Working Backwards: Cost of Delay From the Plan — Writing

Working Backwards: Reverse-Engineering Cost of Delay (and Dollar-Days) From the Plan You Haven’t Written Yet

2026-06-26

“If you only quantify one thing, quantify the Cost of Delay.”
— Don Reinertsen

Here’s the thing nobody tells you about that quote. Everybody nods at it, opens their backlog, points at the task in front of them, and tries to put a number on it. And they get stuck. Of course they do. Because the dollars they’re reaching for were never in the task. They live one level up — in a plan most people never actually wrote.

So let’s work backwards.

The number you can’t compute (and why)

Cost of Delay is simple to define and brutal to estimate cold. Formally it’s the rate you lose value by not having something — dollars per week. That’s it. A feature, a workout, a blog post, a sales call: each one is quietly bleeding some amount of value every week you don’t ship it, and Cost of Delay is the size of that bleed.

The trouble is that “how much is this task worth per week?” is an impossible question to answer in a vacuum. Stare at a single backlog item and your honest answer is a shrug. So you fall back on what you can see — how urgent it feels, what’s due, who’s shouting. Gut-feel. MoSCoW hell. The loudest-voice-in-the-room school of prioritization.

And here’s the part that should bother you: those decisions get made anyway. Every day, you decide what to touch and what to let slide. You’re already running the math — you’re just running it on hidden assumptions instead of shared ones. (That reframe isn’t mine; it’s Joshua Arnold’s, and it’s the most useful sentence in the whole Cost of Delay literature. Quantifying isn’t rigor for its own sake. It’s dragging a decision you’re already making out into the light.)

So the problem was never the arithmetic. It’s that you tried to start at the bottom. The dollars don’t live at the bottom.

Two lenses on the same truth: Cost of Delay and Dollar-Days

Before we climb, one quick detour, because two communities discovered the same idea from opposite ends and never quite shook hands.

The Lean/product world (Reinertsen, Arnold) gave us Cost of Delay — a rate, dollars per week, looking forward: “what are we losing every week we wait?” Maersk Line, in Arnold’s famous case, had a single feature worth about $200,000 a week sit in queues for 38 weeks. Roughly $8 million gone. Five minutes of estimation up front would have changed a dozen downstream decisions. Thirty-eight weeks of waiting seems crazy, doesn’t it? It isn’t. It’s normal. Map your own value stream and you’ll find the same thing.

The Theory of Constraints world (Goldratt) gave us Dollar-Days — looking backward: Throughput-Dollar-Days = the dollar value of a late commitment × the number of days it’s been late. A $1,000 promise, 60 days overdue, is 60,000 dollar-days of accumulated pain. The best possible score is zero.

Notice they’re the same shape: money × time. Cost of Delay is the speedometer — the rate of loss right now. Dollar-Days is the odometer — the total damage that’s piled up. One tells you how fast you’re bleeding; the other tells you how much blood is on the floor. You want both gauges on the dashboard. And — this matters — Cost of Delay must stay a rate. The $8M Maersk number is the area under the curve, not the Cost of Delay. Mix those up and your priorities come out backwards.

Both gauges need the same fuel, though: a dollar figure. Which brings us back to the climb.

Step 0: the plan everybody skips

If you want a real number at the bottom, you have to start at the top. Not with a backlog. With a plan — and not a thin one. A well-rounded one. The kind PRAXOS is built for: not a to-do list but a description of the whole picture.

What’s actually in your mission? What’s the business you genuinely want — not the one you drifted into, the one you’d build on purpose? What does the life around it have to look like for it to count as a win? What are your goals over the next year, three years, five? What strategies are you betting on, and what beliefs sit underneath them? What are your real constraints — your time, your energy, the one resource you actually run out of?

This feels like the soft part. It isn’t. It’s the part where the dollars are born. Every per-week number you’ll ever compute is a slice of something declared up here. Skip it and you’ll spend forever trying to value tasks against nothing — which is exactly why “quantify Cost of Delay” feels impossible to most people. They’re trying to derive a figure from a plan that doesn’t exist.

So write the plan first. Round, multi-dimensional, honest. (This is the five minutes that saves the eight million.)

Building the financial model — reverse-engineering the cash flows

Now we turn the plan into money. Not a forecast you’ll defend to an investor — a model honest enough to rank things. Here’s the walk.

1. Name the business and the business model. From the plan, what does this thing actually sell, to whom, and how does the money arrive? Subscriptions? Engagements? A cohort? A product? The business model is the machine that turns your effort into cash flows, and you can’t value a task feeding a machine you haven’t named.

2. Map the potential cash flows into four buckets. Arnold’s four are all you need, and they catch the stuff people forget:

Increase Revenue — new sales, new customers, things that grow the top line.
Protect Revenue — defending money you already have from churn, competitors, decay.
Reduce Cost — doing the same for less.
Avoid Cost — the proactive kind: risk, compliance, debt, the bill you don’t get later.

Most people only model the first one and wonder why their priorities feel off. Protected and avoided dollars are real dollars.

3. Size the pool, then cut it down. Your plan implies a total — say a practice aiming at $500–700k a year. That’s the pool. Attribute it across the buckets, then down to the handful of drivers that actually move it. A driver that produces, say, $120k a year of new revenue at a 50% chance of landing is worth $120k × 0.5 = $60k/yr to the model.

4. Convert to a weekly rate. This is the move Arnold makes again and again, and it’s the one that makes everything feel real: divide by 52. That $60k/yr driver is ~$1,150 a week. Now you know what one week of not advancing it costs you. Annual numbers are abstract; “$1,150 a week” makes you sit up.

