Revenue Engineering

What Should You Pay Someone to Make Cold Calls?

Per hour, per call, per conversation or per meeting: what each payment model actually buys, which one produces the behaviour you want, and what nobody quotes.

Mauricio Esparza By ·Published ·2 min read
mithub.club
Short answer

Pay per hour and you buy time. Pay per call and you buy dials, which is the worst incentive available. Pay per qualified conversation or per held meeting and you buy the thing you actually want, but you must define qualified in writing first. The model matters less than the definition, and most disputes in this arrangement are definition disputes.

Four payment models, and each buys different behaviour.

Per hour

You buy time. Easy to administer and you carry all the performance risk.

Reasonable for a short test where the goal is to learn what the list does, not to buy an outcome. Poor as a standing arrangement, because nothing in it rewards quality.

Per call

You buy dials. This is the worst incentive available and it produces exactly what you paid for: high volume, short calls, poor listening and a list burned faster than you can replace it.

The only case where it makes sense is pure data collection, where the objective genuinely is coverage rather than conversation.

Per qualified conversation

Better, and only as good as your definition.

Without a written definition, "qualified" drifts toward whatever is easy to produce. With one, this model aligns well: they are paid for finding the right person and having a real conversation, which is the actual job.

Per meeting held

Closest to what you want, and it moves risk to them, which they price in. Expect a higher unit cost and better quality.

The word that matters is held. Booked and held are different numbers, and the gap between them is where this arrangement usually breaks down.

The three definitions to write before agreeing a rate

  1. What counts as qualified. Title, company criteria, expressed need, timing. Be specific enough that two people would agree on a given call.
  2. What counts as held. Showed up, stayed a minimum length, was the right person.
  3. Who owns a no show. Rescheduled once at their cost, then yours. Or some other rule. Just decide it.

Almost every dispute in this arrangement is one of these three. All three are cheap to settle in advance and expensive to argue about later.

What nobody quotes

The list. A caller working unverified data spends much of the day on dead numbers, and you pay for that time in every model. Verify before they start.

The script and the objection handling. If you do not provide them, they will improvise, and you will be represented by whatever they invent.

The handoff. What happens after a meeting is booked, and where the context is written. Without this, your closer starts cold and the meeting is worth less than you paid.

The comparison that decides it

Cost per held meeting with the right person, not cost per call or per hour.

Run that number against what it costs you internally, including the rep hours currently spent on research and dialling. Most companies have never calculated either side, which is why this decision usually gets made on the headline rate.

Frequently asked questions

Which model is best?

Per qualified conversation or per meeting held, with the definition written down. Per hour is fine for a short test where you want to learn rather than to buy outcomes, and per call is almost never right.

What has to be defined in writing?

What counts as qualified, what counts as a held meeting, and who carries a no show. Almost every dispute in this arrangement is one of those three, and all three are cheap to settle in advance.

What is not included in the quote?

The list. A caller working unverified data spends much of the time on dead numbers, and you pay for that time either way. Verify before they start.

Should we use AI instead?

For first contact at volume, it covers ground a person cannot and it never gets tired at call four hundred. For the conversation that needs trust or handles a real objection, a person still converts better. Most working setups use both.

Sources

  1. Building effective agents · Anthropic (accessed 2026-09-22)
  2. A2P 10DLC registration · Twilio Docs (accessed 2026-09-22)
  3. New Lightcast Report: AI Skills Command 28% Salary Premium as Demand Shifts Beyond Tech Industry · Lightcast (accessed 2026-09-22)
Revenue EngineeringOutboundHiring
Mauricio Esparza
Mauricio EsparzaGTM Systems Lead · Revenue Engineer · Founder of MitHub. Designs and runs revenue systems for multi-location businesses: AI voice campaigns, enrichment, CRM automation and attribution. Founded MitHub to teach the method in the open.

Part of Revenue Engineering on MitHub.

Keep going