Cold Calling vs Cold Email in 2026: What 1.29 Million Emails and the Published Numbers Say

Cold calling and cold email get compared as if one of them is going to win, and then the answer arrives as a single percentage that settles nothing. The more useful framing is that the two channels fail in different ways and cost different amounts, so the right question is which failure you can afford. We should say our position before any numbers appear: Sales.co sells managed cold email, and cold calling is a paid add-on for us rather than the product. We have a commercial interest in you reading this a particular way. The email figures below are ours, taken from campaigns we ran. Every calling figure belongs to someone else, and we name who and link to it.

The numbers we can actually stand behind

Between October 2025 and August 2026 we sent 1,288,605 cold emails on behalf of B2B clients across dozens of industries. Those emails produced 34,269 replies. That is a raw reply rate of 2.66%, and it counts every message in every sequence, first sends and follow-ups alike. The full breakdown is in our cold email response rate report.

The 2.66% is the number that gets quoted, and it is close to useless for planning. A reply is any human response, and a large share of replies are people telling you to go away, forwarding you to someone who no longer works there, or setting an out-of-office. When we classify replies by sentiment, 0.70% of the emails we sent got a positive reply. Narrow that further to replies from someone who actually wanted to talk about the thing we were selling, and the rate is 0.24%. One interested reply per 421 emails sent.

Put the two figures side by side and the difference is the whole problem with channel comparisons. A campaign quoted at 2.66% and a campaign quoted at 0.24% can be the identical campaign. If you are budgeting from the first number you will over-forecast by a factor of about eleven. When a vendor quotes you a reply rate, ask which of the two they mean, and ask who classified the replies.

A useful way to hold the scale in your head: per 10,000 emails sent, 266 replies land in an inbox, and roughly 24 of those are from someone genuinely interested. On our own benchmark page we put average at about 2.4 interested replies per 1,000 emails, so a campaign clearing 3 per 1,000 is ahead of most senders. That is the range the whole channel operates in, and any pitch describing 10% interested replies at volume is describing something other than cold email.

Three findings from the wider dataset changed how we run campaigns, and they matter for the calling comparison too. First, 69% of replies come from the first message. We had assumed follow-ups carried more weight, and they do carry some, but the first email is where the outcome is mostly decided, which is why we cap sequences at three emails rather than eight. Second, C-level recipients respond positively about three times as often as directors and managers. That runs against the usual advice to aim below the executive layer to avoid the gatekeeper, and it is one of the few places our data was clearly at odds with what we expected. Third, an informal tone outperformed a formal one by 78% on positive reply rate. Those come from the larger sets behind our cold email statistics report, covering more than two million emails, 161 campaigns and 61,770 sentiment-classified replies.

The reason we can publish numbers at this resolution is unglamorous. Sending platforms log every send, every bounce and every reply automatically, and the reply text sits there waiting to be classified. Nobody has to remember to record anything.

What the calling numbers look like

We have not published cold calling connect rates, dial-to-meeting ratios, or per-meeting call costs, because we have not run the phone at anything like the volume we run email. So the calling figures here come from other people, and we would treat them with more caution than our own.

Belkins published a benchmark study drawn from 175,000+ dials logged through 2025, combining its dialer activity with five anonymized campaign exports. It reports a per-dial connect rate of 9.9%, a per-prospect connect rate of 24.5% across an average of three dial attempts per prospect, and a conversation-to-meeting rate of 4.6%, which works out at roughly 370 dials per meeting end to end. Belkins sells appointment setting, so this is a vendor publishing numbers about the service it sells.

Cognism, working with WHAM, analysed over 200,000 calls for its 2026 cold calling report. It puts the industry-average success rate at 2.7%, up from 2.3% the previous year, and states an average of 1.55 dial attempts to reach a prospect in 2026 against 2.9 in 2025. It also reports its own success rate at 11.3%, more than four times the benchmark it publishes. Cognism sells B2B contact data including verified mobile numbers, and better connect rates are the argument for buying that data, so the report is marketing as well as research.

Notice that these two sources disagree about something basic. Belkins says three dials per prospect, Cognism says 1.55 attempts to reach one. Both are internally consistent and neither is obviously wrong; they are measuring different populations with different data quality and probably different definitions of an attempt. That spread is the honest state of public calling data.

