Dialing technique · 2026

Cold call connect rate benchmarks: what the published data actually says

Last updated September 3, 2026

Search "cold call connect rate" and you get a dozen pages quoting between 2% and 17% as "the 2026 average", mostly without naming a study. Nearly all of them are recycling two vendor datasets and one fifteen-year-old university study, and they mix numbers that do not share a denominator. Below: every figure worth quoting, who paid for it, how old it is, what the funnel looks like when you multiply the rates honestly, and the sample size you need before your own number is worth arguing about.

First, four words that are not synonyms

Benchmarks disagree mostly because vendors count differently and rarely publish the definition. Before you compare yourself to anything, pin these down:

• Dial. One outbound attempt to one number. If you double dial a number, that is two dials. If you use a parallel dialer, every line you opened is a dial, including the ones you dropped.
• Answer. Anything picked up: a human, a voicemail greeting, a switchboard, a fax tone. Carriers bill from here, which is why answer rate matters for your Twilio invoice and almost nothing else.
• Connect. Should mean a human said hello. In practice some tools count any answered call, some count only calls over a duration threshold, some count only the target contact and treat gatekeepers as no-connect. Each choice moves the number by points.
• Conversation. A connect that went past hello. Gong measures set rate against conversations, not dials, which is why its 4.6% looks small next to a "2% dial-to-meeting rate" that is actually larger.
• Contact rate. Share of your list reached across all attempts, measured per lead. Always higher than connect rate, and not comparable to it. Quoting one as the other is the single most common error on benchmark pages.

Every benchmark worth quoting, and who paid for it

All figures checked September 2026 against the publishers' own pages. Note what is missing from that list: there is no independent, academic, cross-vendor study of B2B cold calling connect rates from the last decade. Every current number is produced by a company selling something to cold callers, on a sample of its own customers. That does not make them wrong — Gong and Salesfinity landing within a point of each other from completely different platforms is genuinely reassuring — but it does mean no one should quote them as if they were census data.

One non-vendor number is worth carrying around: the Pew Research Center found that 80% of US adults do not generally answer their cellphone when an unknown number calls, with only 19% saying they do (survey fielded July 13–19, 2020; published December 14, 2020). That is the ceiling every cold calling benchmark is pushing against, and it is why a 6% connect rate is not incompetence.

The funnel, multiplied out honestly

Connect rate on its own decides nothing. Meetings come from connect rate × set rate, and the two move independently: better data lifts the first, a better opener lifts the second. Here is what each combination costs you in dials.

The last two rows are the interesting ones. A rep with a great list and an average pitch and a rep with an average list and a great pitch land 200 and 154 dials from a meeting respectively — both roughly twice as efficient as the published average, neither anywhere near top quartile. Getting to 45 dials per meeting requires both, which is why the top quartile is a quartile and not a technique. If you want the volume side of this, see how to make 100 cold calls a day — but note that page quotes the widely repeated 25–35 dials per meeting figure, and the datasets above are an order of magnitude less optimistic. Believe the datasets.

What a connect-rate change is actually worth

Hold the goal constant — 10 booked meetings a month at a 6% set rate — and vary only the connect rate. Minutes are billed the way carriers bill them (a voicemail pickup costs a full minute even if you hang up in ten seconds), at 40% of dials reaching a greeting and three-minute conversations. Telephony is priced at wholesale Twilio, ~$0.014/min plus one ~$1.15/month number. Hours assume 70 dials an hour on a power dialer.

Read the two end rows against each other. Moving from a 4% connect rate to 13.3% for the same 10 meetings saves 42 hours of dialing a month and $16 of telephony. The phone bill was never the expensive part; the rep was. This is also why per-minute pricing arguments are mostly noise for B2B outbound — see how many minutes cold calling actually uses and the cheapest way to make cold calls.

The other reason the dials-per-day column matters: at 198 dials a day from one number you are well into the territory where carrier analytics start scoring you. Read how many calls a day before your number is flagged before you solve a low connect rate by dialing harder — a flagged number lowers connect rate, so the fix compounds the problem.

How many dials before your own number means anything

This is the part no benchmark page covers, and it is the one that matters. A connect rate is a proportion estimated from a sample, so it carries an error bar. At an observed 7% connect rate, here is the 95% confidence interval at each sample size — that is, the range your true rate is plausibly in.

After 100 dials your 7% is really somewhere between 2.0% and 12.0%. That is the whole published range of the industry, inside one morning of calling. After 1,000 dials it is 5.4% to 8.6% and you can start making decisions. Comparing two things — two lists, two scripts, two hours of the day — is much harder still, because you are estimating a difference between two noisy numbers. As a rule of thumb, detecting a one-point difference around a 7% base needs roughly 9,700 dials per side. Detecting a five-point difference needs about 390 per side. Test big changes, not small ones.

What to log so the number is trustworthy

A connect rate is only as good as the disposition list behind it. Four rules:

1. Log every attempt, not every conversation. The most common way teams inflate connect rate is by only recording calls worth writing a note about. Your dialer should write a row for every dial automatically. In DialSheet that is the "All attempts" view on /calls, separate from the "Conversations" view used for coaching.
2. Separate bad data from no answer. Disconnected lines, wrong numbers and switchboards that will never transfer are list problems, not calling problems. Give them their own dispositions (Bad number, Wrong number, Gatekeeper) and report a connect rate on reachable numbers as well as on raw dials. A 17% dead-number rate — which is what Baylor found in 2011 — moves your headline connect rate by more than any script change will.
3. Decide once whether a gatekeeper is a connect. Either answer is defensible. What is not defensible is changing it halfway through a quarter and reporting an improvement.
4. Timestamp everything in the prospect's time zone. Otherwise your hour-of-day analysis measures your own working day, not theirs. More on that in the best time to cold call.

Then export and check the arithmetic yourself. DialSheet's call export is a CSV with timestamps, dispositions, durations and notes; the team leaderboard shows calls, connected and connect % per rep. Pivot the CSV by disposition, confirm the categories sum to total dials, and only then compare yourself to anything on this page.

Questions people actually ask