Does Cold Calling Still Work? What Four Datasets Show

Four datasets, four different answers, and one number two of them agree on.

You have been told cold calling is dead and you have been told it is the only thing that works. Both claims come with statistics, and if you have spent an hour trying to reconcile them you already know the problem: the numbers do not agree, and almost nobody says where theirs came from. So does cold calling still work? The honest answer is that it depends entirely on which number you pick up, and most of the numbers in circulation are not measuring the same thing.

We went and read the primary sources instead of the listicles. Four of them publish real methodology. Here is what they say, where they disagree, and what a founder with no SDR team should actually do about it.

The statistic problem

Search the question and the top results will hand you figures like "69% of buyers accepted a call from a new salesperson" and "companies that do not believe in cold calling have 42% lower growth." Neither of those carries a named study on the pages that publish them. The second one traces to a careers-content site. This is the normal state of cold calling content: a number gets quoted, then quoted from the quote, and the original either does not exist or dates to a decade when people answered desk phones.

So we set a rule for this piece. A number appears below only if we opened the publisher's own page this session and it discloses what was counted. That cut the field to four.

Four datasets, read honestly

Three of the four are first-party data from companies that sell into this market. That does not make them worthless. It makes them samples of one operation each, and they should be read that way.

Why the connect rates disagree by 2.5x

Gong says the average rep connects on 5.4% of dials. Cognism says 13.3%. That is a two-and-a-half-fold spread on what looks like the same metric, and the explanation is not that one of them is lying. It is that "connect" is a definition, not a fact.

Belkins states its definition outright: a connect includes calls that ended in "not interested," reached the wrong person, or were stopped by a gatekeeper. In its data those three categories are 31%, 13.5% and 13.4% of outcomes. Strip them out and the same dataset tells a much darker story. Cognism counts answered calls on its own dialer, with a 2-minute average SDR call duration, which tells you a large share of those answers went nowhere. Gong compares top-quartile reps against average ones, which is a performance spread, not a market rate.

The practical consequence: a vendor quoting you a connect rate is telling you nothing until you ask what counted as a connect. Ask. The answer is usually revealing.

The one number two datasets agree on

Strip out the definitions and look at the end of the funnel, where there is less room to argue.

Belkins reports roughly 370 dials per booked meeting across its 2025 campaign data. Gong's illustrative monthly figure is 800 dials producing about 2 meetings for an average rep, which is 400 dials per meeting. An agency's managed campaigns and a vendor's platform aggregate, measured differently in different years, land within 8% of each other.

Treat 370 to 400 dials per booked meeting as the planning number for a competent but not exceptional caller. Gong's top-quartile rep gets there in about 45 dials, which is the real argument for skill: not that the channel is good, but that the spread inside it is enormous.

Two corroborations worth noting. Belkins' conversation-to-meeting rate of 4.6% matches Gong's average-rep set rate of 4.6% exactly, from unrelated datasets. And Bridge Group, the only independent survey in the set, found that phone-centric teams average 56 dials and 4.6 quality conversations a day against 28 dials and 3.4 conversations for email-centric teams. Phone-centric teams get about 35% more real conversations per day. They work harder for them.

Does cold calling still work better outside the US?

This is the finding that surprised us, and it is the one most relevant to anyone selling from outside North America.

Belkins' 2025 campaign data breaks out by geography. The US shows a 9.0% live rate and a 0.55% positive-outcome rate per dial. The UK shows 14.7% live and 2.84% positive. Non-US markets overall ran 12–16% live rates, with positive-outcome rates between 1.6x and nearly 7x the US baseline.

Before anyone builds a strategy on that, read Belkins' own caveat: it says vertical mix explains a significant share of the gap, because its US records concentrate in healthcare and retail, both high-volume and low-conversion, while its UK and European records span a broader mix including manufacturing and software. So this is not clean proof that a Turkish or Nigerian prospect answers the phone more readily than an American one. It is one agency's book, confounded by industry.

