Lieben says the offer is the one line nobody tests

Michel Lieben posted that the highest ROI line in a cold email campaign is the offer, and that it is the one line nobody tests. He framed it as a hard pill for anyone running cold email with AI, the point being that teams pour effort into generated copy and enrichment while the thing they are actually asking for goes unexamined.

His spec for that line has four parts. Something free with real value in it. Something intriguing enough that the prospect wants to reply. Something built around what they want from a cold email, tested until it lands. And plain copy around it, because in his model the offer does the work rather than the sentences wrapped around it.

The ranking he is making

The claim is comparative, not just prescriptive. Lieben puts the offer above the whole personalization stack, naming AI lines, Clay signals and Claude skills as things that lose to a good offer.

"A good offer beats every AI line, Clay signal and Claude skill you can stack on top."

That is worth reading with one eye open. Lieben is a regular voice on building outbound inside Claude Code and on the APIs that sit behind it, so he is partly telling senders to stop optimising the layer he works in. It does not make the argument wrong, but it is an assertion about where return comes from, not a test result. No campaign, no volume, no reply rate accompanies the post.

For people sending

The practical read is that an offer test is cheap and most senders have never run one properly. You can hold list, sequence and send infrastructure constant and vary only the ask, which is a cleaner experiment than most copy tests because the variable is a single line and the outcome, replies, is easy to attribute. The harder part is Lieben's third condition. Building the offer around what the prospect wants from a cold email, rather than what you want to sell, usually means offering something you can hand over without a call.

There is at least one data point in circulation pointing the same way. Earlier this month Lucas Feldman said an offer switch moved a client from a single reply to 125 leads in a month, with the list and the channel unchanged. One operator's result on one account is not proof of a rule, but it is the kind of swing that only an offer change explains, and it is the test Lieben is telling people to run before they touch another prompt.

Michel Lieben

@MichLieben

X

A good offer beats every AI line, Clay signal and Claude skill you can stack on top.