The ask · from the corpus
We measured four kinds of cold-email ask across 44 client accounts. The open question beat the meeting request roughly two to one — and asking for a call was the worst-performing CTA in every segment we could split. Here is the ranking, and what to ask instead.
The ranking
Same reader, same offer, different closing line. This is the order they came in, measured on qualified replies, not opens.
“Do you want this, this year?”
The strongest ask we measured. It asks for a decision the reader already owns, and costs them nothing but a yes or a no.
“Is this actually your call?”
Nearly as strong. It flatters the reader's authority and gives them an easy forward if it is not them.
“Shall I send it over?”
Solid, and the most common ask by far. Low threshold, but it makes the reader do the imagining.
“Worth a quick call?”
The weakest, and it was the weakest in every single segment we could split. A first email has not earned a slice of someone's calendar.
EVIDENCE · MEASURED ACROSS 44 CLIENT ACCOUNTS, FIRST AND SECOND TOUCHES. RELATIVE ORDER SHOWN; WE KEEP THE ABSOLUTE RATES PRIVATE.
The other half
Winners carry numbers constantly, so “never use numbers” is the wrong lesson. The real one is about which number. A number that describes you is credibility. A number that predicts the reader’s future is noise. “Seven distributors signed since 2025” earns trust; “increase your margin 30%” spends it. In our corpus the predictive improvement claim performed far worse than its absence, and pairing one with a meeting ask was close to the worst combination we measured. If your CTA promises the reader a result, cut the promise and keep the ask.
THE GRADER FLAGS A MEETING ASK ON A FIRST TOUCH
Questions
An open question the reader can answer yes or no to without leaving their inbox — something like “do you want this, this year?”. Across 44 client accounts we measured, the open question produced roughly twice the qualified-reply rate of the classic “worth a quick call?” meeting request. The order held in every segment we could split.
Because a first email has not earned it. A call is a real cost — a calendar hole, a commitment, a small act of trust — and you are asking a stranger to pay it before you have proven anything. Ask for a reply, not a slot. The meeting request was the worst-performing ask in every segment of our data.
Only if it describes you, not the reader. A number that describes you is credibility (“seven distributors signed since 2025”). A number that predicts the reader's future is noise (“increase your margin 30%”). In our corpus, emails carrying a predictive improvement claim performed far worse than those without one, and pairing an improvement claim with a meeting ask was close to the worst combination we measured.
One. A single, small, answerable ask. Two asks split the reader's attention and neither gets done; a big ask gets ignored. The whole email should build toward one yes-or-no question and then stop.
From the cold email we run for clients: campaigns measured across 44 client accounts, on variants with enough volume to mean something. We publish the ranking and the relative differences, and we label what is measured evidence versus house judgement, because presenting craft as data is how an agency loses the right to be believed.
This is one page of the cold email playbook. The same discipline runs the whole sequence, starts with the subject line, and is what we do for clients as a cold email agency.