What it does establish is that the "cold calling is dead" consensus is a US consensus, built on US saturation and US gatekeeping, and that importing it wholesale into a market where your buyer still answers an unknown mobile number is a mistake. The same pattern shows up in outbound sales in MENA and in emerging markets more broadly: the channel rankings are regional, and the playbook you downloaded was written for somewhere else.

What this means if you are the only person selling

Run the arithmetic on yourself before you run it on a hiring plan.

Twelve dials a day, every working day, buys you roughly one meeting every seven weeks. Four meetings a month from the phone alone needs about 1,500 dials, or 75 a day, which is not a side task. It is the job.

That is the honest shape of it, and it points at one conclusion rather than two. Cold calling produces the highest-quality conversation available to you and consumes the one resource you cannot buy more of. So do not use the phone to find people. Use it on people who have already moved.

Belkins' data shows 19% of call outcomes are warm signals, a requested follow-up or call-back, against 4.6% booked on the spot. Gong's practical advice is the same from the other direction: call prospects with a current trigger, a fundraise in the last three months or a new hire in the relevant function, not names off a list. Cognism's own task mix is 57% calls, 27% LinkedIn, 15% email, and its meetings booked more than four days out showed up only 60% of the time until a pre-meeting sequence pushed that to 80–82%.

Put together, the sequence that respects all four datasets looks like this: automated channels do the finding, you do the calling, and automation does the confirming. Email and WhatsApp generate replies and opens cheaply across a few hundred contacts. You call only the subset that responded, where you are dialling a warm name rather than a cold one. Neither dataset publishes a connect rate for that narrower list, so treat the lift as directional, not a number to plan against. Then a reminder sequence protects the meeting you just earned. The 14-touch multichannel sequence and the pipeline math both assume this division of labour.

Where BOSRAI fits, and what it does not do

BOSRAI does not dial. It runs ICP identification, lead sourcing, personalized outreach across email, WhatsApp and LinkedIn, automated follow-up and a CRM, with a human approving messages before they send. Pricing runs Free, Starter at $79.99, Growth at $199, Scale at $499 and Pro at $999, with a discount for annual billing.

So on the specific question this article asks, the platform is on the wrong side of the line. The highest-converting conversation in outbound is a voice conversation, and that one is yours to have. What BOSRAI is for is the 95% of the work that produces the short list worth calling, and the follow-up that keeps the booked meeting from quietly evaporating. If someone sells you an AI SDR on the promise that you will never speak to a prospect, the four datasets above are the argument against it. We have no published case studies to put on the other side of that, and we are not going to invent any.

Sources

• Bridge Group, 2025 SDR Models & Metrics — 351 B2B companies; 112 median daily activities (44 phone, 41 email, 19 LinkedIn), 4.1 quality conversations a day, phone-centric 56 dials / 4.6 conversations vs email-centric 28 / 3.4, 3.0-month ramp, 60% of reps at quota.
• Cognism, State of Outbound 2026 — 449,933 SDR cold calls in 2025, 13.3% answered, 2-minute average SDR call, task mix 57/27/15 calls/LinkedIn/email, show rates 60% rising to 80–82% with a pre-meeting sequence. The report calls itself directional, not diagnostic, and its own totals disagree in places.
• Belkins, B2B cold calling benchmarks — 175,000+ dials from its dialer plus five client campaign exports, 2025. 9.9% per-dial and 24.5% per-prospect connect, 58% connect-to-conversation, 4.6% conversation-to-meeting, ~370 dials per meeting, outcome mix 31% hard rejection / 19% warm signal / 13.5% wrong number / 13.4% gatekeeper, US 9.0% live vs UK 14.7%, non-US positive rates 1.6–7x the US baseline with a stated vertical-mix caveat.
• Gong Labs, The hidden power of cold calling — published July 2024, "300M+ cold calls," 5.4% average and 13.3% top-quartile connect rates, 4.6% vs 16.7% conversation-to-meeting, 800 dials to ~2 meetings for an average rep. Promotional, with no rep count or collection dates.
• Close, Does cold calling still work in 2026 — read as an example of the sourcing problem described above, not as a data source.