Guides
How to Get Real Feedback on Your Cold Email
2 August 2026
Reply rate takes two weeks to tell you anything and your colleagues will be nice about it. Here is how to grade a cold email before you send it, against the five things that actually decide whether it gets answered.
The usual options for getting feedback on a cold email are all bad in the same way.
Send it and watch the reply rate: honest, and two weeks late, and only conclusive at volume you probably are not sending. Ask a colleague: they will tell you it reads well, because telling somebody their writing is bad is socially expensive and they have their own pipeline to worry about. Post it in a community: you will get five contradictory opinions, at least one of which is from somebody selling a course.
Paste it into a general AI model and it will be polite. Ask for feedback on a cold email and you will get "this is a strong opener, you might consider..." on an email that is not a strong opener.
So the useful thing is not another opinion. It is a fixed set of criteria you can grade against yourself, in ninety seconds, before it goes out.
The five things
We read a large amount of published cold email advice while building our coaching product: 6,794 pages of it. Most of it argues about subject lines and length. Almost none of it is about the thing that decides whether a stranger answers, which is whether the email demonstrates you know something specific about them and their problem.
Grade every email you write against these five. Be harsh. Most emails fail two or three.
1. The first line is about them, and could not be sent to anyone else
The test is mechanical: could you paste this first line into an email to a different company without changing it? If yes, it is not personalization, it is a template with a merge field.
"I saw you're hiring three SDRs" passes. "I noticed you're focused on growth" fails, because everyone is focused on growth. "Congrats on the funding" fails as well, and worse than it looks, because they have had forty of those this month.
2. There is a problem in it, and it is theirs
Not your product's category. Not "improving sales efficiency." A specific bad thing that is plausibly happening to them right now, stated in a way that is falsifiable enough to be wrong.
Being specific enough to be wrong is the point. An email that cannot be wrong cannot be right either, and the buyer can tell the difference instantly. The risk of naming a problem they do not have is much smaller than the certainty of being ignored for naming a problem nobody has.
3. It says what it is asking for, once, plainly
Not "would you be open to a quick chat to explore synergies." A specific ask with a size to it, so they can say yes or no without a follow-up email to establish what you meant.
Two asks is zero asks. So is an ask buried under a paragraph of qualification.
4. There is nothing in it for you to delete
Read it and delete every sentence that does not survive the question "would they miss this?" Almost always this removes: your company boilerplate, the second sentence about what you do, the logo list, and the line explaining why you are reaching out. The email gets shorter and better every time, without exception.
5. It does not perform sincerity
The tells are recognizable because they are taught everywhere. "I'll be brief." "Hope this finds you well." "Just checking in" on a follow-up. "I know you're busy, so." Each one signals that the sender is running a script, and the buyer categorizes the email before reading the substance.
One is worth naming on its own, because it is the most-taught follow-up in the entire body of advice we read, taught on 151 separate websites: "just checking in." It is a follow-up that contains no new information, and it asks the recipient to do the work of remembering why they cared. If your second email is a checking-in email, it is not a follow-up, it is a reminder that you want something.
Grade it yourself, right now
Take the last cold email you sent. Score each of the five out of two: 2 if it clearly passes, 1 if you are arguing with yourself about it, 0 if it fails.
Under 6 out of 10 and the problem is not the subject line, whatever the reply rate is telling you.
The thing to watch is not the total, it is which criterion you fail repeatedly. Almost everyone has one they lose on every time. Ours was 3, for about a year.
Then get a read that is not yours
The limit of self-grading is the same as reviewing your own calls: you cannot catch what you did not notice. You wrote the email believing the problem statement was specific, which is exactly why it reads as specific to you.
It is worth knowing what a general model does on this task before you trust one with it. We ran a blind test: six models wrote a cold email from one identical brief, then judged each other while playing three skeptical buyers, with labels and order randomized so no judge knew who wrote what. Eighteen verdicts. The coached model won 17 of them. Four of the frontier models won none at all. The method and the raw judge reasoning are published, so you can argue with it rather than take our word for it.
The benchmark, with the method
That is what our cold email grader does. It scores against a fixed standard rather than an opinion formed on the spot, it tells you which of the five you failed and quotes the line, and it does not soften it to be encouraging.
Bring one you actually sent, not one you write for the test. The one you wrote for the test will score well and teach you nothing.