Cold Email That Still Works for B2B Services

Cold email still works for B2B services, but almost everything that used to make it work no longer does. It remains one of the most cost-effective ways to start a sales conversation with a decision-maker, with a return on investment often estimated around 42 to 1, because it reaches the right person directly instead of bidding against algorithms for their attention. What has changed is that the volume-and-template version of cold email, the one most people mean when they say “cold email,” is dead, and clinging to it is why so many teams conclude the channel is finished.

The numbers tell the story of a channel that got harder, not one that died. Average cold email reply rates have fallen steadily, from around 8.5% in 2019 to about 5% in 2025 and roughly 3.43% entering 2026 (Instantly). That decline is real, and it has a cause: inboxes are saturated, spam filters are stricter, and a flood of generic AI-written outreach has trained buyers to delete on sight. The channel did not stop working. The lazy way of using it did.

Why the volume playbook died

For a decade the cold email playbook was simple: buy a big list, send a template to everyone, and let volume do the work. Three forces have now made that approach actively counterproductive.

The first is buyer fatigue with generic outreach. Larger, less-targeted campaigns to 500 or more recipients now average around 2.1% response, and every irrelevant email is a small withdrawal from your reputation with exactly the people you most want to reach. More volume against a saturated market produces more waste, not more pipeline.

The second is the AI-generic penalty. Buyers can recognise a ChatGPT-written email, and generic AI-written outreach sees roughly 90% lower response rates (Built For B2B). The tool that made it trivial to send more email also made the average email worthless, which means the volume that AI enabled is the exact thing killing the channel for the people relying on it.

The third is a deliverability crackdown. Since 2024, Google and Yahoo have enforced strict authentication for bulk senders, and by late 2025 Gmail began rejecting non-compliant mail outright rather than just filtering it to spam, with Microsoft following for Outlook. Sloppy senders are no longer landing in spam; they are not landing at all.

What actually works now

The version of cold email that still works is the opposite of the volume playbook in every respect. It is built for relevance, not reach.

Precise targeting comes first. A smaller list of genuinely well-fit prospects beats a large list every time, because relevance is now the entire game. If your reply rate is below 2%, the problem is almost always targeting or deliverability, not your copy.

Real personalisation is second, and it is where the results live. Advanced personalisation, tailored to the recipient’s actual context, roughly doubles reply rates against generic templates, from around 9% to around 18%, and highly personalised campaigns can lift replies by over 140% against blasts (Martal). Yet only about 5% of senders personalise every email, which is precisely why the ones who do stand out.

Follow-up is third and badly underused. Follow-up messages generate about 42% of all replies, and yet nearly half of senders never send a second message, abandoning half their potential responses. A cold email is rarely a single send; it is a short, persistent sequence.

Multi-channel is fourth. Email alone now returns 4 to 6%; email plus LinkedIn reaches 8 to 10%, and adding a phone touch pushes it to 10 to 12%. The same message across coordinated channels outperforms any single channel, because it meets the buyer where they actually pay attention.

AI for research, human for delivery

Given that generic AI writing is the thing killing response rates, the right use of AI in cold email is precise and counterintuitive: use it for research, not for writing.

AI is superb at the work that makes personalisation possible at scale, finding the right accounts, surfacing the signal that gives you a genuine reason to reach out, and assembling the context that a relevant first line is built from. What it should not do is write the email in its own generic voice, because that is exactly the output buyers have learned to ignore. The winning pattern is AI for the research and the targeting, and a human, or a human-shaped, specific message for the delivery. The technology handles the volume of finding and knowing; the message stays personal, because people still reply to people.

The technical floor nobody can skip

None of the above matters if the email does not arrive, and deliverability is now a hard floor rather than a nice-to-have. SPF, DKIM, and DMARC authentication are non-negotiable. New sending domains need weeks of warming before volume. Bounce rates must stay under about 2% and spam complaints under 0.3%, or the mailbox gets throttled or blocked. These are not growth tactics; they are the conditions for the channel functioning at all, and the teams that skip them watch their emails silently disappear while they blame their copy.

Cold email is one channel, not a strategy

The most important reframe is that cold email is a channel, not a system. On its own it is a tactic, and a tactic is not a growth plan any more than a hammer is a house. It works when it sits inside a larger B2B sales and demand system: targeting that decides who to reach, a reason to reach them that demand generation created, a follow-up sequence, and a conversion path for the replies. A brilliant cold email into a broken pipeline still leads nowhere.

That is why cold email belongs inside the owned client-acquisition engine rather than run as a standalone campaign, and why it is one of the ways a firm gets clients without the founder living in the pipeline: systematized, personalised at scale, and running whether or not anyone is manually sending. Outbound that still works is mostly craft, and craft is easier to watch than to read about. If you want to learn it by building your own, The Worldshaper Workshop is Daniel Förster’s daily build sessions, run by Thane Alaric: watch the work happen, then build it yourself. See The Worldshaper Workshop.

Frequently Asked Questions

Does cold email still work in 2026?

Yes, but the volume-and-template version does not. Cold email remains highly cost-effective for B2B services, with an ROI often estimated around 42 to 1, because it reaches decision-makers directly. Average reply rates have fallen to around 3.43% as inboxes saturated and generic AI outreach flooded in, so the channel now rewards relevance and punishes volume. Done well, targeted and personalised, it still books meetings.

Why have cold email response rates dropped?

Three forces. Inbox saturation and buyer fatigue mean generic outreach gets deleted; large untargeted campaigns now average around 2.1%. Generic AI-written emails see roughly 90% lower response because buyers recognise them. And a deliverability crackdown by Google, Yahoo, and Microsoft now rejects non-compliant mail outright. Reply rates fell from about 8.5% in 2019 to roughly 3.43% in 2026 as a result.

What makes a cold email get a reply?

Relevance, built from targeting and real personalisation. A small, well-fit list beats a large one, and personalisation tailored to the recipient’s actual context roughly doubles reply rates against generic templates. Persistent follow-up matters too, since about 42% of replies come from follow-ups, as does running the message across email, LinkedIn, and phone rather than email alone.

Should I use AI to write cold emails?

Use AI for research and targeting, not for writing the message. Generic AI-written emails see about 90% lower response rates because buyers recognise them. AI is excellent at finding the right accounts, surfacing a genuine reason to reach out, and assembling context; the message itself should stay specific and human, because that is what still earns replies.

How important is deliverability for cold email?

It is a hard floor. SPF, DKIM, and DMARC authentication are non-negotiable, new domains need weeks of warming, and bounce rates must stay under about 2% with spam complaints under 0.3%. Since late 2025 Gmail rejects non-compliant mail outright rather than filtering it to spam. Without the technical setup, even a perfect email never arrives.

How does cold email fit into a wider sales strategy?

Cold email is a channel, not a strategy. It works only inside a larger system: targeting that decides who to reach, demand that gives you a reason, follow-up, and a conversion path for replies. Run as a standalone campaign it produces little; run inside an owned client-acquisition engine, it becomes one reliable way to get clients without the founder manually working the pipeline.