The asymmetry between what we can say about email and what anyone can say about calling is not an accident of who bothered to publish. Email volume is logged by the machine that sends it. Dials are logged by whoever pays for the dialer, which means the people with good calling data are the calling vendors, and they publish the slices that support what they sell. There is no equivalent of a bounce log for a phone call. When someone quotes you a connect rate, it came from a company selling either dialers, data, or callers.

One more thing worth stating plainly. A cold email reply rate and a cold call connect rate are not the same kind of event and cannot be compared as percentages. A connect means a human said hello. A reply means a human read something and typed back. The reply is a much stronger signal of interest than the connect, which is why comparing 9.9% against 2.66% and concluding that calling is nearly four times better is a mistake we see in almost every article on this topic. Compare cost per interested conversation instead, and do it with your own numbers.

Dimension by dimension

Ramp time, and why the delays are different

Both channels take time to start working, and the reasons have almost nothing in common.

Email is held back by infrastructure. A new domain with no sending history is treated with suspicion by receiving mail servers, so you build reputation gradually: a handful of messages a day per mailbox, rising over a few weeks, spread across enough mailboxes that no single one carries too much volume. Skip the warming and you do not get an error message. You get delivery to spam, which looks exactly like nobody being interested, and you find out a month later when someone forwards you a screenshot of your email in their junk folder. Recovering a burned domain is slower than warming a fresh one, which is why most people quietly buy new domains instead. We keep clients on a seven-day path to live because the warming happens on infrastructure we prepare in advance rather than starting from scratch.

Calling has no warming problem. A caller with a list and a phone can dial in the first hour. The delay is a person: hiring, training, and the weeks it takes before someone can handle an objection about your product without reading from a page. Rent the capacity and you skip the hiring but not the learning, and you are paying full rate through it.

The LevelUp Leads pricing page is useful here because it publishes activity levels next to prices, which very few agencies do. Its tiers state 150+, 300+ and 700+ calls per business day, alongside 180+, 230+ and 460+ emails per business day, starting at $5,000/mo with a three-month minimum. This is the vendor's own published claim about its own service, not an independent measurement.

Two things fall out of those numbers. The first is a real dials-to-emails ratio from a company selling both: roughly 0.8 calls per email at the entry tier, rising to about 1.5 at the top. The second is stranger. A modern sending stack can push several thousand emails a day without effort, so 460 emails a day is not a machine limit. It is a limit imposed by list building, personalization and the per-mailbox sending caps that protect deliverability. The phone number scales up faster across their tiers than the email number does, which tells you the constraint on high-quality email volume is human work rather than capacity.

Against Belkins' figure of roughly 370 dials per meeting, a LevelUp entry tier at 150 dials a day is about two and a half business days of dialling per meeting, and the top tier is around half a day. We are mixing two vendors' numbers to get that, so treat it as an order of magnitude rather than a forecast. It is still a more honest way to size a calling programme than any cost-per-meeting figure you will be quoted on a sales call.

Compliance, in outline

None of this is legal advice, we are not lawyers, and the rules differ by jurisdiction and change. Get advice before you build a process on any of it. What follows is a description of which regimes apply, so you know what to ask about.

US calling sits under the Telephone Consumer Protection Act and the FCC rules made under it, which govern things like automated dialling and prerecorded messages, and which have been amended repeatedly. Individual states layer their own restrictions on top, including permitted calling hours, and those hours are set by the time zone where the person you are calling is, not where your caller is sitting. The National Do Not Call Registry covers residential subscribers, and business-to-business calls are treated differently from consumer calls. That distinction gets less clean than people expect once you are dialling mobile numbers, sole traders, and people who work from home, which describes a large share of a modern B2B list.

US email sits under CAN-SPAM, which sets requirements around accurate headers and subject lines, identifying the message, including a valid physical postal address, and honouring opt-out requests within a defined period. It applies to commercial messages generally rather than to bulk sending specifically.

In the EU and the UK, both channels sit under GDPR, with PECR adding rules specific to electronic marketing in the UK and the ePrivacy regime doing similar work across the EU. The part that needs advice is your lawful basis for processing someone's contact details and for contacting them at all, and whether the corporate-subscriber treatment applies to the people on your list. Member state implementations differ. Do not assume a process that is fine for a US list transfers.

