Strategy

Your Google Ads Call Tracking Is Counting Wrong - and Smart Bidding Is Acting On It

NPP
Net Profit Positive Advertising Strategy
| | -- min read
📞 All calls counted Junk + qualified
🧠 Smart Bidding sees Inflated conversions
📈 Result Overbidding on bad leads
💸 Account impact Net negative at "good" ROAS
The call tracking feedback loop - bad conversion data trains Smart Bidding to overbid, which spends more budget on the wrong audience.

Most Google Ads accounts with call tracking are running on corrupted data. Not corrupted in a dramatic way - corrupted in a mundane, default-settings way that almost every account manager misses on setup. The result is that Smart Bidding, Google's automated bid algorithm, is trained on a conversion signal that includes wrong numbers, voicemails, and calls that lasted 8 seconds. It treats all of them the same as a 12-minute conversation that closed a $4,000 job.

That is not a minor calibration error. That is the bid engine optimizing toward the wrong outcome at scale, every hour, on your budget.

How Google Counts a Call Conversion

When you set up call tracking in Google Ads, you have two main options: call assets and website call tracking - call assets show a number directly in the ad, while website call tracking replaces your site's number with a Google forwarding number to track calls from visitors who clicked an ad.

Both methods count a call as a conversion when the call meets a minimum duration threshold. The default threshold in Google Ads is 60 seconds - but as you will see when you look at how Smart Bidding actually uses that number, the call duration threshold functions as a signal definition, not just a filter. That sounds reasonable until you think through what a 60-second call actually means in practice.

⚠️
The Default Is Wrong for Most Businesses

60 seconds is Google's default call conversion threshold. For a home services company, a real qualified lead call averages 3 to 5 minutes. For a law firm, often longer. A 60-second threshold means every robocall that got answered, every wrong number where someone explained they wanted Pizza Hut, and every "what are your hours" inquiry counts as a conversion - and feeds that signal into Smart Bidding.

The threshold is just one problem. The deeper issue is that even if you get the threshold right, most accounts mix call conversion types in ways that destroy the signal quality further. More on that in a moment.

What Smart Bidding Does With Bad Data

Smart Bidding strategies - Target CPA, Target ROAS, Maximize Conversions - work by analyzing patterns in your conversion data and raising or lowering bids in real time to target the users most likely to convert. The keyword there is "convert." Google defines a conversion as whatever you told it to count.

If you told it to count every call over 60 seconds, that is what it optimizes for. It will find the audience, device, time of day, and keyword match type that generates the most 60-second calls. Some of those are genuinely qualified leads. Some are people who stayed on hold long enough to give up. Smart Bidding does not know the difference, because you never told it the difference.

"Smart Bidding is not smart about what matters to your business. It is smart about what you measure. If you measure poorly, it optimizes toward the wrong outcome with extraordinary efficiency."

Net Profit Positive

The reinforcing loop makes this worse over time. More budget flows toward the patterns that produce junk conversions. Those patterns get more data. The model gets more confident. Bid adjustments compound. By the time the problem surfaces in cost-per-acquisition numbers, the model has months of bad training baked in.

The Three Ways Call Tracking Breaks Your Conversion Data

These are the specific failure modes we see in nearly every account that relies on phone calls as a primary lead source:

💡
The Three Failure Modes
  • Duration threshold too low - The 60-second default (or worse, no minimum) counts calls that have zero business value. Qualified lead calls for most service businesses average 3 to 8 minutes.
  • Call extensions and website calls mixed into one conversion action - A call from a Google ad call button and a call from someone who clicked through to your website are fundamentally different intent signals. Mixing them obscures which campaign elements are actually driving qualified leads.
  • All call conversions set to "primary" in Smart Bidding - When every call type feeds the bid algorithm equally, the model cannot learn that a 2-minute call from a "near me" search keyword is 4x more likely to convert to revenue than a 2-minute call from a broad-match informational keyword.

