Cold Calling Fundamentals: When It Still Works and How the Mechanics Run

By Laurel · Updated Oct 8, 2026

Cold calling is phoning a business contact who hasn't asked to hear from you, to start a sales conversation. It still works in B2B when the account list is small and well matched, the number reaches the right person, and the caller has a real reason to ring. It fails on stale data, generic openers and unscreened lists.

Most dials don't connect, and most connects don't become meetings, so the mechanics matter. The exact words live in the cold call scripts guide and the gatekeeper and objections guide. The full motion is in the outbound sales playbook.

Key Facts

• Gong, a conversation-intelligence vendor, analyzed more than 300 million cold calls on its platform. The average rep connected on 5.4% of dials, top-quartile reps on 13.3% (published July 2024).
• Belkins, an outbound agency, logged 175,000+ dials in 2025: 9.9% connected per dial, 24.5% of prospects connected across about three attempts, and roughly 370 dials booked one meeting.
• Gong reports a 3.44% email reply rate with calling in the mix against 1.81%. That's its own data, not a controlled test.
• The UK ICO says a number takes 28 days to become active on the TPS or CTPS, so an old screening check can go stale.

When Cold Calling Still Works

Gong's verdict: cold calling is dead for the average rep, who dials when it's convenient, uses canned openers and quits after a couple of objections. Top reps, it found, can get close to a meeting an hour.

What the Connect-Rate Data Says

Both datasets come from companies that sell something close to outbound, so read each as one company's view of its own activity. The connect rates aren't in conflict: Gong divides calls answered by a prospect by dials across its customers' reps, while Belkins counts any live human, gatekeepers included, across its own SDRs and clients.

The shape matters more than the number. A minority of dials connect, a minority of conversations book (4.6% for Gong's average rep and in Belkins' data, 16.7% for Gong's top quartile), and callers vary widely. At 800 dials a month, Gong's arithmetic gives roughly 2 meetings for the average rep and 18 for a top-quartile rep. Swap in your own rates once you have them: outbound pipeline math turns a meetings target into dials, and BDR/SDR metrics and quota math covers per-rep targets.

Where Calling Earns Its Place

Calling pays where a conversation is worth the effort: senior or hard-to-email buyers, larger deals, short account lists, a reason to call now. Belkins' calls produced 33.6% of its 2025 appointments against 50% for email, from far fewer contacts. In its client campaign data, every positive outcome from C-level contacts was a follow-up request, not a live booking, so have the follow-up ready before you dial. Calling is a poor fit for huge low-value lists, numbers you can't trust, or markets where consent comes first.

What Drives Connect Rates

Three levers show up in the data, and the evidence for timing is the weakest.

• Number quality. Gong says top-quartile reps prioritize direct dials and mark bad numbers so nobody redials a dead line. Of Belkins' calls that didn't book, 13.5% were the wrong person or number and 13.4% stopped at a gatekeeper.
• Persistence. Belkins' 9.9% per dial becomes 24.5% per prospect over about three attempts.
• Targeting. Gong's top reps chased prospects likely to buy now, using triggers like funding or hiring. See signal-based outbound.

Getting the Right Numbers

Direct Dials, Mobiles, and Switchboards

A switchboard reaches a receptionist, a direct dial reaches a desk line, and a mobile reaches a person almost anywhere. How B2B contact data works explains where phone numbers come from and why they decay. Tag every number with its type and source so you can compare connect rates by type, build the account layer first as the B2B prospect list article shows, and mark a dead number the first time it fails.

Mobile Numbers and Their Limits

A mobile is the most direct line and the most sensitive. Tied to a named person it's personal data. And phone rules don't exempt mobiles: the UK's TPS, Singapore's Do Not Call rules and US consent rules all reach them, as the rules section below explains. If you can't say where a mobile came from and whether it's been screened, don't dial it.

