Google Ads call tracking: what counts as a phone lead?
Google Ads call tracking can measure calls from ads, calls to a forwarding number on your site, or only a phone-number click, depending on the method. Record the conversion action and evidence being checked; Google’s newer AI-qualified path has additional conditions. Our suggestion: confirm a usable business enquiry separately from the advertising count.
On this page9 sections
What does each call-tracking route actually measure?
Identify the route before interpreting its conversion total. A tap, a measured call and a business-qualified enquiry are different observations.
| Advertising evidence | What the business still needs to establish | |
|---|---|---|
| Calls from an ad | Google’s overview includes call-only ads and call or location assets; the calls-from-ads guide describes recording-based qualification or a configured duration method. | Record the actual conversion method and eligibility. A Google-classified lead is not proof of a booked or paid job. |
| Calls to a website number | The website-call method uses a dynamically generated Google forwarding number after an ad click and a minimum call length you set. | Check that the forwarding route was used and that the conversation met the business’s own acceptance rule. |
| A click on a mobile website number | The mobile-number-click method tracks the tap, not the actual call. | It does not prove connection, answer, duration or a useful request. Keep the action’s name explicit. |
| A qualified business outcome | Google’s overview also describes importing call outcomes tracked in another system. | Define accepted enquiry, appointment or sale using actual business records. Import setup is a separate implementation, not something inferred from a call count. |
The overview actually lists five methods. Its additional ad-interaction method estimates whether a meaningful call occurred after someone clicked an ad or asset, without a Google forwarding number. That estimate is not the same method as tagging a click on a mobile website number.
Our reading: do not combine those two click-related methods under a label such as “all phone clicks are calls”. Record the selected method beside the reported number. For interpreting cost per conversion, use the separate Google Ads cost-per-lead guide.
Does a call count by duration or by Google’s AI qualification?
Read the actual action and current eligibility; there is no single duration rule that describes every call-conversion method. Google’s checked pages describe more than one path.
The overview describes setting a minimum call length. The newer calls-from-ads guide also describes AI analysis of recordings. If you turn off call recording or a call cannot be recorded, it falls back to your configured duration.
The same guide says duration is unavailable if you turn off call reporting or, rarely, a Google forwarding number is unavailable. In those cases, conversions are classified from ad-interaction data.
The separate AI-qualified call-leads guide, checked 20 Oct 2026, limits recording availability to cases where both the dialling and receiving phone numbers are in the United States or Canada. That is a necessary geographic condition, not a promise that every account has recording available.
Google’s forwarding-number country table separately includes Poland, the United Kingdom and the United States. Inclusion in that table does not extend the US/Canada recording condition to every forwarding-number country.
For calls from assets, the setup guide requires an eligible phone-number country code and the relevant call-reporting setup. It begins at Goals → Summary → + Create conversion action → Conversions from phone calls. Verify the route and actual available choices rather than assuming a website tap action measures the same thing.
Keep the selected method with the number
Our suggestion: record whether the number represents a duration-qualified call, Google’s recording-based classification, an estimated ad interaction, a mobile-number click or an imported outcome. The business’s accepted-enquiry decision remains a separate field.
How do you test a website forwarding-number route?
On a mobile page with a tappable number, check both the number the visitor sees and the number the phone will dial, then inspect the corresponding reporting evidence. Seeing a replacement number alone is not a complete test.
Google’s website instructions require the relevant Google tag and phone snippet, website editing and an eligible business. They describe dynamically replacing the website’s business number with a Google forwarding number.
On mobile web pages with a clickable phone number, replace both the number that is displayed and the number that is called when clicked.
- Name the route and action. Our suggestion: record the website page, intended business number, conversion action, method and actual configured duration threshold. Do not assume a default: read the threshold of your own action.
- Check the website implementation. Follow Google’s website-call test instructions for a visit after an ad click, which the instructions say is charged. Check the visible replacement and clickable number’s dial target, including the mobile version.
- Record connection and the selected evidence. Keep the test time, whether the team received it and the relevant Google call/action evidence together. A test of the dial link alone cannot establish advertising attribution.
- Separate the business decision. Mark an artificial test as a test. For a real enquiry, have the responsible person record service fit and the next agreed action rather than infer quality from elapsed seconds.
Google says its forwarding numbers can change and should not be copied and promoted outside Google Ads (forwarding-number guidance). Our suggestion is to preserve the observed test evidence, not replace the business’s published contact number with a copied forwarding number.
If the wider website journey is also suspect, use the website-conversion troubleshooting guide. It covers form and tag investigation; this acceptance test concerns the phone route.
What does a completed call-tracking acceptance record look like?
Our suggested record below is fully invented, including the people, action name, settings, observations and dates. It shows a failed mobile link, a corrected technical route and a separate business classification; it is not a test performed in a real account.
