Personalization is not dead. What died is the signal that used to stand in for it. For most of my career, dropping in a detail about where someone went to school or what they posted last week proved you had done work, and the proof was the point. AI drove the cost of that proof to zero, which means the detail by itself now says nothing about you except that you own a tool, and a buyer who senses that will give you fewer points than if you had sent nothing at all.
TL;DR: Personalization used to mean proving you did the homework. AI made homework free, so the homework no longer proves anything. What stands out now is evidence that you spent something real, money, time, or the willingness to look stupid, on one specific person.
What is personalization in sales, actually?
I define personalization as evidence that you made a decision about this specific person. Not a merge field, not a scraped fact, a decision. Somebody looked at one human being, formed a point of view about what matters to them, and acted on it in a way that would make no sense if sent to anyone else.
The version most teams run is something different. It is research display. You find a fact, you lead with the fact, and then you hard cut to a value proposition the person has no interest in. I get those every day. Hey John, I see you went to the University of Maryland, and then straight into the pitch. That is not personalization, that is a costume, and I actually give fewer points for it than for a clean, honest, generic email.
The distinction I keep coming back to is that personalization without relevance is worse than no personalization at all. The research-display version was always hollow. AI only made it obvious by making it cheap.
Why did AI change what counts as personalized?
Because the cost collapsed. When it took me twenty minutes to find something real about a prospect and tie it to something they actually care about, the detail carried information. It told the buyer I had chosen to spend twenty minutes on them instead of hitting send on a hundred more emails. The fact was never the value. The spend was the value.
Now anybody can generate a first line that reads as researched in under a second, at any volume. The signal still looks the same on the page, but it costs nothing, so it proves nothing. Buyers adjusted faster than sellers did.
You can see it in the numbers. Cadence tools flooded the market, then AI tools stacked on top of the cadence tools, and reply rates went down, not up. This was never a tool problem. The tools put the old problem on steroids and made it impossible to hide.
What does the 'seems like AI slop' button tell us?
A million people have clicked LinkedIn's "seems like AI slop" button. That is the part I want sellers to sit with. Buyers now have a literal button to punish fake effort, on the platform I post to every single day.
That matters because it is the first hard evidence that the commoditized version of personalization has moved from neutral to negative. It used to be that a weak personalized line was wasted motion. Now it is a penalty. Somebody reads it, feels handled, and clicks.
I've drawn my own line on this. I have no problem consuming content created by AI as long as I know it came from AI. What I can't stand is someone using AI and pretending a human did it, in emails, in LinkedIn comments, in chatbots. When I sense that, I stop caring, and I'm not special. Everyone is developing that sensor.
What real personalization looks like: the Olive Garden card
Years ago I wanted to get to Guy Kawasaki. I didn't send him a clever first line about something he posted. I sent him an Olive Garden gift card.
It was a specific bet on something I believed about him, and it cost me money, it cost me effort, and it carried real risk of making me look like an idiot if I had read him wrong. That risk is the whole point. No AI tool is going to spend your money or absorb your embarrassment on your behalf.
Here's the tell that it worked. A week later he sent me an unprompted photo. Not a reply to my follow-up, not a polite acknowledgment, something he did on his own because the thing stuck with him. You cannot buy that with a personalized opening line, at any price, from any tool. It came from the one currency that hasn't been commoditized.
The two currencies of personalization
The simplest way I can put it: personalization used to be about proving you did the homework. Now it's about proving you gave a shit. Those are different currencies, and almost every team I talk to is still spending the one that has been devalued.
Here is how I separate them when I look at a rep's outreach:
- Homework currency: a fact about the person, their company, their post, their alma mater. Cost to produce, near zero. Signal value today, near zero or negative.
- Give-a-shit currency: money spent, time spent, a real point of view attached to your name, a risk of being wrong in public. Cost to produce, real. Signal value, high and rising.
- Homework asks the buyer to be impressed that you found something. Give-a-shit gives the buyer something they didn't have before you showed up.
- Homework can be generated at volume. Give-a-shit cannot, which is exactly why it works.
- Homework is reversible. Delete the email and nothing happened. Give-a-shit leaves a mark, like a photo showing up in your inbox a week later.
