Strategy · 2026-06-03 · 12 min

I Send 200 Hand-Written LinkedIn Messages a Month. Here is the Exact B2B System.

Tools are not the problem. Generic messages and scraped lists are. Here is the exact B2B lead generation system I run: 200 hand-written LinkedIn messages a month, with real research behind every one. List, research, message, cadence, numbers.

This Is Not a Fight With Tools

Let me get this out of the way first. A CRM is fine. Apollo is fine. Lemlist is fine. Clay is fine. Sales Navigator is fine. I am not here to tell you to cancel your stack and live in a spreadsheet. Tools do what tools do.

The problem is not the tools. The problem is what most operators do with them.

Most operators use a tool to scrape a list of five thousand contacts, then use another tool to send the same message to all of them with the first name swapped in. That is the part that does not work. Not because the tools are bad, but because the message is generic and the list was never researched. A senior buyer can tell the difference between a message someone wrote for them and a message a system wrote for ten thousand people. They can tell in three seconds. And once they know, you are done with that buyer for at least a year.

So this Dispatch is not about burning your stack. It is about the two rules I treat as non-negotiable, no matter what tools you use:

One. Do not send generic messages.

Two. Do not let a tool build your list for you without research behind every contact on it.

Everything below is the system I built around those two rules. Two hundred hand-written LinkedIn messages a month, sent by one operator, with real research on every single one. Use a CRM to track them if you want. Use Sales Navigator to find them. Use whatever helps. Just do not break the two rules.

Why 200 Researched Messages Beat 20,000 Generic Ones

The instinct, when you read "two hundred", is to assume the low volume is the constraint. It is not. The low volume is the design.

A senior buyer can tell within three seconds whether the message in front of them was written by a person who understood their context, or sprayed by a tool that knew their company name and their job title and nothing else. The tell is the specificity. A real message references something a stranger could not have known without spending fifteen minutes looking. A generic message references the company name and the job title, because that is all the system had access to.

Once the buyer makes that call, everything downstream changes. The acceptance rate on the connection request changes. The reply rate on the first message changes. The tone of the reply changes. And the buyer's willingness to refer you sideways to a peer changes. Generic outreach almost never produces a sideways referral. Specific outreach produces one in roughly every twenty-five conversations.

Two hundred messages, written by hand, with real research, will produce more Qualified Conversations in a month than twenty thousand sprayed sends. I have run both. The numbers are not close.

The Exact System // Five Parts

Part One // Target List Build

The list is built manually, in batches of fifty accounts a week. Not fifty contacts. Fifty companies. The contacts come second.

You can absolutely use Sales Navigator or a similar tool to find the companies. That part is fine. What you do not do is hand the filtered list straight to a sequencer. You read every company.

The filter is narrow on purpose. B2B SaaS companies and custom software agencies between five and fifty employees, in markets where I can hold a real conversation in the buyer's working language, and shipping something I can describe in one sentence after reading their site for ninety seconds. If I cannot describe what they sell in one sentence, they come off the list. A confused list produces confused messages.

Within each account I identify, at most, two decision makers. Usually the founder or CEO, and one operational counterpart. A COO, a Head of Product, a CTO, depending on what the company actually does. Two contacts per account is the ceiling. More than two and you start to look like a vendor working the org chart, which is exactly the signal you do not want to send.

That is fifty accounts, roughly one hundred contacts a week. Four weeks, two hundred contacts, two hundred messages. The math is intentionally tight.

Part Two // Pre-Message Research Protocol

Before I write a single word, I spend roughly seven minutes per contact looking for four specific signals. Not "looking around". Looking for these four things.

One. A recent post, comment, or article from the contact in the last sixty days that reveals something they actually believe, not something their marketing team would have approved. The strongest opener in B2B outbound is one that references an opinion the buyer has stated publicly and that most of their peers would disagree with. It tells them you read carefully, and it flatters the part of them that wrote it.