When to Call: What the Timing Data Does and Doesn't Show

Best-time-to-call claims sound authoritative, and they disagree. ZoomInfo, a data vendor, analyzed more than 1.4 million outbound calls to new business accounts and reported Tuesday as the day with the highest connect rate, with Tuesday and Wednesday making up 44% of its sample's demos. In Belkins' 2025 data, Wednesday had the best positive-outcome rate at 0.95% of dials, Tuesday 0.61% and Friday the lowest at 0.51%. By hour, Belkins called 3 to 4 PM local time the most consistent daily peak (1.05% to 1.22%) and the usual 9 to 11 AM block only middling (0.65% to 0.83%).

Three weaknesses to keep in mind:

1. Different outcomes. ZoomInfo counted connects and demos. Belkins counted a booked meeting, a follow-up request or a scheduled call-back.
2. Different customers. Belkins says its hour data reflects specific client verticals, calls the windows a starting hypothesis, and notes most dial volume lands at 9 to 11 AM, so a popular window means more competition.
3. Geography. Neither source reports Southeast Asia, where Singapore, Malaysia and the Philippines share a time zone, Vietnam and western Indonesia sit an hour behind, and Indonesia spans three zones.

Treat the windows as a hypothesis in the prospect's local time. Log dial time and outcome, split similar accounts between two windows for a few weeks, and compare conversations, not dials.

Preparing in Two Minutes

Prep tier-one accounts before each call and batch the rest the day before.

1. Check the record (20 seconds). Number type, opt-out flag, screening status, last touch.
2. Confirm who and when (20 seconds). Name, title and their local time.
3. Find a reason (40 seconds). One fact tied to a problem their role owns, like a new office or a hiring push.
4. Write the question (20 seconds). The one you'll ask after the opener.
5. Set the ask and fallback (20 seconds). Two meeting times, and what you'll send if they say "email me."

If step three comes up empty, that's a finding about the list. Tighten the ideal customer profile for outbound.

The First 30 Seconds

An opener has three jobs: say who you are, say why you called, and make it easy to answer. In the UK that's also a rule: the ICO requires live marketing callers to display their number and say who is calling. A permission-based opener names the cold call and asks for a short slice of time. A pattern-interrupt opener skips the greeting ritual and leads with something specific. Both examples are illustrations with placeholder names.

Permission-based: "Hi Ms Tan, this is Alex from [Company]. I'm calling out of the blue, so I'll say that upfront. We help [type of firm] with [problem]. Can I take thirty seconds to say why I called, and you tell me whether it's worth continuing?"

Pattern-interrupt: "Ms Tan, Alex at [Company], and you don't know me. I saw you're hiring three account managers. Who owns onboarding for new customers right now?"

Gong tested four openers on its 300M+ calls. "Did I catch you at a bad time?" scored a 2.15% success rate, "How's your day going?" 7.6%, a tailored permission opener 11.18%, and "Have you heard our name tossed around?" 11.24%. Gong published this alongside a book, and reps who pick stronger openers may differ in other ways, so treat the ranking as a direction. The "name tossed around" opener only works if it's true, since implying contacts you don't have is deception. What the strong openers share matters more than the label: a stated reason and a specific hook.

Voicemail

Most dials end in voicemail or nothing, so a voicemail has one job: be recognized later. A workable structure is name and company, one specific reason, a callback number, and a pointer to the email you've sent or will send.

One legal distinction applies. A message a rep speaks is treated differently from a recorded or synthetic message played by a system, which is often regulated more tightly. In the US, the Telephone Consumer Protection Act restricts artificial or prerecorded voice calls to cellular numbers without prior express consent, and the FCC ruled in February 2024 that this covers current AI technologies that generate human voices. Drop-message tools and AI voice agents need a legal review before they touch mobiles.

Combining Calls With Email

Email scales; a call can turn silence into a conversation. In Gong's data the email reply rate was 3.44% with calling in the mix against 1.81% without, even when calls didn't connect. But that's an association, since teams that call may differ in other ways.

