The Call Duration Threshold That Tells Smart Bidding What a Lead Is Worth

The duration threshold in Google Ads call tracking is the most misunderstood setting in the platform. Most account managers treat it as a spam filter - set it high enough to exclude obvious junk and move on. That framing is wrong. The threshold is a signal definition. When you set it to 60 seconds, you are telling Smart Bidding: "Any call that lasts 60 seconds is a qualified lead." Smart Bidding believes you. It builds a model around what those 60-second calls have in common - their device, their location, their search query, their time of day - and it bids higher for users who match that profile. If 40% of your 60-second calls are people waiting on hold before hanging up, Smart Bidding is spending real money to find more of them.

This is not a fringe scenario. It is what the default setup produces in most service-business Google Ads accounts. Fixing it requires understanding what the threshold actually does - not just raising the number.

Why Duration Is a Proxy for Intent

You cannot measure intent directly in Google Ads. There is no signal that says "this person is ready to book a job." What you can measure is behavior, and call duration is the behavioral signal most correlated with buying intent in phone-dependent businesses. A call that ends in 8 seconds is almost never a qualified conversation - it is a misdial, a hang-up, or someone who heard the voicemail greeting and quit. A call that runs 5 minutes almost always involves a real exchange: the caller described their problem, asked questions, heard pricing, and either booked or declined.

The question is where the threshold between "junk" and "real" sits for your specific business. That number is not Google's default 60 seconds for most service businesses - and it is not the same across industries, offer types, or even campaign types within the same account.

"Think of a physical store. You would not count someone who opened the door, glanced around for 10 seconds, and left as a customer interaction. The duration threshold is where you draw that same line for phone calls."

Net Profit Positive

The 60-second default made sense when Google introduced it as a way to filter out the most obvious junk - calls under 10 seconds. But it was never calibrated for service businesses where a real intake conversation takes 3 to 6 minutes. Using it unchanged is not a neutral choice. It is an active decision to train Smart Bidding on a degraded signal.

How Smart Bidding Uses Duration Data

When Smart Bidding records a call conversion, it does not just log the fact that a call happened. It logs the full signal context present at the time of the click that led to the call: the keyword that triggered the ad, the match type, the device, the geographic location, the time of day, the ad copy variation the user saw, any audience segments the user belongs to, and the recency of their last interaction with your site.

It then uses that set of signals to build a probability model: given this combination of signals, how likely is a click to produce a conversion? Over time it adjusts bids upward for signal combinations that historically produced conversions and downward for combinations that did not. The model gets more confident the more conversion data it accumulates.

The result is campaigns that are optimized toward volume, not quality. They hit conversion targets because there are plenty of 60-second calls. They fail to hit revenue targets because the calls the model learned to find are not the ones that actually book jobs.

Why Campaigns Look Fine on the Dashboard

The insidious part of this problem is that it does not produce obvious red flags in standard reporting. Your CPA looks reasonable because conversions are counting. Your conversion volume looks healthy. Smart Bidding shows "eligible" status. Nothing in the Google Ads dashboard signals that 40% of your conversions are worthless.

The only way to see the problem is to look outside the dashboard - at your call recordings, your booked-job rate, your revenue per lead - and trace those numbers back to the ad data. Most accounts never do this systematically, so the problem persists for years.

Finding Your Business's Right Threshold

The correct threshold is not a number from a benchmarks table - it is derived from your own call data. The process takes about 30 minutes the first time and produces a starting point that is dramatically more accurate than any default.

Step 1: Sample 30 calls you know resulted in a booked job or qualified conversation
        (Pull from your CRM or call recording software - match to calls that converted)

Step 2: Record the duration of each call in seconds

Step 3: Sort the durations and find the median (the middle value of your sorted list)

Step 4: Multiply the median by 0.70

Result = your starting threshold

---

Example:
  30 qualified calls - median duration: 4 min 20 sec (260 seconds)
  260 x 0.70 = 182 seconds (3 min 2 sec)

  Start your threshold at 3 minutes.
  Review junk rate at 30 days and adjust upward if >20% of counted calls
  are still not qualifying to a real conversation.

The 0.70 multiplier exists because you want to capture calls that were genuinely qualified but ran slightly shorter than your average qualified call - early commitments, quick rebooking calls from existing customers, or situations where the customer had already done their research. Setting the threshold exactly at the median would exclude too many of those.

The Data Behind the Problem

The 3.1x CPA ratio deserves explanation. It means that accounts running a 60-second threshold report a cost per acquisition roughly 3 times lower than their actual cost per qualified lead once junk calls are removed from the denominator. An account that looks like it is running at $60 CPA is often actually running at $180 CPQL. Whether that math works depends entirely on what a qualified lead is worth to you - but most accounts have no idea they are looking at a 3x inflation factor until they do the call review.

How Fast Smart Bidding Re-Stabilizes

The implication for low-budget accounts is significant: if you spend under $1,000 per month on Google Ads, expect 7 to 8 weeks of uncertain performance after changing the threshold. That is not a bug - it is Smart Bidding rebuilding its model from a smaller data pool. The instability is real and can cause bids to spike or collapse during the learning period. Plan around it, do not panic through it.

What Happens When You Change It Mid-Campaign

This is the section most guides skip, and it is the one that causes the most preventable account damage.

The best practice for mid-campaign threshold changes is to make the change, set a calendar reminder for 35 days out, and only evaluate performance against the new signal at that point. Any performance assessment before the learning phase completes is measuring noise, not signal. If you are running Target CPA or Target ROAS, you may also need to adjust the target during the transition - because the conversion denominator just got smaller, and the algorithm will initially read that as underperformance.

When the Threshold Alone Is Not Enough

For most service businesses, a correctly set duration threshold is sufficient to get Smart Bidding training on the right signal. But duration is only one of two structural problems - mixing call conversion types from call assets and website tracking into a single action compounds the threshold problem, especially for businesses with very high call volume - typically 500 or more calls per month - where duration alone is still a noisy proxy.