form may produce better-prepared leads.
- Desktop traffic. On a laptop a tel: link may open a calling app or do nothing, so a plain visible number often works better there.
- Your audience prefers text. Some buyers would rather chat or message. Our live chat tests show that extra contact widgets can lose as well as win.
How to build it
A click-to-call link uses the tel: scheme defined in RFC 3966.
Write the number in full international format so it works for every visitor, for example <a href="tel:+15551234567">Call (555) 123-4567</a>.
- Make it big enough to tap. WCAG 2.2 sets a minimum target size of 24 by 24 CSS pixels at Level AA; a primary call button should be well above that.
- Say what happens. "Call now: free quote in 5 minutes" sets an expectation that "Call" alone does not. See our CTA copy tests for how wording shifts clicks.
- Show the number, not only an icon. People like to see where they're calling, and a visible number also serves desktop visitors.
- Watch auto-detected numbers. Safari on iOS detects phone numbers by default and turns them into links. Those taps happen outside your own tel: links, so your click tracking may miss them. Mark up every number on purpose.
Call tracking: counting calls as conversions
Most call button tests go wrong at measurement.
A form submission shows up in your testing tool; a phone call happens on another network, often minutes later.
If you can't connect the call to the variant the caller saw, you can't call a winner.
Taps are not calls
The easy metric is clicks on the tel: link.
Google's own docs are clear about its limit: for clicks on a phone number on a mobile site, "we can only track these clicks, not the phone calls themselves" (Google Ads help).
People tap by accident, hang up before it connects, or call the wrong office.
Our database has a warning here. Digital NRG's above-the-fold contact CTA for Enhance Insurance raised CTA clicks by +133.38%, but that counted clicks, not completed forms.
A big lift in a proxy metric can shrink a lot by the time it reaches revenue.
The ways to track a call
| Method | What it counts | Good for | Weak spot |
|---|---|---|---|
| Click event on the tel: link | Taps on the button | Fast, free, works in any A/B testing tool | No idea if a call connected or how long it lasted |
| Google Ads call conversions | Calls from ads, calls to a Google forwarding number shown on your site, or mobile number clicks | Paid search traffic | Only covers visitors who arrived from Google Ads |
| Dynamic number insertion (DNI) | Real calls, tied to a visitor session | A/B tests: each variant's calls can be matched to the session | Needs a big enough pool of tracking numbers |
| Offline import from CRM | Calls that became qualified leads or sales | Judging lead quality, not just volume | Slow; needs a clean ID that links call to visit |
Google Ads lets you count a call as a conversion when it lasts longer than a minimum length you choose, which filters out wrong numbers and hang-ups (Google Ads help).
The same page notes that tracking calls from ads or from your website number relies on Google swapping your number for a forwarding number.
How dynamic number insertion works
DNI swaps the number on your page for a tracking number from a pool.
Each new visitor session gets its own number, and a call to that number is credited to that session's source and pages (CallScaler docs).
When the session expires, the number goes back into the pool.
The catch is pool size. If more visitors are on the site at once than you have numbers, the tool falls back to a default number and those calls lose their session data.
For a test, that means some calls can't be assigned to a variant, so size the pool to your peak concurrent visitors.
How to A/B test a call button
- Pick a primary metric past the tap. Qualified calls (connected and above your minimum length) plus form leads is a fair "total leads" metric when you compare a call button against a form-only page.
- Count both channels. A call button that adds 30 calls but removes 30 form fills did nothing. Measure calls and forms together, per variant.
- Check lead quality later. Import outcomes from your CRM. Mercantil Bank's result is useful because it reported qualified lead rate and CAC, not just submissions.
- Plan for low volume. Calls are usually rarer than clicks, so work out the sample size before you start. A test on taps will finish sooner but answers a weaker question.
- Segment by device and hours. A call button can win on mobile during business hours and lose at night or on desktop.
For more on the form side of the same decision, see form optimization and our lead-gen A/B test results.
You can browse every lead-gen test or search the full test database for contact and CTA changes in your industry.