Belkins recommends calling as an escalation after unanswered email steps, and its follow-up data found over 53% of email-sourced meetings came at step three or later. An illustration to adapt, not a benchmark: email on day 1 (how to write a cold email), call and voicemail on day 2, second email on day 4, a call in a different window on day 6, then a last call and a close-the-loop email on day 11. The cold email cadences guide has fuller sequences. One rule overrides any pattern: if someone asks you to stop, stop on every channel.

Logging Outcomes

A call that isn't logged can't be counted, coached or screened. Use one outcome list so a connect rate means the same thing next month. Belkins separates a dial (any attempt), a connect (a live human, gatekeepers included), a conversation (the person engaged beyond the greeting) and a meeting (a booked appointment). Capture the time, the number and its type, the outcome, the next step with a date, and an opt-out flag.

The categories point to different fixes. Of Belkins' calls that didn't book, 31% were hard rejections and 19% were warm signals, a follow-up request or scheduled call-back. Wrong numbers and gatekeepers, covered above, are data and process problems, not pitch problems.

If you record calls, check consent rules first: the FTC notes that US state laws vary on whether the other party must consent, and other countries differ too. Whichever tool places the call, the outcome belongs on the contact record. Rework's Sales app keeps accounts, contacts and activity against each deal, where a disposition and a do-not-call flag can sit beside the pipeline.

Calling Rules: What the Regulators Publish

This is general information, not legal advice. Rules differ by country, by whether you call a company or an individual, and by what you sell, so confirm with local counsel before a program scales.

United Kingdom: TPS and CTPS

For most types of call, the ICO says you can phone people and businesses when they haven't objected and the number isn't on the Telephone Preference Service (TPS) or, for corporate subscribers, the Corporate TPS (CTPS). Registration is free, includes mobiles, and takes 28 days to become active. The ICO's guidance says a supplier's claim that a list is screened needs a recent check behind it, and you stay responsible either way. Sole traders and some partnerships count as individual subscribers, and the ICO's B2B guidance says PECR's rules generally apply to both kinds of subscriber, with the email consent rule the main exception. When someone objects, add them to a suppression list rather than deleting the record.

Singapore: The Do Not Call Registry

The PDPC runs three Do Not Call registers: voice calls, text messages and fax. For messages the rules cover, a sender needs valid confirmation that a number isn't listed, and the guidelines (revised 1 February 2021) set a check's validity at 21 days. Messages sent to an organization for any of its own purposes are excluded. The PDPC's example: ringing an HR manager's business number to promote something the company would buy is outside the rules, but asking about a product for her personal use brings it back in. Senders must identify themselves and not withhold their calling line identity.

United States: Do Not Call and the TCPA

The FTC's Telemarketing Sales Rule exempts most B2B calls, except those selling nondurable office or cleaning supplies and those soliciting individual employees to buy for their own use. The FTC treats software and computers as durable goods. Where the rule applies, it bars calls to consumers' homes before 8 a.m. or after 9 p.m. local time unless they've agreed otherwise.

The separate Telephone Consumer Protection Act, 47 U.S.C. § 227, restricts calls made with an automatic telephone dialing system or an artificial or prerecorded voice to cellular numbers unless the called party gave prior express consent. Its wording doesn't distinguish business mobiles from personal ones, and the FCC's 2024 ruling applies the artificial-voice language to AI-generated voices. Whether a given dialer counts as an automatic one is a legal question, so have counsel review dialers and voice tools before they call mobiles.

Everywhere Else

Malaysia, the Philippines, Vietnam and Indonesia each regulate personal data under their own laws, which the outbound sales playbook lists with sources. Business culture in Southeast Asia describes a shared high-hierarchy, relationship-first pattern with Singapore as the outlier, which suggests a warm introduction or earlier email touch may help a first call. Some habits hold everywhere: identify yourself, use numbers whose source you can explain, check registers where they exist, and honor a stop request across your tools.

Run One Week of It

Pick 40 tier-one accounts, find direct dials where you can, and tag each number's type. Prep each call in two minutes, dial in two different windows and log every outcome with the codes above. After a week you'll have a connect rate that's yours, split by number type and time of day.