How do you build this into a cadence without blowing up the math?
The objection I always get is volume. If real personalization costs money and effort, you can't do it for five hundred accounts. Correct. That is a feature, not a bug.
What I teach is to use AI for the part of the work that genuinely scales, which is the research, the pattern-finding, the first draft, the account prioritization, the summarizing of ten earnings calls into the two things a buyer is actually being measured on. Then spend the time AI gave you back on fewer accounts, with something that costs you something.
The reps who lose are the ones who take the time AI saves them and convert it into more volume. More emails that look researched and cost nothing, into a market where buyers have a button to flag exactly that. That is apathy with a productivity dashboard attached.
The JB Sales Approach
I teach this as the give-a-shit factor, and it runs through both Filling the Funnel and Driving to Close. The test is simple. Before anything goes out, I ask one question about it: what did this cost me, and would the person on the other end be able to tell?
If the answer is nothing, it is not personalized, no matter how many of their details are in it. If the answer involves money, real thinking, a genuine point of view, or the possibility of looking foolish, you're in the right currency.
In practice, here is the workflow I run reps through:
- Use AI as a force multiplier on research, not on sending. Let it compress an hour of account work into ten minutes.
- Take the time you saved and reinvest it into fewer, better touches instead of more touches.
- Make a specific bet about what this person cares about and commit to it in a way that would be absurd to send to anybody else.
- Accept that you will be wrong sometimes. The willingness to be wrong in public is the cost that makes it credible.
- Never use AI to produce something and then pretend a human made it. Buyers sense it, and the moment they do, you've spent your credibility instead of earning it.
What to do this week
Pull your last twenty outbound touches and sort them into the two currencies. Mark every one where the personalization was a fact you found, and every one where you actually spent something. My guess is the first pile is almost all of them.
Then pick three accounts, just three, that matter enough to justify real spend. Not your whole list, three. Do something for each of them that costs you money, time, or the risk of looking stupid, and send it with no pitch attached. Watch which pile generates a reply you didn't have to chase.
If you're a leader looking at this and wondering whether the issue is your reps or your motion, that is the replace, rebuild, or retrain question, and it's worth answering deliberately rather than by attrition. The framework I use for it is at learn.jbarrows.com/pages/replace-or-rebuild. If you want your team working this way as a system, Filling the Funnel and Driving to Close are the team programs where we build it in, at learn.jbarrows.com/pages/team-training-packages. For individual reps who want direct access to me, Individual PRO is $497 a year at learn.jbarrows.com/pages/individual-packages. And if you're a sales leader who wants the diagnostic version, JB Sales Advisory is a 3-month strategic engagement that starts with a written diagnostic of your sales motion, AI readiness, and team structure, and ends with a 90-day plan you can execute, at learn.jbarrows.com/pages/advisory.
Frequently asked questions
Is AI personalization in sales emails still worth doing?
Using AI to research an account and draft faster is absolutely worth doing. Using AI to generate a personalized-sounding first line and calling that personalization is not, because buyers can tell and now they have a button to flag it. Use the tool for the research, spend the saved time on something that actually costs you something.
Why are my reply rates dropping even though we personalize every email?
Because the thing you call personalization costs nothing to produce, so it no longer signals effort. Cadence tools flooded the market, AI tools stacked on top of them, and reply rates went down while volume went up. The problem was never the tools, it is that the old proof of effort stopped proving anything.
What is the difference between fake personalization and real personalization?
Fake personalization displays a fact about the buyer and then hard cuts to a pitch they have no interest in. Real personalization is evidence you made a decision about that specific person and spent money, time, or the risk of being wrong on it. One asks the buyer to be impressed. The other gives them something.
Doesn't real personalization make it impossible to hit activity targets?
It makes it impossible to hit activity targets the old way, which is the point. I'd rather a rep run three accounts a week with real spend than two hundred with scraped first lines. Let AI compress the research so you can afford to go deep on the accounts that matter.
Should I tell prospects when I've used AI to write something?
Yes, and I hold myself to this too. I've no problem consuming content created by AI as long as I know it came from AI. What kills trust is using AI and pretending a human did it, because when someone senses it, they stop caring about anything else you say.