The Numbers: What This Actually Costs

The impact is not abstract. Here is what inflated call conversions do to the math of a typical local service business running Google Ads.

0 Avg Junk Call Rate Calls under 90s in service accounts
0 CPA Inflation Factor True CPA vs. reported CPA
0 Avg Bid Overage Smart Bidding overbid on junk signals
0 Avg Reset Time For Smart Bidding to re-calibrate

That 2.4x CPA inflation number deserves a closer look. If your Google Ads dashboard shows a cost per lead of $65, and 38% of those "leads" are calls that lasted under 90 seconds and had no real buying intent, your actual cost per qualified lead is closer to $105. If your average job value is $800 and your close rate on qualified calls is 30%, you need to acquire qualified leads for under $240 to break even on ad spend alone - before COGS, overhead, and Google's platform cut.

At $65 reported CPA, that math looks fine. At $105 true CPA, it is tighter than most people think. At $105 CPA with Smart Bidding actively training toward the patterns that produce junk calls, the number trends upward.

True Cost Per Qualified Lead vs. Reported CPA

Impact of junk call rate on actual acquisition cost - home services industry baseline

$200 $150 $100 $50 10% 20% 30% 40% 50% Junk call rate
Reported CPA (stays flat)
True cost per qualified lead

The reported CPA line is flat because Google's dashboard only knows what you told it to count. The true CPQL climbs because the junk-to-qualified ratio gets worse as Smart Bidding learns to optimize toward junk signals. Both lines look calm from the outside. Only one of them reflects reality.

How to Fix Your Call Tracking Setup

This is a configuration problem, not a platform problem. Google gives you everything you need to measure calls properly. Most accounts just never configure it correctly after initial setup. Here is the specific sequence to correct it.

Step 1 - Set the Right Duration Threshold

Go to your Google Ads conversion actions and find your call conversion. The duration threshold should not be 60 seconds. The right number depends on your business, but the way to find it is straightforward: pull a sample of 20 to 30 calls that you know resulted in booked jobs or qualified sales conversations. Average their duration. Set your threshold at 70 to 80% of that average.

For most home services businesses, this lands between 2 and 4 minutes. For professional services and B2B, often 4 to 6 minutes. For e-commerce customer support, it varies widely and call conversions are usually the wrong metric anyway.

📊
Benchmarks by Industry

Based on accounts we have audited: HVAC and plumbing qualified calls average 3 min 40 sec. Legal intake calls average 5 min 20 sec. Dental practices average 2 min 55 sec. Roofing and remodeling average 4 min 10 sec. Use these as starting points, not absolutes - your specific market and offer structure will shift the number.

Step 2 - Split Call Conversion Actions by Source

Create separate conversion actions for: calls from call assets (formerly call extensions) in search ads, and calls from website call tracking. Do not combine them into one action.

Call asset calls happen before someone hits your site. They clicked your ad's phone number directly from the search results page. Website call tracking calls happen after they visited your site - they saw your content, evaluated your offer, and then called. Those are fundamentally different intent signals and they perform differently by campaign type, keyword match type, and audience segment.

Mixing them obscures which parts of your funnel are driving quality calls versus informational calls. You cannot optimize what you cannot see separately.

Step 3 - Set Bidding Priority Correctly

In your conversion action settings, Google asks whether each conversion action is "primary" (used in Smart Bidding) or "secondary" (tracked but not used for bidding). Most accounts set everything to primary. The correct setup is more specific.

Set your qualified-duration call conversion actions to primary. If you have the capability to review calls and tag them by outcome (booked, not interested, wrong number, etc.), set the raw duration-based action to secondary once you have enough call outcome data to train Smart Bidding on actual qualified calls. If you do not have call recording review in place, duration threshold is your proxy - use it as primary, but set it correctly per Step 1.

Real Account Example
Account Audit Finding

Scenario

A regional HVAC company spending $18,000/month on Google Ads, running Target CPA bidding at a $90 target. Dashboard was reporting 200 call conversions per month at an average CPA of $88. Account manager considered the account healthy - close to target, consistent volume.

