1Sales Team of One

Chapter 8

Chaos in the Pipeline

Post-its

The Post-it note says DALLAS: proposal? and Marcus can't tell what he was supposed to do about it.

It's his handwriting. It's about a mid-size lender whose head of ops had a good conversation with him about servicing transfers. He can't tell whether he sent the proposal or only meant to.

The note has been there for weeks. There are eleven other notes on the monitor. One of them says only TMBA - 45 min.

Forty-three unread emails. Fifteen are live conversations. Three are follow-ups he owed people last week. One is Karen Walsh, asking whether he can move the working session. One is Mike Donovan at First National, who wants the assessment and says he'll circle back in a few weeks. Marcus sends it from his phone, then realizes he can't remember exactly what else he has sent Mike or when. He can't tell whether this is enterprise politeness or the start of something.

The calendar is worse. Discovery calls are stacked back to back. There is a demo for a lender in Colorado. In a few weeks he speaks for forty-five minutes in a Fort Worth ballroom, at the Texas Mortgage Bankers Association conference. He agreed in four minutes and has avoided thinking about it since.

David stops at the desk and looks at the monitor.

"Busy?"

"Busy month."

"That's not what I'm looking at." He nods at the notes. "What's that one?"

"I don't know. That's the problem."

David pulls up a chair. He doesn't usually.

"You don't have a system," he says. "You've got Post-its."

"It's working, though. Four months ago I had nothing."

"Right, and now you have a pile of conversations and a monitor covered in Post-its." David leans back. "Do you remember year one, when you wanted me to ship the officer app in six weeks and I said no?"

"You said we'd build it twice."

"I said no technical debt. Build it right or build it twice." He points at the monitor.

Marcus has no answer.

"I'm dropping things," he says.

"You're going to drop Karen." David says it flatly.

Rachel's second ninety days are running, and she wants one signature.

"So what do I do?"

"Same thing we did with the product." David stands. "Clean it up before you add anything."

Refactor

Marcus calls Tom that night and opens with "I think I'm failing. David says I've got Post-its instead of a system." Tom laughs.

"How many live conversations?"

"Fifteen. Maybe twenty. I don't know exactly."

"And in February?"

"Zero."

"So you've gone from no pipeline to more pipeline than you can hold. That's a good problem to have. It can still sink you. What does David do when the code gets too big to keep in his head?"

"Refactors."

"So refactor. Map what you actually have, then build something to hold it. This weekend, not next month. Which one can't wait?"

"Karen's."

"Right."

Marcus writes Karen's on a Post-it and sticks it to the monitor.

What the Machine Does, What You Do

Split your work into volume and judgment, and give all the volume to the machine.

The machine's half: it logs what happened and remembers what you sent and when. It notices that somebody hasn't replied in nineteen days and drafts the obvious follow-up. It schedules, reminds, transcribes, scores and sequences. Before a call, it pulls up what they told you last time.

An SDR and a coordinator used to do this work. The machine does it now, and that is what lets Marcus keep track of twenty-seven conversations. It doesn't do his writing or his selling.

Yours: you decide which conversations are real, what to say to this person, and when to push, wait or walk away. You also catch the system when it is wrong, because the machine can't.

Stages, so nothing lives in limbo. Don't rely on the stage names that come with a CRM. Write down what has to be true, and what evidence shows it, before a deal can advance. A deal can't move up because you feel good about it.

BANT-Plus. BANT stands for Budget, Authority, Need and Timeline. Add two more, Fit and Trust.

Fit: does this solve their problem, or one next to it? Selling to somebody who buys and then fails costs you the referral, the case study and other deals in their network.

Trust: ask them whether they think you can deliver.

The machine fills in most of the first four from what has already been said. It read the transcript, so it knows whether anyone mentioned a budget number. It can't fill in Fit or Trust.

Which fields earn their place. Ignore the fields the software ships with. Keep the ones that predict your deals. Over time, your won and lost deals tell you which those are. Marcus has only five beta lenders and twenty-seven conversations, so the four fields he picks are best guesses, built mostly from what Karen told him. He wants to know how many loan officers the lender has, whether they can see their own retention number, what they have already bought and abandoned, and who inside the building owns the problem. The retention-number and abandoned-purchase fields are not standard in most CRMs. He expects those two to predict the outcome best, and he will check that against the deals that follow.

