Naming one bottleneck and putting a value on it

You have a map and a friction list. Most of what's on them is real, and most of it doesn't limit the aim. This lesson narrows everything down to one step: the bottleneck for this loop. You'll say what kind it is, name the person who leads it, and put a value on it that a finance director would accept.

One bottleneck, not a shortlist

The test is a single question. If this step moved, would more reach the customer? Only one step passes it at a time, because the work can only move as fast as its slowest step. Fix two things at once and you can't tell which one moved the aim, so your prediction can't be scored.

The trouble is that the bottleneck is rarely the most visible candidate. People reach for the work they can see, measure easily, or most dislike doing.

So pick with the map and the list open in front of you. Look for the step where work piles up, where the steps after it wait, and where friction clusters on the list. Write down why it's this one and not the others: that sentence goes on your prediction sheet, and the leadership team confirms it.

Three kinds of bottleneck

The Theory of Constraints sorts bottlenecks into a few kinds. Three matter here, and the kind tells you where the fix will be found.

• Capacity. Too few hands for the work arriving. Every quote waits for one estimator. The fix is to get more from that step: take other work off it, never let it sit idle, never feed it bad work.
• Policy. A rule or habit holds the work, not a lack of effort. Discounts wait for Thursday's leadership meeting; invoices wait for a monthly batch. The fix is to change the rule, and it's often free.
• Market. Every step could handle more than arrives. The limit is demand, price or reach. Faster work inside changes nothing; the fix is in sales, pricing or the offer, and new offers run as experiments with kill criteria.

Policy bottlenecks are easy to mistake for capacity ones, because both produce a pile. The difference shows when you ask what the work is waiting for. If it's waiting for someone to have time, it's capacity. If it's waiting for a meeting, a batch, a signature or a form, it's policy.

Give it a lead

A bottleneck lead is the one person who answers for the bottleneck this loop. They map it, value it, write down the prediction, and check the handoffs between teams: every two weeks during the first loop, then monthly. The bottleneck often sits between teams, which is why it needs a named person. Without one, a handoff belongs to nobody.

Pick someone with a few hours a week and the standing to change how the work is done at that step. If the bottleneck is a leader's own sign-off, pick someone else to lead it, with the sponsor's backing.

Value it in the aim's unit, never in hours saved

This is where most AI business cases go wrong. They add up hours saved, multiply by a salary, and call it value. But you saw in module 1 that hours saved away from the bottleneck are a mirage, and hours saved at the bottleneck only count once they turn into more output.

So value the bottleneck by the extra output the whole organisation gets when it moves, in the unit your aim uses. For a quoting team, that's quotes out a week, and then what those quotes are worth:

• two more quotes a week,
• times the historic win rate of 30%,
• times the average deal of $40,000,
• is about $24,000 a week. (Illustrative.)

That number has three things an hours figure doesn't. It's in the same unit as the aim, so day 90 can check it. It's honest about conversion, because not every quote wins. And it puts the value where it lands: in what the organisation sells, not in anyone's diary. Speed at a step that isn't the bottleneck is valued at zero until it's converted.

Not every aim has a price. If yours is a faster month-end close, the unit might be days to close, and that's enough on its own. What matters is that the value is counted in what the aim measures, so the prediction you write next can be checked against it.