Audit found: call threshold set to default 60 seconds, call extensions and website calls in a single conversion action marked primary. We pulled a 90-day call log sample. 41% of counted conversions were calls under 2 minutes. Of those, a manual review of 30 calls found 26 were wrong numbers, "what are your hours" inquiries, or calls that reached voicemail. Smart Bidding had 18 months of training data that included these calls as equivalent to a $3,500 AC installation job booked by phone.

After Fix (Day 60) True CPQL dropped from $149 to $91 - same budget, 39% more qualified leads

The Formula for True Cost Per Qualified Lead

Once your call tracking is clean, this is the number to watch. Not reported CPA - true CPQL. It accounts for the junk call rate, which will not be zero even with a good duration threshold, and gives you an honest cost per outcome that matters.

True Cost Per Qualified Lead (CPQL)

CPQL = Total Ad Spend / (Total Call Conversions x Qualified Call Rate)

Where:
  Qualified Call Rate = (Calls that resulted in quote/booking) / Total counted call conversions

Example:
  Monthly ad spend:         $8,000
  Counted call conversions:   180
  Qualified calls (reviewed):  72
  Qualified call rate:        72 / 180 = 40%

  CPQL = $8,000 / (180 x 0.40)
       = $8,000 / 72
       = $111.11

  Google's reported CPA:    $44.44
  Actual CPQL:             $111.11

  The gap is 2.5x. At a $300 average job value and 35% close rate,
  break-even CPQL = $300 x 0.35 = $105. This account is underwater.

The qualified call rate is the piece most accounts skip because it requires reviewing calls, not just counting them. You do not need to review every call forever. Review a statistically meaningful sample - 50 to 100 calls per campaign over 30 days - establish your qualified call rate per campaign, and then apply that rate as a correction factor to your reported conversions going forward. Update the sample quarterly or when you see significant performance shifts.

This Week's Action

Pull your Google Ads call conversion report for the last 90 days. Filter for calls between 60 and 120 seconds. That is your highest-risk junk conversion window. If those calls represent more than 20% of your total call conversions, your Smart Bidding target is calibrated toward the wrong outcome. Fix the threshold first, then re-evaluate your CPA target after 30 days of clean data.

What Good Call Tracking Looks Like

For reference, here is the setup that produces reliable conversion data for service-business Google Ads accounts:

Setting Default (Most Accounts) Correct Setup
Duration threshold 60 seconds 70-80% of average qualified call length
Conversion action structure One action for all calls Separate: call asset calls + website calls
Bidding priority All actions set to primary Primary = qualified-duration only; secondary = raw volume
Call review process None - counting by duration only Quarterly sample review to validate threshold
CPQL tracking Reported CPA only True CPQL formula applied monthly
Smart Bidding target Set against reported CPA Recalculated from true CPQL after clean data period

The last row matters more than people expect. After you fix call tracking, you will likely need to adjust your Smart Bidding targets upward - sometimes significantly. The reported CPA that felt comfortable was artificially low because it was counting junk. Your new, accurate CPA will be higher. That is not the account getting worse. That is you finally seeing what the account actually costs, so you can price your jobs and set your bids accordingly.

The goal is not a low reported CPA. The goal is a CPQL that is reliably below the revenue-per-customer number that keeps your business net positive. Those are different targets, and confusing them is how ad accounts spend six figures a year and still fail to generate the profit they should. If you want to run that math for your specific numbers, the calculator below does it in about a minute.

Free Calculator

Find your actual break-even CPQL

Enter your job value, close rate, margin, and current ad spend. Get your exact qualified lead cost target - and see how far off your current setup might be.

Calculate Mine →
NPP
Net Profit Positive Advertising Strategy Team

We build ad campaigns around what lands in your account, not what looks good on a dashboard. Every piece of content on this blog is grounded in data from real accounts across home services, professional services, and local B2B.