The practical difference between the channels is where the obligations attach. Email compliance is largely about the content of the message and the mechanics of unsubscribing, both of which you can build once and apply to everything you send. Calling compliance is about who you dial, when, and from what, which has to be enforced per call and per list, and which changes when you cross a state line.

When to combine them

The version that works is not two parallel programmes. It is email finding the interest and the phone converting it.

Start with what our data says about timing. 69% of replies come from the first message, so a sequence that is going to produce a written signal usually produces it early. That gives you a list within a couple of weeks: people who replied positively, people who replied with a soft no, and people who never replied at all. Those three groups deserve different treatment, and only one of them is worth a caller's hour.

The positive replies that then went quiet are the best calling list you will ever have. Someone said they were interested, a meeting did not get booked, and the thread died. A call there is not cold. It costs the same dial and lands on a person who already knows your name. If you are only going to add the phone for one thing, add it for this.

Deal size decides most of the rest. If a closed deal is worth a few thousand dollars, the arithmetic will not carry a caller's hourly cost through the connect rate and the meeting rate. Somewhere above roughly $20,000 in average contract value the phone starts to pay, and well above that it becomes hard to justify not calling. That threshold is a judgement from what we see rather than a figure we have measured, and it moves with your gross margin and your close rate.

Then ask a question most people skip: does your buyer answer unknown numbers? Some roles still do, because the phone is how their industry works. Owners of contracting businesses, clinic managers, and plant supervisors answer. Engineering leads at software companies largely do not, and a caller can spend a full day producing voicemails. You can test this cheaply before you buy anything, by having one person dial thirty accounts for two afternoons and counting how many humans answer. That test costs a day and it is worth more than any benchmark in this article, including ours.

The last condition is internal. A live call creates an interested person on the phone right now, and if nobody on your side can take that handoff within the hour, you have spent the money and lost the moment. Email is more forgiving here, though not as forgiving as people think. We treat a reply older than a few hours as materially colder. Our appointment setting and outsourced SDR pages describe how the handoff works when someone else is doing the calling, and our comparison of cold calling companies covers the vendors who sell the phone as the main product rather than an add-on.

When we would and would not add calls

Since we sell the add-on, here is the arithmetic without the sales gloss. Our plans are $1,000/mo for Intent Takeover, $2,000/mo for TAM Takeover, and $2,500/mo for both. Cold calls are a $1,500/mo add-on on top of a plan, not a standalone product. So the $1,000/mo plan plus calls is $2,500/mo, and the $2,500/mo combined plan plus calls is $4,000/mo. You are paying 2.5x the entry price to add a second channel. All of it is on the pricing page, month to month, no setup fee.

We would argue for adding calls when several of these are true at once. Your average contract value is high enough that one additional meeting a month covers the $1,500 several times over. Your target list is named and finite, in the hundreds rather than the tens of thousands, so a caller can cover it properly instead of skimming. Your buyers demonstrably answer the phone, tested rather than assumed. And email is already producing positive replies that stall before a meeting gets booked, which is the specific failure a caller fixes.

We would argue against it in more cases than for it, honestly. If your contract values are in the low thousands, the add-on will not return its cost and you are better off putting the same money into a second Sales.co plan or better list data. If your product is self-serve and the buying process does not involve a scheduled conversation, a phone call interrupts a flow that was working. If your addressable market is very large, coverage beats depth, and the same $1,500 spent on email reaches a couple of orders of magnitude more accounts. If your buyers are in roles that screen every unknown number, you are buying voicemails at retail. And if there is nobody on your team who can pick up a warm handoff the same day, the calls create opportunities you will not catch.

For a lot of our clients we would not add it, and we say so on the call. The pattern we most want to avoid is a team buying calls to fix an email programme that is underperforming for a different reason: wrong list, weak offer, or a message that assumes the reader already knows what you do. Adding a channel does not fix any of those. It repeats them into a second medium at higher cost. Our cold email service page covers what the base plan includes, and the tools versus managed service breakdown is the comparison to read if the real question is whether to run any of this in-house. If what you want is the data rather than the delivery, our one-time verified email lists are $295 flat and you can dial or send from them yourself. And if you are comparing vendors, the field is laid out in our cold email agency comparison.

