Phone Call Conversion Length Thresholds: Which Call Durations Predict a Booked Job — and What Counting Wrong Does to Your ROAS

The Problem Nobody Audits

If you're running Google Ads for a local service business — HVAC, plumbing, roofing, dental, legal — your campaign is almost certainly counting phone calls as conversions. That's the right instinct. Calls are often the highest-intent action a local prospect takes.

The problem is the default setup.

Out of the box, Google's call conversion tracking fires on any call that meets a minimum duration — and that default is often set to 60 seconds or less, sometimes as low as 30 seconds. A caller who reaches voicemail, listens to your greeting, and hangs up can meet that threshold. So can a wrong number that the recipient politely explains. Neither of those is a booked job.

When every call counts as a conversion, you're feeding corrupted signals into Smart Bidding. The algorithm believes it's hitting targets it isn't. Your reported ROAS looks healthier than reality. And the actual jobs you booked are buried inside a pool of noise.

Why Duration Is a Useful Proxy (and Its Limits)

Call duration isn't a perfect proxy for booking intent — but it's the most actionable signal available inside Google Ads without third-party call intelligence software.

The underlying logic is sound: a caller who has booked an appointment, received a quote, or been scheduled for an estimate has to exchange enough information — name, address, problem description, availability — that the call will almost always run longer than one that dead-ends.

Google itself has noted in its help documentation that businesses can improve conversion quality by adjusting the minimum call duration to better reflect meaningful interactions, though it stops short of prescribing a universal number — because the right threshold varies by business type.

Here's a rough framework by category:

| Business Type | Likely Booking Call Duration | Suggested Threshold Range | |---|---|---| | Emergency services (plumber, locksmith) | 2–4 min | 90–120 sec | | Scheduled appointments (dental, legal consult) | 3–6 min | 120–180 sec | | Complex estimates (roofing, remodeling) | 4–8 min | 150–240 sec | | Simple bookings (salon, oil change) | 1.5–3 min | 75–120 sec |

These are illustrative ranges, not published benchmarks. Your actual threshold should be derived from your own call data (more on that below).

The Audit: How to Find Your Real Threshold

You don't need call recording software to start this audit, though it helps. Here's a four-step process:

Step 1 — Pull your call log. Export call detail reports from Google Ads (Conversions → Call Details) for the last 60–90 days. You'll see call duration for every tracked call.

Step 2 — Cross-reference against your CRM or booking system. For every call that resulted in a booked job, note the call duration. If you don't have a CRM, your front desk or technician dispatch log works. You're looking for the minimum duration that reliably appears in your booked-job calls.

Step 3 — Map the distribution. Group calls into buckets: under 30 sec, 30–60 sec, 60–90 sec, 90–120 sec, 120–180 sec, 180+ sec. Count how many booked jobs fall into each bucket. In most local service businesses, you'll find a sharp drop-off in bookings below the 90–120 second range — though your data will tell you your own inflection point.

Step 4 — Set or update your threshold. In Google Ads, go to Tools → Conversions → select your call conversion action → Edit → adjust the minimum call duration. Set it to the inflection point your data revealed, not the Google default.

Modeling the CPA Inflation: A Worked Example

Here's a labeled illustrative model to show what corrupt call tracking does to reported numbers.

Scenario (illustrative):

• Monthly ad spend: $3,000
• Reported call conversions (all durations): 60
• Reported cost-per-conversion: $50
• After audit, calls that lasted 120+ seconds (booking-correlated): 22
• True cost-per-booked-call: $136

That's a 172% CPA inflation versus what the platform dashboard shows.

Now layer in ROAS. If the average job value is $400 (illustrative), reported ROAS is 8:1 on paper. Actual ROAS against real booked jobs is closer to 2.9:1 — still potentially profitable, but far below what the account appears to show, and far below what Smart Bidding thinks it's achieving.

Why this matters for Smart Bidding specifically: Target CPA and Target ROAS bidding strategies train on conversion events. If the algorithm is rewarding ad clicks that produce 30-second calls at the same weight as clicks that produce 4-minute booking calls, it will actively optimize toward traffic that looks like your noisy conversions — not your real ones. You're paying to teach the algorithm bad habits.

For a related look at how traffic quality shapes conversion rates before a call even happens, see our article Google vs Meta Ads: Traffic Quality & Conversion Rates.

Secondary Conversion Actions: Don't Throw Away the Data

When you tighten your primary conversion threshold, your reported conversion volume will drop — sometimes significantly. That's correct and healthy. But don't delete the short-call data entirely.

Instead, demote shorter calls to a secondary conversion action (set 'Include in Conversions' to No). This keeps the data visible for analysis without letting it feed Smart Bidding targets.

Short calls are still useful signals:

• A spike in sub-60-second calls after a new ad launch may signal a headline or offer mismatch — people clicking expecting something your ad implied but your team can't deliver
• High short-call volume from a specific keyword cluster is a negative keyword audit trigger
• Seasonal short-call spikes can correlate with wrong-number traffic from directory listings — worth cross-checking, especially if you've been running campaigns through seasonal transitions (see Ad Spend Seasonality Lag: What It Costs Local Businesses for context on how seasonal signal drift compounds)

The goal is a clean primary signal and a preserved diagnostic layer — not just trimming numbers to make dashboards look tighter.

Landing Page and Offer Alignment Matters Too

Fixing your call threshold solves the measurement problem. It doesn't automatically fix the upstream traffic quality problem.

If a meaningful share of your calls are short because prospects are confused about what they're calling about, the fix is pre-call — on the page where they clicked. A landing page that clearly states the service area, the type of jobs you handle, and what the caller should expect will filter unqualified traffic before they dial.

This is part of why the choice between offer pages and service pages for paid ads matters more than most businesses realize. If you haven't read Offer Page vs. Service Page for Ads: Which Wins?, that's a useful companion read to this audit — better page-to-call alignment means fewer wasted calls to filter in the first place.

The Clean Tracking Flywheel

Here's the compounding payoff of doing this right:

1. Accurate threshold set → primary conversions reflect booked jobs 2. Smart Bidding trains on real signal → algorithm finds more of the traffic that converts to actual revenue 3. ROAS reporting reflects reality → you make better budget decisions (scale what works, cut what doesn't) 4. Short calls stay visible as secondary → you catch quality issues early instead of months later

This isn't a one-time fix. Call behavior shifts with seasonality, new campaigns, and changes to your service mix. A quarterly call audit — pull the log, cross-reference bookings, sanity-check your threshold — is a lightweight habit that protects the integrity of everything downstream.

---

If you want a second set of eyes on your current call conversion setup — including whether your threshold, bidding strategy, and landing pages are aligned — book a call with the Nika Spark team. We'll show you exactly where your reported numbers diverge from real booked revenue, and what to do about it.