Two. A concrete operational change at the company in the last quarter. A new hire at the VP level, a product launch, a market expansion, a funding round, a public case study, a rebrand. Something dated, sourced, and verifiable. This is the bridge between "I read your profile" and "I have a reason to be in your inbox today specifically".

Three. A peer or near-peer company that has already solved the problem you suspect they have. Not a competitor. A peer they would recognise and respect. Buyers do not move on abstract logic. They move on the social proof of someone they consider their equal having done the thing first.

Four. The single sentence that describes the gap between where the buyer is and where the peer is. Not a pitch. A diagnosis. If you cannot write that sentence in plain English after seven minutes of research, you do not have enough to send a message, and the contact comes off the list for this cycle.

Four signals. Seven minutes. If any of the four is missing, I do not write. Skipping a contact is free. Sending a weak message costs you that contact for the next twelve months.

This is the part most operators skip. They let a tool build the list, and then they send a message that fits any contact on it. The tool is not the problem. The skipped research is.

Part Three // Message Architecture

Every message I send follows the same three-part structure. The structure is not creative. The content is the part that varies, and the content is what the research feeds.

The opener is a single sentence that references signal one or signal two. It is specific enough that the buyer immediately knows this message could not have been sent to anyone else. No compliments. No "I have been following your work". A real reference to a real thing they said or did, with enough detail that no template could have produced it.

The observation is one or two sentences that connect what they said or did to a pattern I have seen elsewhere in their market. This is where signal three lives. It is the moment the message stops being about them and starts being useful to them. The observation has to be true, and it has to be something they probably suspect but have not heard said out loud.

The micro-ask is the last line. It is not a meeting request. It is not a calendar link. It is a single, low-friction question that respects the fact that they have not agreed to anything yet. "Worth a fifteen minute conversation, or is this not where your head is right now?" That phrasing, or one very close to it, gives the buyer a clean way to say no without losing face, which is exactly why they say yes more often.

Three parts. No images. No PDFs. No links to my calendar in the first message. The message is shorter than a paragraph in a book. It takes me, with the research already done, about four minutes to write.

Part Four // Cadence // The Second Touch Rule

The most damaging thing most operators do after sending a first message is the long follow-up sequence. Three pings. Five pings. The breakup email. The "just bumping this to the top of your inbox" message that everyone learned to hate years ago.

I send one follow-up. Ever. Twelve to fourteen days after the first message, and only if the buyer has done something publicly visible in that window. A post, a comment, a company announcement. Something that gives me a legitimate reason to write again. If they have been quiet, I stay quiet. The follow-up references the new signal, not the original message. It does not start with "circling back". It does not start with "just checking in". It starts the same way the first message did, with a specific reference to a real thing.

If the second touch does not produce a reply, the contact goes into a list I review once a quarter. Not a sequence. A list. The next time I write to them, it will be because something material has changed on their side, and the message will read like a peer noticing, not like a vendor chasing.

Part Five // Tracking

Track this however you like. A CRM works fine for this. A spreadsheet works fine for this. The format is not the point. The discipline is the point.

I personally use a single sheet with eight columns. Account, contact, role, date of first message, signal used, reply or no reply, next review date, notes. If you already pay for a CRM, use it. Just make sure the tool serves the system, not the other way around. The minute you start spending more time entering data than reading buyers, the tool has won and the system has lost.

The sheet (or CRM view) is reviewed every Friday morning. Replies move into an active conversations view. Non-replies get the next-review date set. Anyone who explicitly declined gets archived and is not contacted again. That last rule is non-negotiable. The market is small, and people remember.

The Numbers

Two hundred messages a month, run through the system above, produces the following, averaged across the last twelve months and across multiple engagements in B2B SaaS and dev agencies.

Connection request acceptance, when a request is required, sits between thirty-eight and forty-six percent. Industry benchmarks for generic outreach are between fifteen and twenty-two. The difference is not the request copy. The difference is that the profile the request is coming from is clean, and the request is going to people who have a real reason to recognise the name.

