Call conversions: count every call or only ones over 60 seconds?

I keep running into the same question when setting up call conversions for local plumbers or dental clinics. Google counts a call as a conversion once it passes a duration you pick, and most people leave it at the 60 second default. The trouble is that plenty of 20 second calls are someone asking whether you handle emergencies in their area, which is a real lead. And a 90 second call can just as easily be a wrong number or a spam bot that stayed on the line.

Still, my view is that a high threshold keeps Smart Bidding focused on people who actually want to talk to you, while a low one rewards quick hang-ups and inflates the count. So I'd keep the threshold high for bidding and track the shorter calls as a separate action for reporting only. How do you compare call logs to booked jobs when they don't line up?

Agreed on keeping the bidding threshold high, but I'd set the number from your own data instead of leaving the default. Pull a month of call logs with duration and CRM outcome, group the calls into buckets (under 20 seconds, 20 to 60, 60 to 120, over 120), and see what share of each bucket turned into a booked job. For a plumber, the short bucket might be the emergency inquiries that actually book. If that's where your bookings sit, the 60 second default is quietly cutting your best leads out of bidding.

The bucket test is the right move. Before you touch the threshold, though, check whether the short calls are coming from a different source. Emergency searches on mobile often produce 20 second calls from people who are about to book anyway, so a low threshold can pull in exactly the leads you want while also pulling in misdials from the same keywords.

Agreed, and before touching the threshold I'd split the short calls by keyword and hour. Inside one keyword, a 15 second emergency call and a 15 second misdial look identical in the logs, but the timing usually isn't. For a plumber, real emergencies tend to cluster after hours and on weekends. Misdials from broad "plumber near me" terms show up during normal working hours, often someone who tapped the wrong number on a bus.

If the night-time calls under 60 seconds book at a decent rate, that's a case for a separate conversion action just for those calls. A global threshold change would sweep in the broad-match noise too. I'd run that cut first, because it tells you whether the short bucket is one population or two.

Splitting by hour is smart, but I'd look at the search terms before the clock. A broad match on "plumber near me" can pull in misdials at any hour, so your night calls might just be noise from a different query. Rule of thumb: if a short-call bucket doesn't book at a rate at least a few times that of a random misdial (roughly, the rate of your sub-10-second calls), don't give it its own conversion action.

Agreed on the sub-10-second baseline. A separate short-call action doesn't have to touch bidding, either. Set it as a secondary conversion and Google reports it without feeding it to Smart Bidding, so your 60-second primary keeps driving bids while you see how the short bucket actually books. The call history report gives you caller area code, duration and call status for each call, which is enough to pull the sub-10-second baseline and the night-time buckets from one export.

The weak spot is matching calls to booked jobs, because area code and timestamp don't tell you who booked. If the front desk asks for a name or phone number at intake and you match that against the CRM, the booking rate for each bucket is real. Without that, you're comparing your misdial rate against a guess.