Yesterday I wrote on LinkedIn about coming back after two months away, and what had kept running while I was gone. The capture side held. The content stopped. The conversations slowed down
There was a fourth thing I left out, because it needed more room than a LinkedIn post gives.
Two conversations started during those two months. Both of them came from an old post someone happened to find, not from anything I did.
That is the part worth sitting with. The list grew a little, the open rates were fine, and the only two conversations that happened were accidents. Nothing in my setup caused them, and nothing in my setup would have made them happen more often.
That is not a Lead Letters problem. I see it in almost every founder-led setup I look at, including ones with much better numbers than mine.
The model most of us use is missing a stage
Attention.→ Capture.→ Nurture.→ Sale.
Attention brings people in. A lead magnet captures the email. A newsletter nurtures them. Eventually, some of them buy.
The problem is that word "eventually".
Between nurture and sale there is a step almost nobody designs, and it is the one where a subscriber becomes a conversation. Most systems leave it to chance. The hope is that if you write well enough for long enough, someone will reply, or click something in the footer, or find the booking link on your site.
Sometimes they do. But that is not a step in a system. That is a lottery running quietly at the end of a lot of work.
Five stages, not four
Attention. → Capture. → Nurture. → Conversation. → Customer.
The fourth one is a joint, and joints are where things come apart. Everything before it is content work. Everything after it is sales work. Most of us are competent on one side and treat the joint as something that happens on its own.
Designing it means you can answer two questions at any moment.
Who on my list has shown they are in the situation I fix?
And what happens to those people next?
If the answer to the second question is "they get Thursday's issue like everybody else", then the joint is not there.
A version that runs alongside a job
This is the one I use, because it has to survive rotating shifts.
One: pick a behavior that means something. Not opens. Something with intent behind it. Clicking one specific link. Replying to an issue. Downloading a second asset after the first one. Landing on the page about the exact problem you solve.
Two: make that behavior visible. A tag, a segment, a saved filter, whatever your platform calls it. The point is that these people stop being anonymous inside a list of a thousand.
Three: write one email only they receive. Not a broadcast. Something short that references what they did, offers one specific thing, and asks one question. One, not three.
Four: reply personally when they answer. This is the part you never automate. It is also the entire reason the other three steps exist.
That is the whole thing. Four steps, and roughly an hour to set up once.
Where automation earns its place, and where it does not
Steps two and three are automation's job. Tagging on behavior, triggering a sequence, moving someone onto a different track without you watching it happen. This is what your email platform was built for, and most of us use maybe a tenth of it.
AI is useful in a narrower place than the marketing suggests.
It is not good at writing the email in step three. The whole value of that email is that it is not generic, and a model writing to a stranger produces something that reads exactly like it was written to a stranger.
It is good at the reading job. Connected to your email platform and your LinkedIn activity, it can tell you which subscribers match the situation you sell into, which posts brought in people who are still paying attention, and which fifteen names are worth a look this week.
Then you write the fifteen emails yourself. Fifteen is not a lot. That is one Saturday morning.
This is the split I keep coming back to. Machines do the reading and the routing. You do the talking. The mistake is doing it the other way around, which is quietly what most AI content advice recommends.
One thing to do this week
Before you write another issue, pick one behavior that would tell you a subscriber is close, and set up a way to see who does it.
You do not need a new funnel. You need one filter, and one email that only fires for the people who pass through it.
What is next here
I am rebuilding this for myself before I build it for anyone else, and the next few issues will be that build in the open, including the parts that do not work.
If you reply with the behavior you would count as your signal, I will tell you whether I think it is a real one. That is a useful five minutes for both of us.
Founder, LeadLetters
AI-Supported Inbound & Outbound Systems for B2B SaaS
Newsletters, Lead Magnets & Automation → Qualified Conversations