(Be brave here. An estimate that’s wrong by 2× still ranks your work correctly almost every time, because real portfolios spread 10:1, 50:1. The precision you’re missing wouldn’t change the order. The number you’re missing absolutely would.)

“But half my life doesn’t have a price tag”

This is where most people bail. The revenue work dollarizes fine — but the workout? The hour of learning? The post nobody asked for? Surely those are priceless, which is a polite way of saying zero, which is how they end up last every single day.

They’re not priceless. They’re un-priced, and that’s a choice you can reverse.

A skipped workout isn’t free. A physically inactive US adult runs about $2,025 a year more in healthcare than an active one — call it $5.50 a day, before you count the sick days you don’t take or the years you don’t lose. That’s a Cost of Delay. Put it in the model.
An hour of real learning isn’t free either. Returns to skill compound; the honest range is a few percent of earnings a year, which on a working income is thousands. It’s small per session and enormous per decade — which tells you something important we’ll get to in a second.
A content post isn’t free. This is “surface area for luck” — your odds of a good thing finding you scale with how consistently you’re visible. The data is blunt: consistent creators see multiples of the engagement and opportunity of sporadic ones, and the curve shrinks the longer you go dark. Each post is a small expected-value bet on serendipity, and the bets compound.

Put real numbers — even rough, even editable — on these, and they stop being the things you “should” do and start being line items that compete fairly. That’s the whole game: make the invisible legible so it can win on merit.

Now — finally — Cost of Delay falls out

With a plan up top and a model in the middle, the famous “four steps” become almost easy, because you’re no longer inventing value — you’re reading it off the model:

Value: pull the driver’s contribution from your buckets. Already done.
Urgency profile: ask how the value changes over time. Most things decay slowly and the four steps handle that with a flat rate. But some have a cliff (a launch window, a deadline, a tax date) where all the value is at risk inside a shrinking number of weeks. And some — this is the one people miss — get more urgent the longer you neglect them.
Peak benefit: the most it’s worth if you nail the timing. (Be brave. Again.)
Combine: value shaped by urgency, expressed as $/week. The area between “what it could’ve been” and “what you actually got” is the Delay Cost. The gap is the loss.

And here’s the prize you came for: divide that $/week by the hours of real focus the task costs, and you have a number with a unit you already understand — return per hour of your scarcest resource. Not “4.38.” Something like “$600 per focus-hour.” You know instantly whether that’s hot or cold, because you’ve spent your life calibrated in dollars. No trend required.

Urgency that grows in the dark, and the dollar-days that pile up

Two refinements the plan unlocks, both pure TOC.

First, urgency isn’t only about due dates. Stop running for two weeks and your fitness measurably erodes — the steepest losses land somewhere around days 10 to 35, then plateau. Go twenty days without posting and the luck-curve’s area is visibly shrinking. A task with no deadline at all can still be getting more expensive every day you ignore it. So let urgency rise with days-since-you-last-did-it, not just days-to-due. The thing quietly rotting in the corner deserves a rising number, and now it gets one.

Second, this is where Dollar-Days earns its keep. Cost of Delay tells you the weekly bleed; Dollar-Days tells you the bill that’s accumulated while you looked away. Every day a commitment sits late, or a driver sits neglected, its dollars-times-days climb. You don’t optimize Dollar-Days to some target — like the speedometer, the number only means something in motion. Watch it trend. Rising? You’re falling behind on the things that matter. Falling? The system’s healing.

(Don’t fixate on it as a standalone score. It’s a comparison and a trend, not a grade. That’s true of the whole apparatus, honestly — these numbers are legible relative to each other and to your plan, which is the only kind of legible that helps.)

One more turn of the screw: which horizon are you playing?

The plan asked you for one, three, five years. That wasn’t idle. Time horizon doesn’t just scale the numbers — it reorders them.

A big one-off win dominates a short horizon. But a cheap daily action — the workout, the post, the hour of practice — is an annuity, and its present value climbs steeply the longer you let it compound. Run the math and the small daily thing overtakes the big one-off somewhere around eighteen months. So the honest answer to “what should I do today?” genuinely depends on which game you’re playing. Pick a six-month horizon and the one-offs win. Pick five years and the boring daily compounders quietly take the lead. A tool that lets you turn that dial isn’t a gimmick; it’s surfacing a real fork in your strategy.

Tilting the field

A warning, in Arnold’s spirit: this isn’t a framework you “implement.” It’s not a target state you grind toward. It’s a set of design criteria — things to try, because they tilt the odds.

What you’re really doing by reverse-engineering from the plan is tilting the playing field. The default field is sloped against you: the urgent shouts down the important, the measurable beats the meaningful, and your scarcest hours leak into whatever’s loudest. Start at the top instead — a round plan, a real model, dollars on everything including the “priceless” stuff, urgency that rises in the dark, a horizon you’ve actually chosen — and the slope changes. Your best hours start rolling, on their own, toward the highest return per hour, anchored in the life and the business you actually said you wanted.

Some of this will feel counterintuitive and a little wrong at first. Putting a dollar figure on a workout. Letting a no-deadline task outrank a “real” one. Trusting a number you estimated in thirty seconds while braver than felt comfortable. Sit with it. The discomfort is just the field leveling out.

Reinertsen was right — if you quantify one thing, quantify Cost of Delay. He just left out where the dollars come from. They come from the plan. So write the plan, build the model, and let the numbers fall down the chain to the task in front of you.

It’s five minutes. It’s eight million dollars. It’s the same five minutes either way.

Companion to the DailyFlowScore paper (the per-task scoring mechanics) and the dollar-days sourcing note (the verified figures behind the health/learning/content numbers above). This piece is the why-and-where-the-dollars-come-from; those are the how.