Cold email personalization
Personalization used to mean proving you knew who someone was. Now that every tool does that automatically, it has to mean proving you understand what they are dealing with.
Michael Sparks · Published 2026-08-18 · Last reviewed 2026-08-18
Direct answer
Why token personalization stopped working
A first name in a subject line was once a small proof of effort. It is now the default output of every sequencing tool, which means it carries no information about whether the sender thought about the recipient at all.
Belkins analysed 7,530,489 cold emails sent in 2025 and reported an average of roughly 0.45 percent replies measured against total sends, about 20 percent lower in the second half of the year than the first. Volume and automation went up over the same period. Buyers adapted faster than templates did.
Signals that still change response
- A public commitment with a timeline, because it implies work already underway.
- A recent change of ownership in the function you sell into, because priorities are being reset.
- A specific operational constraint the buyer has described publicly, in a talk, post or report.
- A comparison the buyer is already making between two approaches you can speak to credibly.
Each of these gives you a sentence that could not have been sent to anyone else. That is the actual test.
Signals that are mostly noise
- Congratulating a funding round, which every sender does within the same week.
- Referencing a blog post the buyer did not write and probably did not read.
- Technology detection that is frequently inaccurate and rarely interesting to the buyer.
- Generic industry statements dressed up as insight about their specific company.
What to send instead of a generic sales deck
A deck asks the buyer to sit through your narrative structure. A short message plus a tailored destination lets them jump to the part they care about and share it internally without a meeting.
- A page stating the situation you observed and what you think it means, with sources.
- The single customer story closest to their industry, size and problem.
- A comparison of the two or three approaches they are realistically weighing.
- One next step, sized to the trust you have earned rather than the one you want.
Keeping quality when volume rises
Quality collapses when personalization depends on individual effort. It holds when brand, structure, claims and proof come from an approved system, and the rep's judgement is spent only on which situation and which evidence apply.
That is also the practical reason to cap the list size. If nobody has time to read each draft before it goes out, the personalization is decoration.
What this means in practice
Drop the tokens you cannot defend, keep two or three real signals per account, move the detail out of the email and into a destination built for that buyer, and only run the depth your team can review.
| Signal | Strength | Research cost | Fails when |
|---|---|---|---|
| Public commitment with a date | High | Low | The initiative is already staffed |
| New function owner | High | Low | They are too new to decide |
| Stated operational constraint | High | Medium | The quote is taken out of context |
| Funding announcement | Low | Very low | Every competitor sends the same note |
| Technology detection | Low | Low | The detection is simply wrong |
Common questions
- What should I send a prospect instead of another generic sales deck?
- Send a short message plus a destination built around their situation: what you observed, what you think it means, the closest customer proof and one next step. It lets the buyer skip to what matters and share it internally without scheduling a call.
- Does personalization still increase reply rates?
- Personalization of the argument does, because it demonstrates understanding of the buyer's situation. Token personalization such as first name and company name does not differentiate a message anymore, since nearly every sender automates it.
- How much research is enough per account?
- Two or three verifiable signals that the buyer would recognise as accurate and current. Beyond that, additional research usually improves the sender's confidence more than the buyer's experience.