Frequently asked questions

Is cold calling better than cold email in 2026?

Neither wins in general, because they fail for different reasons and cost different amounts. Cold email is cheap per touch and can cover a whole addressable market, but most messages are never read and the honest interested-reply rate is small: across 1,288,605 emails we sent between October 2025 and August 2026 we saw a 2.66% raw reply rate and a 0.24% interested reply rate, one interested reply per 421 emails. Cold calling costs far more per attempt and cannot cover a large market, but a connected call is a conversation rather than a message in a queue. A large addressable market favours email; a short list of high-value named accounts is where the phone earns its cost.

What is the conversion rate of cold email compared with cold calling?

The two are not measured in the same unit, which is where most comparisons go wrong. For email we measure replies per email sent: 2.66% raw, 0.70% positive, 0.24% interested across our 1,288,605 emails. For calling the published figures are connect rates per dial, a different event. Belkins reports a 9.9% per-dial connect rate and a 4.6% conversation-to-meeting rate across 175,000+ dials logged in 2025, roughly 370 dials per meeting. Belkins sells appointment setting, so that is a vendor's number about its own service. We have not published calling figures of our own.

Which is cheaper per meeting, cold calling or cold email?

Email is far cheaper per touch and usually stays cheaper per meeting, though the gap narrows as the contact list gets smaller. Sending costs a fraction of a cent; the real expense is the list, the copy and the person watching deliverability. A dial costs a slice of a caller's paid hour and a caller makes a few hundred dials a day at most. Our published rates are $1,000/mo for Intent Takeover, $2,000/mo for TAM Takeover, $2,500/mo for both, with cold calls a $1,500/mo add-on on top of a plan, so $1,000/mo becomes $2,500/mo. That is worth paying when one extra meeting a month covers it several times over.

Should a B2B startup start with cold calling or cold email?

Start with email in almost every case, for a reason unrelated to which channel performs better. Email produces a written record of what you said and what people replied, at a volume large enough to learn from within weeks. If the message is wrong, a few thousand emails tell you cheaply. Calling teaches you more per conversation but far more slowly, and early on the thing being tested is the message rather than the channel. Add the phone once you know which segment replies and which pitch lands.

How long does each channel take to ramp?

Cold email is gated by domain and mailbox warming. New infrastructure has to build sending reputation before it carries volume, typically over a few weeks, and skipping it damages deliverability in a way that is slow to reverse. Our clients are live in 7 days because the warming happens on infrastructure we prepare in advance. Cold calling has no warming step, so the first dial can happen on day one, but it is gated by people: you hire and train callers or rent them, and either way it takes weeks before a caller sounds credible on your product.

Is cold calling legal for B2B in the US?

This is not legal advice, rules differ by jurisdiction, and you should check with a lawyer before building a process on any of it. In outline: US telemarketing calls sit under the Telephone Consumer Protection Act and related FCC rules, states add their own restrictions including permitted calling hours, and the National Do Not Call Registry applies to residential subscribers, with business-to-business calls treated differently from consumer calls. That difference is narrower than people assume once mobile numbers and sole traders are on the list. Cold email in the US is governed by CAN-SPAM, which sets requirements on headers, subject lines, identification, a physical address and opt-outs. In the EU and UK, GDPR and PECR apply, and your lawful basis for contacting someone is the part to get advice on.

Do cold calls work better after a cold email?

That sequencing is what most teams settle on and it is the version we would defend. Email is the cheap way to find out who is interested; the phone is the expensive way to move an interested person forward. In our data, 69% of replies come from the first message, so if a sequence is going to produce a written signal it usually does so early. Calling the people who opened a conversation and then went quiet is a much better use of a caller's hour than dialling a cold list, because the account has already shown it is in market.

Where we would leave it

Run the test rather than the comparison. Send to a segment you understand, count interested replies rather than replies, and have one person dial thirty of the same accounts for two afternoons and count how many humans answer. Two weeks of that is worth more than every benchmark on this page, ours included, because it uses your list and your offer. If you want the reply-rate distribution to check your own results against, it is in our response rate report, and the prices for having us run it are on the pricing page.