Is cold calling dead? The data says no.

The short answer

No. Cold calling isn’t dead. Lazy cold calling is.

AI is taking over most of what surrounds the call: the research, the list, the first draft of every email, the notes afterwards. That’s good news, and you should use all of it. What AI can’t do well, legally or socially, is have a live conversation with a stranger who didn’t ask to talk to it. That part is still yours.

• Buyers still want a person. 69% of B2B buyers check what AI tells them with a sales rep (Gartner, 2026).
• The phone is reaching people. It took 1.55 calls on average to reach a prospect in Cognism’s 2026 data, down from 2.9 the year before.
• Fewer reps are doing it. 48% of sellers say they don’t have the bandwidth for proper cold outreach (Salesforce, 2026).

Put those together and you get the opportunity: a channel buyers still answer, with less skilled competition on it every quarter.

What AI is genuinely good at

Start with the honest part. AI makes reps more productive, and anyone telling you to ignore it is going to lose to someone who doesn’t.

In Salesforce’s 2026 State of Sales survey of 4,050 sales professionals, the average seller spent only 40% of the week actually selling. Sellers using AI agents expect them to cut prospect research time by 34% and email drafting time by 36%. 87% of sales organizations already use some form of AI.

Two cautions keep this in proportion. First, measured savings are smaller than the promises: across US sales jobs, generative AI saved about 1.5% of total work hours in mid-2026, averaged over users and non-users alike (Bick, Blandin and Deming’s tracker). Second, more tools isn’t the same as more output. Gartner predicts AI agents will outnumber sellers 10 to 1 by 2028, yet fewer than 40% of sellers will say the agents improved their productivity.

The lesson: AI gives you hours back. What you do with those hours decides whether it pays off.

Where AI stops: the live cold call

The rules

In February 2024 the FCC ruled that AI-generated voices count as “artificial” voices under the Telephone Consumer Protection Act. Calls that use them generally need the person’s prior express consent. A cold call is, by definition, someone who hasn’t given you that. If you’re thinking about putting an AI voice on outbound, talk to counsel first. (This isn’t legal advice; it’s a reason to ask for some.)

The phone itself is screening

Since iOS 26, iPhones can ask unknown callers for their name and the reason they’re calling before the phone even rings through. A real person with a real reason can answer that question. A blast can’t.

The buyer

Gartner’s 2026 survey found buyers lean on AI to research (45% used generative AI in a recent purchase) and then turn to a person to check it: 69% validate AI-generated insights with a sales rep. Gartner also predicts that by 2030, 75% of B2B buyers will prefer sales experiences that put human interaction ahead of AI.

What we haven’t found

We have not found an independent study showing AI voice agents booking cold B2B meetings at scale. There are plenty of vendor claims. When someone publishes real numbers with a method behind them, we’ll add them here.

Why the phone is wide open

Here’s the part most “cold calling is dead” posts miss. As inboxes fill with AI-written email and teams cut their callers, the live call gets rarer. Rarer means it stands out.

The phone also produces far more live conversations than email produces replies. In Belkins’ 2025 data, 1,000 dials led to about 57 real conversations and roughly 2.7 booked meetings. 1,000 cold emails led to about 4.5 replies of any kind, positive or not.

And the gap between average and good is skill. Cognism puts the average cold call success rate at 2.7% in 2026, against 11.3% for its own trained team. Meanwhile, 47% of Gen Z reps say they don’t get enough roleplay before customer calls (Salesforce, 2026). Most teams are handing their reps AI tools and no practice. That’s the opening.

The split: the orchestra and the solo

We say it this way: AI is the orchestra. The call is the solo. Let the tools carry everything around the call, and train the human for the part only a human can do.

A 15-minute AI prep routine before a call block

Use whatever assistant your company approves, and keep confidential customer data out of tools that aren’t cleared for it.

1. Research (5 minutes for your top accounts)

2. Openers (5 minutes)

Then read them out loud and rewrite the one you’d actually say. If it sounds like an email, it’s not done.

3. Objections (5 minutes)

Write your own answers, then say each one out loud before the first dial. That’s what Practice mode is for. (More prompts like these in the prompt library.) Then close the AI tab and dial.

Where AI backfires

• Everyone sounds the same. When every rep pastes the same prompt, every opener sounds alike. Use AI for the first draft and your own words for the last one.
• Volume that burns your number. Auto-dialing at scale gets numbers labelled as spam, and then nobody picks up, AI or not.
• Facts you didn’t check. Saying something wrong about a prospect’s company ends the call faster than any objection. Verify anything you plan to say out loud.
• Saved time that turns into busywork. The hours AI gives back should go to more dials and more practice, not a tidier CRM.

For sales leaders

• Measure conversations, not tool adoption. Live conversations and meetings booked tell you whether AI is working. Logins don’t.
• Spend the saved hours on the phone. If AI saves each rep five hours a week, put at least two of them into call blocks and two into practice.
• Bring back roleplay. Nearly half of young reps say they don’t get enough of it. Ten minutes of out-loud practice before a block is the cheapest edge you can give a team.

Sources

1. Gartner (May 20, 2026). Survey of 645 B2B buyers: 69% turn to sales reps to validate AI-generated insights.
2. Gartner (Aug 25, 2025). By 2030, 75% of B2B buyers will prefer sales experiences that prioritize human interaction over AI.
3. Gartner (Nov 18, 2025). By 2028 AI agents will outnumber sellers 10x, yet fewer than 40% of sellers will report improved productivity.
4. Salesforce (Feb 3, 2026). State of Sales, 4,050 sales professionals in 22 countries.
5. Bick, Blandin and Deming. Generative AI time savings, sales and related occupations (Real-Time Population Survey, via FRED).
6. Cognism. Cold Calling Report 2026 (200K+ calls; vendor data).
7. Belkins. B2B cold calling benchmarks, 175,000+ dials in 2025.
8. Belkins. Cold email response rates, 7.53 million emails in 2025.
9. Cooley (Feb 15, 2024). FCC: AI-generated voices are artificial under the TCPA.
10. Euronews (Oct 9, 2025). How iOS 26 Call Screening handles unknown callers.