LinkedIn DM outreach to CEOs, sized for twenty-five named targets rather than two thousand

A CEO does not read your message, they open an investigation

The sequence is almost always the same. A notification arrives with your name and headline attached. If the first line survives, they open the message. If the message makes any claim at all, they open your profile to see whether the claim holds. If the profile holds, a good number of them scroll your recent activity, because activity is the only part of a profile you cannot write in an afternoon. Then, and only then, they decide about your ask.

So the DM is roughly one quarter of the encounter, and it is the only quarter that outreach advice discusses. Every template you can find optimises the words in the box. None of them mentions that the box is a doorway to three other surfaces that were finalised weeks before you pressed send.

This is not a general truth about outreach. It is specific to writing upward. A department head reading a relevant message may reply without checking you, because the cost of a wasted twenty minutes is low. A chief executive has a calendar that is defended by other people and a reflex, built over years, of establishing who is writing before deciding what to do about it. The inspection is the norm, not the exception, and it is the thing your message has to be built for.

The template farms are solving a problem you do not have

Almost every outreach template in circulation was written for a sales development rep working a list of thousands, where the correct design goal is to minimise the cost of being ignored. At that scale the message genuinely is the whole encounter, because nobody inspects a stranger they will forget in four seconds. A founder with twenty-five named accounts is running a different machine with different arithmetic, and the difference is not stylistic.

Illustrative arithmetic. Every rate below is an assumption you should replace with your own history. The structure of the comparison is the part that holds.

Read the second to last row honestly. On cost per conversation, the volume route wins, and that is exactly why it dominates the published advice. It is also the wrong metric for anybody with a named list, because a conversation with a company you filtered into a spreadsheet is not the same asset as a conversation with one of the twenty-five companies you actually want. Set your own multiple on that difference and the comparison inverts at anything above 1.4 times.

The more important number is hidden in the top row. The volume regime above converts at roughly one percent end to end, from target to reply. Apply that same rate to twenty-five targets and you get a quarter of a reply. A founder who copies a sequence template and sends it to twenty-five people is not running a campaign with a low yield. They are running a campaign that is mathematically incapable of producing anything, and no amount of subject line testing changes that.

The Ra-Aha Inspection Order, and where each message actually dies

Five checkpoints, in the order a recipient encounters them. Each one has a pass condition and a characteristic failure, and a message that dies at checkpoint one never gets credit for being excellent at checkpoint five. Work them in order, because effort spent on the ask is wasted if the headline already classified you.

Price the ask against what the inspection can find

Match your ask to the row that describes what a stranger could actually discover about you in ninety seconds. Asking above your row is the most common reason a well written message gets read and ignored, and it is invisible to the sender because the message itself looks fine.

The ask ladder. Find the highest row your surfaces genuinely support, then ask no higher than the right hand column.

The bottom row is the one people skip because it is slower and requires asking a favour. It is also the only row where somebody else does the vouching, which removes the inspection entirely. If two of your twenty-five targets are reachable that way, they are not part of this campaign and should be handled separately.

The top row deserves defending, because it looks like giving up. A single answerable question is a powerful opening precisely because it is cheap to answer and impossible to answer generically. It also converts the encounter from a pitch into a correspondence, and correspondence is the only format in which the ask can escalate one step at a time.

The swap test, and the three lines it usually deletes

Read every sentence and ask one question: could this be sent to two hundred other people with the name changed? If the answer is yes, the sentence is proving nothing about whether you looked at this company, and proof of looking is the entire currency of the named regime. Delete it, or make it unswappable.

1. The credential line. Whatever you have done is true of you regardless of who receives the message, which means it carries no evidence that you read anything about them. Even with a number attached it is unverifiable from the reader's chair, which is why it does far less work than it feels like it does.
2. The compliment line. Praise for a recent post is swappable by construction. It survives only when it contains a disagreement or a consequence, because both of those require having actually read the thing.
3. The value proposition. It was written once, for everybody, which is the definition of swappable. Save it for the second exchange, where it answers a question they asked rather than one they did not.

What survives the test is narrow: a sentence that could only be written after looking at their company, and a question only they can answer. That is a smaller message than most people want to send and a much harder one to write, which is where the forty minutes per target goes.

One consequence contradicts the standard advice, so it is worth stating plainly. Short is correct for volume, because brevity lowers the cost of being ignored. In the named regime the constraint is different. The message must contain enough specificity to prove that research happened, and specificity has a word cost. Four to six sentences that pass the swap test beat two sentences that do not.

A composite rewrite, annotated against the checkpoints

Both messages below are fictional and written for illustration. The names, the company details and the claims are invented. Copy the structure and never the sentences, because the entire mechanism depends on the specifics being true of you and of them.