First-message reply rate sits between eleven and fifteen percent. Of those replies, roughly sixty percent are positive or neutral, meaning the conversation continues. The remainder are explicit no's, which I treat as a gift, because a clean no is more valuable than a slow ghost.

Qualified Conversations, defined as a scheduled, attended call with a decision maker who fits the Ideal Client Profile, land at between eighteen and twenty-six per month off two hundred messages. That is a conversion of roughly nine to thirteen percent from sent message to attended call, and that is the number that matters. Every other number on the page is a leading indicator for this one.

Pipeline value generated, in markets where deal sizes sit between fifteen and seventy-five thousand euros annually, lands at between one hundred and eighty thousand and four hundred and twenty thousand euros per quarter, depending on the season and the engagement. Close rates from Qualified Conversation to signed agreement vary by buyer category, but the floor is roughly one in eight, and the ceiling, in the warmest months, is closer to one in four.

What Breaks at 200 a Month

The system is honest about its constraints. Three things break, predictably, and the discipline is in catching them early.

Research fatigue breaks first. Around message one hundred and twenty in a given month, the temptation to skip the seven minutes and write from the profile alone becomes real. The discipline is to either skip the contact for the cycle or to do the research properly. There is no middle option that produces a useful message.

Message templating breaks second. After enough cycles, you notice your own openers starting to rhyme. The same sentence structures. The same hinge words. Buyers in tight markets compare notes, and the worst possible outcome is two contacts at the same company recognising your template. The fix is a monthly audit of the last forty messages sent, looking for repetition, and rewriting the muscle memory before it sets.

Pipeline visibility breaks third. Because the conversations live inside LinkedIn, it is easy, around month three, to lose track of which conversations are real and which are slowly dying. The fix is the Friday morning review, and the willingness to mark a conversation dead when it has not moved in three weeks, instead of pretending it is still warm.

Every one of these failure modes is solvable. None of them is solved by adding more tools.

Why This Will Not Scale to 2,000

The obvious question, once the numbers above are on the table, is whether the system can be scaled. Five operators, a thousand messages a month. Ten operators, two thousand.

It cannot, and the part that cannot scale is not the messages. It is the judgment.

The system works because one operator owns the entire chain. List, research, message, follow-up, conversation, close. Every decision compounds with every prior decision. When the same operator has written to two hundred buyers in a market for twelve months straight, they develop a feel for which signals matter, which messages will land, which conversations will close, and which will quietly waste a quarter. That feel cannot be put in a playbook. It can be modelled, taught, and slowly transferred to a second operator, but it cannot be replicated by hiring six junior people and giving them the template.

When operators try to scale this past one person without spending eighteen months developing the second operator, the system collapses back into generic outreach with a hand-written costume on. The messages get longer. The research gets thinner. The acceptance rate sags. The reply rate halves. The pipeline value per message drops by an order of magnitude. And the senior reputation the original operator spent two years building gets spent down in three months.

The point of two hundred messages a month is not that it is the maximum a person can do. The point is that it is the maximum a person can do well, and "well" is the only setting that produces the numbers above.

What to Do With This

If you are a B2B SaaS founder or a dev agency owner reading this, the takeaway is not "fire your tools tomorrow". Keep your CRM. Keep Sales Navigator. Keep whatever helps you find the right companies faster.

The takeaway is the two rules at the top. Do not send generic messages. Do not let a tool build your list for you without research behind every contact on it. The tools are fine. The shortcuts around those two rules are what kill the pipeline.

The system above is what I run. It is what the SENT Protocol is built around. And it is what produced the engagements documented in the Studies on this site.

If you want to see whether it would produce the same numbers inside your business, the next step is a Strategy Call. Bring the current numbers. Sends, replies, Qualified Conversations, pipeline. I will tell you, honestly, whether this system would move them or whether the bottleneck is somewhere else.