The Decision Chain

Every business wins the same way. Customers that cost less to win, and are worth more over the time they stay.

You know your version of those two numbers, or close enough. So does every competitor about theirs. Knowing them doesn’t tell you what to do on Monday, because neither number is something anyone does. They’re the sum of hundreds of smaller calls.

That’s why a strategy built only on the two numbers stalls. “Lower acquisition cost” and “raise lifetime value” are goals. Nobody can act on a goal. People act on the decision in front of them.

Every company knows its two numbers. Few can name the decisions that move them.

Follow a customer through the business and you pass through five links. At each one, somebody makes a call that pushes one of the two numbers.

Here’s the chain for an illustrative 150-person engineering services firm, one call per link. A real one has more, but each link usually has one call that matters more than the rest.

Notice what the chain isn’t. It isn’t an org chart, and it isn’t a process map. It doesn’t say how a proposal gets written or how a project gets scheduled. It says where the calls are made that decide whether the business wins.

That’s the right altitude for an executive. You don’t need to know how every process runs. You need to know where the calls get made, and you almost certainly already do.

For each link, three things are worth writing down: who makes the call, what they make it on, and what it moves downstream.

Who makes it. A person, not a department. “Sales decides which bids to chase” hides the fact that it’s whichever principal saw the RFP first.

What they make it on. The information they actually look at, which is rarely what the system was designed to give them. Old proposals. A spreadsheet. Memory. A phone call to whoever was there last time.

What it moves. A fee set too low at Win doesn’t show up at Win. It shows up months later at Deliver, as a project that loses money, and at Keep, as a client relationship strained by the scramble to rescue it.

A decision’s cost usually lands in a different part of the business than the one that made it.

Once the chain is written down, the weak links are usually obvious. They tend to break in one of three ways.

• A handoffThe fee is set at Win without seeing what similar projects actually cost to deliver. That number exists, in timesheets and accounting. It just lives in Deliver.
• Half the pictureStaffing at Deliver is decided on who’s free, without what the principal promised the client at Win.
• Nobody decidesAt Keep, “which client needs a call” is everyone’s job, so in practice it happens when a client complains.

The most common is the first. The information one call needs is produced by another link, by other people, in another system. Each team does its own job well and the chain still fails between them.

This is also why AI added at one desk so often fails to move the business. A Federal Reserve Board staff note put it plainly in July: “A 10% improvement on a task does not necessarily lead to proportional gains for a firm if adjustment costs or other bottlenecks lie elsewhere in the production process.”

Speeding up one link doesn’t move the numbers when the bottleneck is the next one.

Building the intelligence into a decision means giving the person who makes it what they couldn’t see before, at the moment they decide.

It doesn’t mean taking the decision away from them. For the engineering firm, the call worth building into first is the fee at Win. Every past project’s actual hours and margin already exist in timesheets and accounting. The principal setting the fee can’t see them. Put that history in front of the principal when they price the bid: what similar projects actually took, where they overran, which clients pushed scope. The fee is still the principal’s call. It’s just no longer a guess.

Pick the first link with two tests:

Does better information change the outcome? Some calls are made well already. Some go wrong for reasons information won’t fix. Look for the call where people would decide differently if they could see more.

Does that information already exist somewhere in the business? If it does, the work is connecting it. If it doesn’t, the first job is to start capturing it, and that’s a slower project.

The best first link is rarely the most visible one. It’s the one where the payoff is clear and the information is already sitting in another part of the business.

AI doesn’t improve outcomes. It improves decisions, and decisions move outcomes.

Once you know which decision comes first, you know what it needs to see. That tells you what the foundation has to be.

For the fee decision, the list is short and specific. The data: project hours, margins, and the proposal archive, where they already live. The connections: accounting, timesheets and the shared drive, linked to one AI setup the principals already use. Written down: how the firm decides which bids to chase, so the AI knows the firm’s rules and not just its numbers. An owner: the principal accountable for how bids get priced.

The next decision reuses most of it. Staffing at Deliver needs the same project history plus what was promised at Win. Then Keep needs both. Each link you build into makes the next one cheaper. That’s why the foundation comes first when you build, even though it’s the last thing you can work out.

Foundation first is the order you build in. It’s the last thing you can scope.

This is also where the question of AI tools gets a useful answer. The useful question isn’t how many AI accounts the company has. It’s which of them touches a decision on the chain. The ones that do are worth bringing into one setup. The ones that don’t are spend to question.

Everything above is what the first call covers. Ninety minutes, at the level you run the business, drawing your chain together.

No prep and no deck. We won’t go through how each process runs; that comes later, from the people who own each one. These are the questions:

1. Where do you want the business to be in three years, and what has to be true about winning and keeping customers to get there?
2. Which five or six calls most move what a customer costs to win, or what one is worth?
3. Who makes each one, and what do they look at when they do?
4. Where does one call depend on information another part of the business holds?
5. Which call do you wish someone were making that nobody is?

The second call is an hour, a week or so later. I walk you through a written roadmap: which decisions to build into first and in what order, what each one needs to see, what you already have to build on, and the foundation underneath. You keep the roadmap whether or not we go further.

If you’d like to see where you stand first, sketch your own chain below. It’s optional. Half a chain is a perfectly good place to start, and the blanks are often the most useful part.