Prioritisation Under Real Constraints
Choosing between two defensible options, naming what the chosen path costs, and surviving the choice organisationally. Runway decides which arguments are even admissible.
Chapter 4 · 4 ideas · 13 checks · 18 min
Start here
Two things are on the table and you can fund one.
The first is a rebuild of the onboarding flow. Three of your last five enterprise losses mentioned it. Your head of sales wants it and has wanted it for a year.
The second is a set of integrations two large existing customers have asked for by name. Both are up for renewal in the spring. Together they are 19% of revenue.
Both are good. Neither is a mistake. Every argument you can make for one, someone competent can make against — and they will, in the meeting, and they will be right about the parts they are right about.
The chapters before this one gave you tools for telling a good proposal from a bad one. This is different. Here both proposals are good, and the previous tools return yes on both.
Notice what people actually do in this room. They argue about which option is better, for an hour, and the argument does not converge, because it cannot: the options are not comparable on a single scale. Eventually someone senior decides on feel and the meeting ends with everyone slightly unsatisfied.
The question this chapter answers: when both options are defensible, what decides — and what does a good decision look like on paper afterwards?
Core concepts
Each idea is a short read, one picture and one quick check.
Idea 1 of 4
Compute the constraint before you argue about the choice
Before the onboarding-versus-integrations argument is worth having, there is a number nobody in the room has said out loud.
Given current revenue, its growth rate and current costs, and assuming no new funding, does the company reach profitability before the money runs out? If yes, it is default alive. If no, it is default dead. This is not a mood. It is arithmetic on four numbers you already have: monthly revenue, its growth rate, monthly costs and cash in the bank. The months of cash left at the current burn (how much more you spend each month than you earn) are your runway.
It goes first because it decides which arguments are admissible. A default-dead company does not get to make bets whose payoff lands after the money runs out, however attractive they are. Once the answer is on the table, an argument about vision becomes an argument about constraints, and constraint arguments converge.
Picture it
Q3 plan · summary page
- Four new hires and two new initiatives.
- Revenue £120k a month, growing 5% a month. (highlighted)← you have this
- Costs £160k a month; £195k with the four hires. (highlighted)← and this
- £900k in the bank. (highlighted)← and this
- A raise is expected in Q1. (struck out)a story, not a sum
- Runway and break-even: not stated. (circled)the line to add
Every number you need is on the page.
Check yourself
A company that is default dead, with nine months of cash, must pick one of the two bets. Which argument is admissible?
Go deeper
How to use it. Compute it before the debate, not during. State it at the top of the plan, and scope the plan to it. The characteristic error is discovering the answer late, when every remaining option is bad, and the reason it is discovered late is that the calculation forecloses things the team is emotionally committed to.
How to spot it. A resourcing plan with new headcount, multi-quarter payback and no runway figure anywhere. There is always a plausible story about the next raise or the next big customer, and the story is doing the work the arithmetic should be doing.
One thing that has changed. Lower build costs shift the number and rarely reverse the conclusion. Headcount, distribution and sales remain the dominant costs for most companies, and cheaper building tends to be spent on more surface area rather than banked. "AI makes us more efficient" is a hypothesis with a measurable answer, not runway.
Idea 2 of 4
Comparison is what makes a tradeoff visible
You have read this document many times. It recommends one course of action. It is well argued. Its risks section lists three things that could go wrong in execution: the timeline might slip, a vendor might be late, a hire might take longer than expected.
Read it again and ask what else was considered. Usually: nothing that made it to the page.
A single-option proposal is a decision that already happened. The document is not making a choice, it is transmitting one. That is fine as long as everyone knows which is going on. The trouble is that it looks identical to a decision document, so readers evaluate it as though there were something to decide.
The fix is real alternatives, each with its cost, including doing nothing. Three is a reasonable floor. Watch for the fake set: two obviously worse options flanking the intended answer, which is advocacy wearing a comparison's clothes.
Picture it
A real set
- Onboarding rebuild · cost: the renewals go at risk
- Integrations · cost: win rates stay flat for a year
- Do nothing · cost: both problems grow a quarter
A fake set
- Integrations (the one we want)
- Hire ten engineers and do both
- Do nothing and lose both customers
Check yourself
A memo recommends moving every customer to annual billing. Which version of its options section shows that something was actually compared?
Go deeper
How to use it. Require the real alternatives with their costs, including the option of doing nothing. If the author is genuinely confident, the comparison costs them nothing. It gives readers something specific to disagree with, which is how disagreement gets surfaced early instead of six weeks in.
