Twice the Pipeline (And Counting) From the Same Reputation

Every morning a digest lands in my inbox telling me who I should reach out to that day, and why.

A new CMO, a few weeks into the job at a company that fits our ICP almost perfectly. A company that just closed a growth round. Another that opened four marketing roles in a single week. Each one comes with a short research brief and a draft of the first line I might send.

Then I have to go do it.

I was describing this on a call a few weeks ago with a friend who runs another independent consulting firm, and heard myself say something I’ve been chewing on since:

“It can build all the technical stuff, but someone’s still gotta tell it what to do. It’s giving me these daily checklists of things I can do with my network, but I gotta go do that. The system doesn’t have those relationships.”

Why we built it

Like a lot of B2B companies, we were swimming in signals with no real playbook for acting on them. We had the tools. What we didn’t have was a system connecting a reason to reach out with the right person, the right offer and a real next step. First meetings didn’t reliably turn into projects, and follow-up faded once the call ended. We’ve gotten that wrong too.

So we built it with Claude. Strategy first: who we’re for, which signals matter, what we’d offer and what counts as a real opportunity. Then the process. Then the technology to run it.

Lord knows the build was the easy part.

There are four pipeline engines running now

The signal machine. Four families of triggers, each with its own cadence. New marketing leaders, who have roughly 90 to 120 days to assess and declare a plan. Funding, M&A and board changes, which usually turn into a pipeline-coverage question within a couple quarters. Hiring for demand gen, ABM, marketing ops and RevOps, which tells you a company is about to spend on headcount before the system that headcount will run in even exists. And the “AI transformation” titles showing up increasingly across marketing orgs.

The daily CMO invitations. A handful of marketing leaders identified every business day, each one invited into a timely and contextual conversation with no pitch attached.

The research sponsor hunt. We publish industry research every year and sell a sponsorship slot on each study. This one finds the companies whose own marketing already runs on research, then finds the right executive inside them.

The conference machine. Who’s attending, who matters, who I should try to see and what happens to all of it afterward.

They all end in the same place. A to-do list, on my desk, with the guidance and context I need.

What these machines are already really good at

1. Volume, consistency and memory. They harvest, score and route without getting tired or bored, and they don’t drop an account because it fell off the bottom of a spreadsheet. That’s the work robots should do.

2. The research brief, and a first draft of the one line that matters. Each top account gets a short brief: what changed, why it matters to that person right now and how to open. Claude drafts the personalized line. Anything going out under my name to a top account, or to anyone I already know, waits for my approval first. Every time. Most need edits to make them sound like me and that’s fine, the machine still saves me an incredible amount of time.

3. Every touch ends on something useful instead of an ask. Each one is anchored to the clock the signal created, and features a free diagnostic rather than a meeting request. Urgency comes from the signal’s clock, never from discounts.

4. Every reply goes to a human. The sequence stops the moment someone answers, and a person on our team replies, ideally within the hour.

5. Nothing goes quiet. Open, stalled and lost opportunities stay subscribed. When something real changes at one of those accounts, the machine drafts a note that references it. Follow-up gets triggered by a change in the buyer’s world rather than by a reminder sitting in somebody’s calendar.

What the machines can’t touch and cannot do

Trust somebody has to already have.

The research sponsorship work made this one obvious. The machine can build a complete list of every company whose marketing runs on research, find the right executive and get a well-written first note out the door. But nobody puts their logo on a study because a well-written invitation arrived. That decision gets made by somebody who has seen our research before, who trusts our brand or who trusts somebody else that already does.

Our biggest source of new business is still past executive sponsors. The person, for example, who’s had four CMO jobs over the last 18 years and still trusts us. The machine can tell me the day she starts a new role, and that’s useful. Whether she’s glad to hear from me was decided years ago, by the work we did together as well as the value we continued to deliver after the project ended.

Personalization a machine can’t fake.

Claude can find the funding announcement and write a sharp line about it in seconds. That’s table stakes now, because every vendor in her inbox has the same tools and the same press release. What it doesn’t know is what we talked about over coffee a few years ago, which of her peers she trusts or how she likes to be approached.

The research gets a note to relevant. Everything past relevant is on me.

Knowing when the list is lying to you.

On one pass through our sponsor list, a surprising number of the contact records came back wrong. Not broken records. People who had changed jobs, where nobody told the database and the machine kept right on recommending them.

The data decays for the same reason the relationships matter. People move. And the only way I knew they had moved was that I know them, or know somebody who does.

A list nobody recognizes will still arrive on time, in the right format, with the wrong names on it. Relationships are what keep the list true.

Showing up, person to person.

The digest tells me who to reach out to. It can’t take the meeting, read the room or notice what a CMO doesn’t say on a first call. My part is about 45 minutes a day: read the digest, approve what’s queued, write personal notes, answer every reply the same day and spend 15 minutes commenting on what the people I hope to work with are posting.

Every first meeting is supposed to end on a real outcome, with a recap out inside a day. “Let’s stay in touch” doesn’t count. The machine can hold me to that rule and draft the recap. It can’t be the reason the next step exists.

Reputation and brand.

A cold note from me gets opened at a higher rate than is normal for most campaigns we run for clients. And the reason has nothing to do with the quality or complexity of our workflows. It’s from years of writing in public, podcasts and communities, showing up on stages and at events. As I put it to my friend a few weeks ago, you can maintain that through digital channels but you build it by being in front of people.

We wrote one guardrail into all of this that is arguably more important than any other: the CMO Coffee Talk community stays in the giving direction. Nobody gets a follow-up because they were in the room. The community exists to help its members, and the moment it starts to feel like a lead source it stops being worth showing up for.

Where the opportunities have actually come from

All of these digital and semi-automated sequences are great. And I’m drawing a conclusion off a few weeks of running them, so hold the size of it loosely.

Every opportunity we’ve created this quarter traces back to reputation. Mine and the firm’s. Work published over the past 18 years, podcasts and communities, clients who already knew what we do and executives who trusted us somewhere else first.

About half of it arrived on its own, through inbound and repeat engagements with people who came looking. That half would most likely have shown up with or without a single sequence running.

The machine produced the other half, and that is the whole case for building one. Same reputation, applied with consistency and scale. Every signal caught, every account scored, nothing dropped out the bottom and no follow-up left to whether I remembered.

So far that has doubled the natural yield. And the machine has been running for weeks rather than quarters, which makes 2X a floor. I’d expect the impact to compound.

The upkeep on all of these systems is real too. Every rule, list and prompt needs tending, and none of it holds up on its own.

The very first build item was fixing contact records that arrived without LinkedIn URLs, which is about as unsexy as it gets (and was the hidden reason our LinkedIn runs kept coming back with zero sends).

Every morning the machines hand me a list. Then, as I told my friend, I’ve got to go do that, because they don’t have those relationships.

Neither one works alone

Relationships and trust without systems leaves too much on the table. Too much to chance, too much to the hope that the phone rings enough times each quarter to get you to your number. The relationships are real and the pipeline still comes out lumpy.

Systems without relationships and trust is a commodity. Every competitor in your category can buy the same machines, point them at the same signals and write from the same press releases. Everyone has access to that now.

Which leaves the one thing that’s really hard to copy. How you invest in human-to-human trust is your moat, and it’s also the only thing a machine has to work with. For a lot of us that’s getting a fraction of the time and attention it deserves.

Build the machines. Then spend the time they give you back in front of people.

This post originally appeared on Matt Heinz’s Substack.