Number strategy · 2026

How many Twilio numbers do I need for cold calling?

Last updated September 3, 2026

This is a division problem, and almost every article that answers it skips the division and gives you a number between two and ten. The count depends on exactly two inputs: how many dials your reps make in a day, and how many dials one number can carry before carriers start scoring it. We argued the second input at length in how many calls a day before your number is flagged as spam — published guidance runs from 30 to 250 a day and the honest answer depends on your answer rate and call length. This page takes that ceiling as given and does the buying decision: how many numbers, in whose name, in which area codes, and what happens if you overbuy.

The arithmetic, and the one input that matters

The formula is trivial. The divisor is not. Set it from how your calls behave, not from a blog post:

• ~100/day if your answer rate is above roughly 15%, calls run over a minute, your numbers are registered and A-attested, and you take some inbound.
• ~75/day for ordinary B2B cold calling: 5–15% answer rate, 30–60 second average, outbound only. This is the default used below.
• ~40/day if answer rates are under 5%, most calls die inside 20 seconds, or the list was bought.

The divisor swings the answer more than headcount does. One rep at 150 dials a day needs 4 numbers at a 40-dial cap, 3 numbers at a 60-dial cap, 2 numbers at a 75-dial cap and 2 numbers at a 100-dial cap. So the cheapest way to need fewer numbers is to raise your answer rate, not to buy more numbers — better lists and better openers reduce your number count as a side effect.

Numbers to buy, by team size and daily dials

At the 75-dial default. Each cell is the whole team's pool, with the monthly cost of that pool at ~$1.15/month per US local number (Twilio list price, checked September 2026).

The largest cell on the table — ten reps at 200 dials a day, 30 numbers — costs $34.50 a month. That is the entire point of running a dialer on your own Twilio account rather than renting seats: numbers are a commodity at wholesale, so you size the pool to the calling rather than to the invoice. Tools that "include a phone number" with a seat are including a ~$1.15/month item and charging you $30 to $170 for the privilege — see cold calling software with a phone number included.

Where the pool sits in the total bill

Five reps at 100 dials a day, with a quarter of dials hitting voicemail and 8% reaching a human for about two minutes, bill roughly 4,305 minutes a month across the team — carriers bill a whole minute for a voicemail pickup, which is the line most calculators forget (the full method is in how many minutes cold calling uses). On their own Twilio that is about $72 of usage a month, plus $58 of software on the $29-per-3-seat plan: $130 all in. Of that, the 10 numbers are $11.50 — about 16% of the Twilio bill and 9% of the total. The minutes dominate, the numbers do not. Do not economise here.

One number per rep, or a shared rotation pool?

Two structures, and most teams want the first until the arithmetic forces the second.

The rule that survives contact with a real team: one primary number per rep, plus pool numbers only where the arithmetic or the area codes demand them. A rep at 150 dials a day gets their own number and one more. A rep at 60 dials a day gets one and does not need a pool at all, however much rotation is sold as best practice.

Area codes: how local presence multiplies the count

Local presence — showing a caller ID in the prospect's area code — is the one legitimate reason to hold more numbers than raw volume requires, and it multiplies fast. Three territories with two numbers each is six numbers per rep, not two. At ~$1.15/month that is still under ten dollars a rep, but every one of those numbers needs warming and none of them build history quickly, because you are dividing the same dials across more caller IDs.

The discipline that keeps it sane: rank your lead list by area code, buy numbers for the top two or three only, and let everything else fall back to a single default number. A rep whose leads are 60% in one metro gets far more from two numbers in that metro than from twelve numbers spread across a state. The setup, the attestation implications and the line between local presence and illegal spoofing are covered in local presence dialing with Twilio and how to get a local number for cold calling.

How many you can warm up at once

This is the constraint that stops you buying the whole pool on day one. Warm-up requires deliberately low volume per number, stepping up over one to two weeks — and those dials come out of the same daily budget as your real calling. A rep who makes 150 dials a day and gives three new numbers 20 dials each has spent 60 dials on warm-up and has 90 left for numbers that are already trusted. Give six new numbers 20 dials each and the day is gone before you have made a single call from a number anyone answers.

So the practical ceiling is two to three numbers warming per rep at a time, which means a six-number local-presence pool takes about a month to bring fully online. Plan the pool a quarter ahead of the headcount, not the week you hire. The day-by-day schedule and the pre-flight registrations are in how to warm up a new phone number.

When to retire a number, and when not to

Numbers get replaced on evidence, not on a rota. The evidence worth acting on, in order:

1. Answer rate divergence. One number in the pool answering at half the rate of its siblings on comparable lists is the earliest reliable signal, and you will see it before any label does.
2. A confirmed label. A test call to handsets on all three major US carriers, or a lookup at the free caller registries, showing the number rendering as Spam Likely or Scam Likely.
3. Failed remediation. You have submitted a correction, waited, changed nothing about the calling behaviour, and the label came back.

Point three is the one that matters. Buying a fresh number after a flag, without changing the list quality or the per-number volume that caused it, buys a few weeks and burns an asset. Aged numbers with real history are worth more than clean ones; treat retirement as a loss, not a maintenance task. If you have a flagged number already, work through fixing a Twilio number flagged Spam Likely before you release it — releasing a number puts it back in the pool for someone else, and you cannot get it back.

The honest downside of a big pool

Since numbers cost ~$1.15/month, the temptation is to buy forty and rotate hard. Three reasons not to:

It is the spam pattern. Spreading a small volume of short, unanswered calls over many caller IDs to stay under per-number thresholds is exactly what high-volume spam operations do, and the analytics engines behind carrier labelling — Hiya for AT&T, TNS for Verizon, First Orion for T-Mobile — are built to detect it. A pool sized to your dials looks like a sales team. A pool five times that size looks like someone evading a threshold, because it is the same shape.

Thin history is its own risk. Numbers with no track record get mislabelled more often than numbers with one. Forty numbers each carrying a handful of calls a day never accumulate the history that makes a caller ID trusted; four numbers carrying a real load do.

Registration does not scale for free. Voice does not require A2P registration, but if you also text from those numbers, every one of them has to be attached to a registered 10DLC campaign with its own recurring fees, and a large pool of lightly-used sending numbers is precisely the pattern the 10DLC regime was created to stop. See the A2P 10DLC registration guide. On the voice side, keeping the pool small also keeps your STIR/SHAKEN story simple: full A attestation depends on calling from numbers you own on your own account, which is covered in STIR/SHAKEN attestation for cold callers.

Questions people actually ask