Scoring, and how it misleads you. Score your leads on fit, engagement, budget, authority, timeline and how much the problem hurts, so you know where your hours go. Let the machine keep the scores current. It doesn't get tired or optimistic.

You choose the weights, whether you notice or not, and you choose them from the deals you have closed. So the model finds more deals like the ones that already closed. It does this fast and with complete confidence.

It will mark down a large deal that is still early. Early enterprise deals are missing the usual signals (no confirmed budget, no named authority, no stated timeline) for ordinary reasons that look the same as a bad lead. And for a deal unlike any you have closed, the model has nothing to base a score on, but it gives you a number anyway, and that number looks like all the others.

Use the score to sort your week, and read it as a guide, not a verdict. When you override it, write down why, as Marcus does below, and check later whether you were right.

The weekly review. Spend one hour at the same time every week and ask four questions in the same order.

What moved? Which deals changed stage, and on what evidence?

What stalled? Anything untouched for two weeks is stuck, and a stuck deal is usually waiting on a person you haven't found yet.

What is the single next action on each live deal, and who owes it? If the answer is "follow up," you do not have a next action yet. Name the specific thing you will send or ask.

Which of these am I lying to myself about? Most pipelines have at least one deal that is there because you want it to be.

The machine prepares what you need for the first three. It hands you the stage changes, flags everything untouched for two weeks, lists the deals with no scheduled action, and shows you which forecasts you got wrong last quarter and by how much. The fourth question is yours.

The Weekend

Saturday morning, Marcus clears the calendar.

He has been using the free tier of the CRM he signed up for in year one as an address book. He rebuilds it. He adds the four fields he thinks predict a VaultPath deal, stages with entry criteria, and sequences for the people who aren't ready.

Then he enters all of it, every email, message and Post-it, with context. It takes eleven hours across two days and is the dullest work he has done all year. When the last note is entered, he takes the notes off the monitor, all but one, and burns them in the sink. Karen's goes with the rest, since she is in the system now.

By Sunday evening he can see what he actually has.

He finds three real opportunities, five qualified leads, twelve suspects worth nurturing and seven that were never opportunities. Another lender is one of the five. It has been a beta pilot since before any of this, and Marcus has never asked them to convert it. The seven he archives. They had been taking up his attention every day just by sitting on the monitor.

Dallas never got the proposal. He'd drafted it and not sent it. He writes a short note saying so, and the reply on Monday is warmer than he expected.

Then he turns on the scoring because he built it and it's satisfying. He set the weights from his five beta lenders and what Karen told him, which is thin, and he knows it. The model ranks everything in about a second, on a scale of zero to a hundred:¹

Marcus reads that twice.

He set a cutoff of sixty on Saturday for which deals get his hours, and fifty-eight is under it. By that rule he should stop spending real time on the largest company that has ever written to him.

The model isn't wrong on its inputs. Mike has no confirmed budget, no timeline, and an evaluation process nobody has described to him. Every field is soft. Fifty-eight is fair.

What the score can't see is that Mike wrote to him first. There is no field for that, because Marcus never built one.

He doesn't change the score. Instead he adds a column that he fills in himself. He calls it conviction. Beside First National he writes high and two sentences explaining why, so that in November he can see what he thought and whether he was right.

Monday morning the dashboard is clean.

His phone rings.

"Marcus, about the working session," Karen says. "My CEO wants to meet you. Can you do Tuesday at two instead?"

He opens her record. It's all there: every conversation, her own score of six out of twenty, the three questions she couldn't answer, his notes from the first call, and who owns what in her building.

"Yes. What does he care about?"

"Implementation. He believes the problem now. He wants to know whether it's six weeks or six months and whether it breaks anything."

"That I can do."

After she hangs up he logs it, sets the prep task, and notices how little is left on the monitor.

Except one note is still stuck to the bottom of the monitor.

TMBA - 45 min.

He never took this one down. It doesn't fit the new system. It has no stage, score or close date, and by the rules he built this weekend it is not a deal.

He leaves it on the monitor.

Endnotes

  1. Marcus's scoring weights, the scores of 85, 78 and 58, and the pipeline counts are invented for the story. The point they illustrate is not invented: a hand-built scoring model encodes the shape of the deals you have already won, and it will hand that assumption back as a number that looks as authoritative as every other number on the screen.