When we audit accounts that run phone call conversions, the first thing we ask is: what qualifies as a conversion? The most common answer is "any call over 60 seconds." Sometimes it's 30 seconds. Occasionally it's any call at all, regardless of duration.
None of those thresholds measure what the business actually cares about. A 90-second call from someone asking for directions isn’t a lead. A 45-second call from an existing customer confirming an appointment isn’t a lead. A two-minute call from a job applicant isn’t a lead. All of them count as conversions under the standard setup, and all of them are pulling your cost per conversion down to a number that looks good in a report but means almost nothing for the business.
We cover the broader conversion tracking problem in our post on why your Google Ads conversion tracking is probably wrong. Call tracking is where that problem shows up most often, because phone calls are harder to qualify than form fills and the default setup makes it easy to count everything.
What Google’s Default Call Tracking Actually Measures
Google Ads offers two types of call conversions: calls from ads (someone clicks the call button on your ad directly) and calls from your website (someone clicks an ad, lands on your page, and calls the number shown there). Both can be configured with a minimum call duration threshold.
The duration threshold is the only quality filter Google gives you out of the box. If a call lasts longer than your threshold, it counts. Google has no way to know if the caller was a prospective customer, an existing patient, a vendor, or someone who dialed the wrong number and stayed on hold for two minutes before hanging up.
According to Google’s call conversion documentation, the recommended minimum duration is 60 seconds, but that recommendation is about giving a call enough time to be intentional, not about qualifying the caller as a real lead. A 60-second minimum is a noise filter, not a lead filter.
What most accounts end up with is a conversion action that counts any non-trivial phone call as a lead. The algorithm is doing its job. The data is wrong.
If you’re spending $8,000 a month and reporting 120 conversions at $67 each, but 40 of those conversions are existing customers, vendors, and wrong numbers, your real CPA for new leads is probably closer to $100. Smart Bidding is optimizing toward $67 because that’s what the data says.
The Gap Between Calls and Qualified Leads
Here’s what that looks like in a real account. A home services company running plumbing ads was reporting 85 calls per month at a $58 CPA. When they dug into their call recordings (they had call tracking software with recording enabled), they found:
| Call Type | Volume | % of Total |
|---|---|---|
| New service inquiries (actual leads) | 41 | 48% |
| Existing customer callbacks | 19 | 22% |
| Job applications | 8 | 9% |
| Supplier and vendor calls | 6 | 7% |
| Wrong numbers / hung up | 11 | 13% |
| Total reported conversions | 85 | 100% |
Their actual new lead volume was 41, not 85. Their real CPA for new leads was $120, not $58. The account had been running for 14 months. Smart Bidding had been optimizing toward a $58 CPA the entire time, trained on a dataset where half the “conversions” were not leads.
When they reconfigured their conversion actions to import only confirmed leads from their CRM (via offline conversion imports), the algorithm recalibrated. CPA jumped to $115 in the short term because it was finally looking at real numbers. Over the following three months, it came back down to $88 as Smart Bidding learned what an actual lead looked like. $88 is higher than $58. It’s also accurate, and the decisions they made based on it were real.
The Three Ways to Fix It
Option 1: Integrate with a call tracking platform. Tools like CallRail or CallTrackingMetrics let you listen to calls, tag them by outcome (new lead, existing customer, wrong number, etc.), and import only the tagged leads back into Google Ads as conversions. This is the most accurate approach because a human reviewed the call. It’s also the most work. For businesses that run on phone leads and spend more than $5,000 a month on ads, it’s worth it.
Option 2: Use offline conversion imports from your CRM. If your team logs every call in a CRM and marks whether it was a new inquiry, you can export those logged leads and import them into Google Ads as offline conversions. Google matches them to the clicks that generated the calls and trains Smart Bidding on real lead data. Google’s offline conversion import documentation walks through the technical setup. It requires consistent CRM discipline, but it produces the cleanest signal.
Option 3: Raise your duration threshold significantly. This doesn’t fix the problem, but it reduces the noise. If you raise the threshold from 60 seconds to 3 minutes, you eliminate most wrong numbers and a lot of quick existing-customer calls. You’ll still count some non-leads, but you’ll count fewer of them. If options 1 and 2 aren’t feasible right now, this is a reasonable interim step. Just know you’re still working with an approximation.
Do not leave the default 60-second threshold in place and call those conversions "leads." They're phone calls. Close, but not the same thing.
What This Does to Your Bidding
The reason this matters beyond reporting accuracy is that Smart Bidding trains on your conversion data. Every call you count as a conversion teaches the algorithm something about who converts. If 50% of your “conversions” are existing customers calling for service, the algorithm learns that the people who search “schedule a plumbing appointment” (an existing customer query) are high-value. It bids more aggressively for those queries and less aggressively for “emergency plumber near me” (a new customer query), because the data says the first group converts at a higher rate.
This is backwards. You want to spend money acquiring new customers, not taking calls from people who already found you. But Smart Bidding doesn’t know the difference between a new lead and a repeat customer call unless you tell it. The conversion data is the only way you have to communicate that distinction to the algorithm.
Our Google Ads management always starts with a conversion audit before we touch campaigns. If the conversion data is wrong, every optimization decision downstream is wrong too. If you want us to look at what your call conversions are actually counting, request a free audit and we’ll pull the data.
