AI made the personal first line free

A free personal line reads as a template

The personal first line used to prove you did your homework. Now a tool writes one in a second, so almost every sender has one.

Open a sales inbox this year and you will likely see the same opener. A compliment, a nod to a recent post, then the pitch.

When every sender can add the same personal detail, it tells the buyer nothing about you. It stops being a sign of care and becomes part of the template.

That is the whole problem. The line was a signal because it was costly, and AI made it cheap. My view is simple: stop faking it.

The wider problem got a name last year. A major US dictionary named slop its 2025 Word of the Year. It defines slop as "digital content of low quality that is produced usually in quantity by means of artificial intelligence." [2]

Most decision makers say AI-written email bothers them

The sharpest number comes from a 2026 cold email report. It pairs 31 million emails its users sent in 2025 with two surveys.

One survey asked senders, and the other asked decision makers who get cold email. 69% of US decision makers said it bothers them if AI wrote the email. [1]

The report's summary adds a condition: unless the result feels truly human. The team behind it says its earlier surveys found more willingness to accept email that AI helped write. [1]

Decision makers in the survey also named what fails. 65% said cold emails feel too pushy or sales-focused, and 48% called out generic, impersonal messages. [1]

They have not given up on cold email, though. 28% said they never get a valuable one, but 58% get something useful at least once a month. [1]

Sales reps on Reddit's sales forum argue about this. One thread says that with so much AI slop in sales email, standing out has never been easier. [4]

Another thread shrugs off the worry about AI-written email to buyers altogether. [5]

I side with the first thread. Most messages do go unanswered, as the numbers below show, but the template is a big part of why.

Personal details and a human edit still win

The data does not say you should stop being personal. Personal details still pay, and so does a person's hand on the draft.

67% of decision makers said personal details from public information make them more likely to reply. Emails with two custom fields in the body got a 5.6% reply rate, against 3.6% with none. [1]

Emails a person edited by hand beat fully automated ones, 5.2% to 4.4%. Lists of 21 to 50 people got a 6.2% reply rate, while lists of more than 500 got 2.4%. [1]

Put the three pairs side by side and the pattern is hard to miss. In each pair, the smaller or more personal version won.

A second benchmark, built on billions of cold email interactions from 2025, points the same way. The average reply rate was 3.43%, and the top 10% of campaigns got 10.7% or more. [3]

Its top senders kept first emails under 80 words on average. They led with the buyer's problem and asked for one thing. [3]

I think small lists and hand edits work for the same reason. They force you to know who you are writing to before you write a word.

A dated fact, or nothing personal

Here is the rule I would hold any writer to, person or tool. A personal line must point to something the buyer said or did, with a date.

Take a made-up example to see the line. A post from September 30 about a move to a new loan system counts. A job title scraped from a profile does not, and neither does a compliment.

If no such fact exists, drop the personal line. Open with a dated company event instead, like a funding round or a burst of hiring. Signal-based selling walks through that check.

One mistake to avoid is the compliment opener, such as telling someone you loved their recent post. That shape reads as a template, and it proves nothing.

A person's edit is the last check, so read the draft as the buyer would. Cut any line you could not defend if they asked where it came from.

Our own runs often say no

We built Mach on this rule, so we can show you what it costs. For one firm we work with, Mach read each person's recent public posts before writing.

Across 294 runs it read 6.9 posts per person on average, and only 2.4 of them were about work.

It found an honest personal opener for 188 of 294 people, or 64%. The other 106 got no personal line at all. Here is the full count.

For a software services firm we work with, Mach was asked for 167 first messages over two days. It kept 62 and refused 105. In the end, 46 of the 87 people it wrote for got a ready message.

Each refusal stores its reason in plain words, so your rep sees why instead of getting a weak draft.

Most refusals were drafts that still broke the writing rules after one fix. Another 9 were declined because no honest opening existed. Here is why it says no.

That is a lot of no, and some of it was our own checker's fault. Still, I would rather send 62 messages a buyer can check than 167 I could not stand behind.

We do not publish reply rates for these runs yet, because that data is too young to defend.

Rewrite one small list this week

Pick 50 people you plan to write to. For each one, find a dated fact in their own words or in their company's news.

Where you find none, cut the personal line and lead with their problem. Then compare the replies with your last big list.