The template version

Hi Priya, hope you are doing well. I am the founder of a firm that helps B2B software companies fix their revenue operations. We have worked with dozens of companies and typically see a meaningful lift in pipeline efficiency. Would you be open to a quick fifteen minute chat next week?Fictional example, written to be criticised

Four sentences, and all four are swappable. The greeting proves nothing, the description of the firm is true on every day of the year, the results claim cannot be checked from where she is sitting, and the ask is a fifteen minute call requested at the moment when the evidence available is a title. It fails checkpoint two and then fails checkpoint five for good measure.

The named-target version

Priya, your team shipped the self-serve tier in March and the pricing page still routes enterprise enquiries into the same form as trials. I have watched two companies at your stage lose most of a quarter inside that handoff before anyone noticed it. One question before I say anything about myself: is the shared route deliberate while you gather data, or is it on the list? I have a two page write-up of how the other two untangled it, and I am glad to send it either way, with no call attached.Fictional example, written to survive the inspection

• Sentence one is unswappable. It could not be sent to anyone else, which is the only thing that proves research at this stage.
• Sentence two makes a claim, which is what sends her to the profile. Checkpoint three now has to corroborate it within one screen, or the message has written a cheque the profile cannot cash.
• Sentence three is the ask, and it is priced at the top row of the ladder: one question, answerable in a sentence, no meeting requested.
• Sentence four removes the trade. The artefact goes either way, which is what separates an opening from a transaction and leaves a second exchange available.

The rewrite is longer than the template, and that is deliberate. It is also considerably more expensive to produce, which is the honest reason most people send the first version to two thousand people instead of the second version to twenty-five.

Fix these seven things before you send any of the twenty-five

The pre-flight matters more here than in any other outreach context, because the twenty-five targets are not replaceable. Sending before the surfaces are ready does not lower your response rate. It permanently spends a name from a list of twenty-five, and there is no version of next month in which you get that name back.

The follow-up rule is arithmetic rather than etiquette. With 2,500 targets, burning a contact costs 0.04 percent of the campaign and the sequence templates correctly recommend five touches. With twenty-five targets, one burnt contact costs four percent of everything you have, which is a hundred times more expensive. That single ratio is why volume playbooks and named playbooks give opposite advice about persistence, and why copying the wrong one is so costly.

What to do when they reply, and when they do not

Escalate the ask by exactly one step per exchange. A reply is not a conversion event, it is evidence that the previous ask was correctly priced, and the most common way to waste it is to answer a question with a calendar link. The correspondence is the asset. Protect it for one more exchange than feels necessary.

• They answered your question. Answer back with something useful and do not request a meeting in the same message. The meeting request belongs in the following exchange, where it is now supported by two rounds of evidence rather than none.
• They asked you to send something. Send the four minute artefact, not a deck. Length here reads as effort spent on yourself rather than on them.
• They replied with a polite deferral. Take it literally, ask when the situation would change, and put a date in your own calendar. A deferral from a named target is worth more than a fresh name.
• No reply after ten days. One follow-up that contains new information, never a message whose only content is that you sent a previous message. Then stop, and leave the name intact for a different opening in six months.

There is also the case where the connection is accepted and nothing else happens, which is common enough to plan for rather than to be surprised by. That silence is a distinct problem with its own repair, and the wrong response to it is a pitch, which is why it is handled separately in what to say after they accept.

If the twenty-five list is producing acceptances and no replies at all, the diagnosis is almost never the message. Check checkpoint four first. An accepted request means the headline and top card passed, and a total absence of replies after that usually means the recent activity is either empty or is quietly telling every recipient that selling to people like them is your subject.

Automation, priced honestly for this specific regime

At twenty-five targets, automation removes the only input that makes the regime work. The forty minutes per target is not overhead sitting around the message, it is the mechanism that produces the acceptance and reply rates in the table above. Automating it saves perhaps fifteen hours and converts a named campaign back into a very small volume campaign, which the arithmetic already showed cannot work.

That is a separate argument from the platform risk, which is real and is worth pricing properly rather than waving at. The exposure calculation, and the specific things that turn a warning into a restriction, are set out in how to price automation risk. For a founder with twenty-five names, the simpler point stands on its own: there is nothing to automate, because the expensive part was never the sending.

One closing note on cadence. Twenty-five targets at forty minutes each is under seventeen hours, which is a fortnight of mornings, not a quarter. Spreading it across six weeks is usually better than compressing it, because the earliest replies teach you which of your five checkpoints is actually failing, and you still have twenty names left to apply that to.