The business offers planned repair assessments. Its invented website-call action uses a 75-second threshold; that is the business’s example setting, not a Google default or recommendation. The deliberately longer 125-second test avoids relying on the boundary between “at least” in the overview and “longer than” in the website guide.
| Recorded observation | Acceptance decision and owner | |
|---|---|---|
| Route and date | Website visit after a Search ad click, using the forwarding-number route; action “Site calls — assessment”; controlled test on 20 Oct 2026. | Mira, the advertising operator, records the action and chosen reporting columns before the test. |
| Initial mobile check | The visible number changed, but tapping it still dialled the original business number. | Technical check fails. Clara, the website owner, corrects the clickable dial target on 20 Oct. This does not say the original number cannot receive calls. |
| Retest | The visible number and the tap target both use the observed forwarding number; the office receives the 125-second test call. | The replacement/connection check passes. Record the time and evidence instead of assuming a visible replacement proves every step. |
| Google evidence | The operator finds one conversion for the selected website-call action and reporting dates in this invented example; the office separately records receiving the test. | Mira records the action and date basis. This does not claim a Google individual-call entry for the website route, or establish a sale from an aggregate count. |
| Business classification | The office marks the conversation TEST, not a genuine new customer enquiry. | Ben, the office owner, keeps it out of the accepted-enquiry total; the Google action count and business customer count answer different questions. |
| Handover | Clara retains the implementation change; Mira retains action/report evidence; Ben owns the business classification. | On 21 Oct, the team checks that the real-enquiry procedure uses the same fields. This is an invented business checkpoint, not a Google processing deadline. |
Our reading: a completed technical test can produce an advertising conversion while adding zero genuine new business enquiries. That is expected for the marked test, not evidence that Google is broken or that every counted call is unsuitable.
Why do the call report and conversion column disagree?
First compare the metric, action, date basis and reporting level. Two totals can differ without referring to the same set of calls.
For call or location assets, Google’s call-reporting instructions let an operator add aggregate columns through Columns → Call Details: Phone calls, Phone Impr. and PTR.
The page separately describes individual-call details, such as start/end time, missed or received status and call type, and links to instructions for seeing each call. These are distinct from the aggregate columns and are Google Ads data, not the product’s call-level data.
For call and location assets, the same page says Phone calls use the date the call occurred, while phone call conversions use the date of the ad click. Our suggestion: write both date bases on an export before comparing them; a call after the original click can appear in a different reporting period.
Google also says manually dialled calls are counted at campaign and ad-group level but are not available for keywords and ads. An ad-level sum therefore need not reproduce the campaign’s complete call total. Preserve the reporting level beside the number.
For a duration-based method, the call count and conversion count can additionally differ because of the configured threshold. For newer AI-qualified or estimated methods, keep that classification visible rather than describing every difference as a short call.
For actual qualified or closed outcomes, use the separate offline-conversion guide. Recording a useful conversation in the office and uploading an eligible outcome to Google are distinct tasks.
What can Variafy inspect about call tracking?
Variafy is in the free beta, with Google Ads support live and read-only. For an authorised connection it reads returned CALL assets, their phone/country and attachment/status information, conversion-action type, primary setting, counting and windows, plus available aggregate conversion totals, subject to coverage limits.
It does not read the configured call-duration threshold, individual calls, their duration, recordings or whether your team answered and qualified them. It does not install website call tracking or change Google Ads settings. A call-level Google report and the business’s conversation record still need a person to inspect them.
A suggested request is: “List the returned call assets, conversion actions of the account and this campaign’s conversion goals, their settings and available totals. State what is missing or incomplete. Do not treat a number-click action as proof of an answered business enquiry.”
What should you hand over after the test?
Our suggestion: send the failed or accepted step, its evidence and a named next owner. “Calls do not track” is too broad to tell the website owner or account operator what must change.
Copyable call-route handover
Page/ad route and conversion action: __. Measurement method and eligibility checked: __. Visible number / dial target: __. Test time, received status and evidence: __. Configured threshold if applicable: __. Google metric, column, date basis and level: __. Business classification: __. Failed step or remaining unknown: __. Next action, owner and date: __.
In the invented record, the immediate defect was the mobile dial target, not the team’s qualification rule. The retest accepted the technical route, while the test conversation remained outside the genuine lead total. Preserve both conclusions when assessing the next real enquiries.
How we checked
Checked 20 Oct 2026 against 7 different Google Ads Help documents: phone-call conversion overview, calls from ads and website phone calls.
Also checked: clicks on a mobile website number, call-reporting results, Google forwarding numbers and AI-qualified call leads. Their duration, estimated-interaction and recording-based descriptions are kept distinct.
The filled test, people, threshold, duration and dates are invented. They illustrate an acceptance record, not a measured account, a Google default or a guarantee of attribution, qualification or reporting timing.
Questions
Does a click on a phone number count as a call?
The mobile-website-number click method measures the click, not an actual phone call. Google separately describes an estimated meaningful-call method from ad or asset interactions without a forwarding number. Record which method produced the number; neither should be silently substituted for a confirmed business enquiry.
How long must a call last to count as a Google Ads conversion?
For duration-based tracking, inspect the minimum configured for the actual action. Google also describes AI-qualified recording analysis and ad-interaction fallback methods. This guide does not use a universal default; its 75-second setting and 125-second test are invented examples.
Why does my website show a forwarding number but calls are missing?
On a mobile page with a tappable number, Google’s website instructions require replacing both the visible number and the number a tap dials. Then inspect the chosen action, reporting dates and connection evidence; a visible replacement alone is not a complete attribution test.
Are Google forwarding numbers available in Poland?
Poland appears in the checked Google forwarding-number country table, as do the United Kingdom and United States. That does not make every call feature available: the checked AI-qualified guide limits call-recording availability to cases where both phone numbers are in the United States or Canada.
Why are phone calls and phone call conversions on different dates?
For call and location assets, Google’s call-reporting page says phone calls use the date they occurred, while phone call conversions use the date of the ad click. Compare the selected metric, dates and reporting level before assuming calls are missing.
Can Variafy show call duration or recordings?
No. It reads returned call assets, action type, primary setting, counting/windows and aggregate results. It does not fetch the call-duration threshold, individual call records, duration, audio or the business’s qualification decision, and it does not change Google Ads or install the website tracking.