How to spot it. The risks section lists execution risks only. No belief risks: nothing of the form "this only works if X is true, and we do not know that X is true."
One thing that has changed. Generating the alternative branches is now nearly free, which removes the last excuse for a single-option proposal. It also floods the comparison with plausible options nobody has evidence for. Three generated alternatives are not three considered alternatives; the work moved from producing options to pruning them.
Idea 3 of 4
Say what the chosen option costs
Every real choice sacrifices something. This is not a moral point, it is a structural one: if an option cost nothing, it would not be a choice.
So a proposal in which every consequence is a benefit (faster and cheaper and better for users and strategically superior) has either not found the cost or is declining to name it. From the reader's side those are the same problem, because either way the reader cannot evaluate it. This is the costless tradeoff.
The fix is a named cost: what the option sacrifices and who bears it, written in the document in the author's own voice. It is the highest-signal line in most product documents, in both directions. Its presence is the strongest evidence of real thinking; its absence is the most reliable tell of shallow thinking.
Picture it
Choosing the integrations · what we give up
- Enterprise win rates stay where they are for a year. (highlighted)← the segment not served
- Sales is right about this, in public, for four quarters. (highlighted)← who bears it
- The onboarding rebuild waits until next year. (highlighted)← work deferred
- Some short-term disruption is expected. (struck out)names nothing
Each says what, who or how long.
Check yourself
Each line is from a proposal's downsides section. Does it name a cost?
Two engineers for six weeks, so the reporting project moves to Q4.
The timeline may slip if the vendor is late.
A third large customer who wanted the same thing is told no this quarter.
Faster for users and cheaper to run.
After the first month, switching back gets expensive.
It will be a big change for the team.
Go deeper
How to use it. Ask what the option sacrifices and who bears it. Look specifically for: work deferred and by how long, the segment not served, the complexity added, the reversibility lost, the person or team who now has a worse quarter. Then write it down, in the document, in the author's own voice. A named cost is what turns advocacy into a decision.
How to spot it. Search for the sentence beginning "what we give up." If it is not there in some form, it has not been thought about, because thinking about it is not the kind of thing you do and then forget to mention.
One thing that has changed. Fluent drafts are especially prone to costless tradeoffs, because the register of confident business writing has no natural place for "and this will hurt." The cost line is one of the small number of things you have to add on purpose.
Idea 4 of 4
Find the load-bearing assumption before writing the plan
Plans allocate money and people against assumptions. Most plans never say which ones.
Before writing the plan, list what must be true for it to succeed, and mark each item evidenced or assumed. Then find the item that is both load-bearing and unevidenced. That is not a footnote to the plan. It is the first thing the plan should do, with the smallest test that could change your mind.
The plan then either gets evidence or gets cancelled early, and early cancellation is the cheapest possible outcome. It also gives the team something concrete to go and find, instead of a quarter of building.
Picture it
Integrations plan · what must be true
- The two customers are 19% of revenue. (highlighted)← evidenced: billing data
- The suite takes one quarter to build.an estimate, fair enough
- Each customer needs the full suite. (circled)assumed
- Both renewals turn on getting it. (circled)assumed, and load-bearing
Only one line has evidence behind it.
Check yourself
"Both renewals turn on getting the integrations" is the load-bearing assumption. Which is the smallest test that could change the plan?
Go deeper
How to use it. Write the list. Mark each item evidenced or assumed. Find the one that is both load-bearing and unevidenced: that is this week's work. Then check that the test could actually come back "no".
How to spot it. Same tell as the single-option proposal, and it recurs because it is the tell: a risks section made of execution risks only.
Back to the two options. With these four ideas, the room from Start here is a different room. Compute first: if the runway is nine months and the onboarding rebuild pays back in twelve, the argument is over before it starts, and it was never really an argument about product. Compare properly: the rebuild, the integrations, and at least one option neither side proposed. Name the cost: choosing integrations means enterprise win rates stay where they are for another year, and the head of sales is right about that, in public, for four quarters. That sentence belongs in the document, not in the meeting. Test the load-bearing line: ask both customers what their renewals turn on before the quarter is committed.
One thing that has changed. Cheap building makes "just build it and see" look rational, and for small bets it often is. What matters is whether the learning is cheap, not whether the building is. A fortnight's build that cannot answer the question the plan rests on is still a wasted fortnight.
Worked examples
Make the call yourself first. The reasoning opens after you choose.
Example 1 of 3
The plan that got smaller and shipped
Tomas runs a 24-person company selling compliance software. Revenue is £310k a month, growing 4% a month. Costs are £395k a month and there is £2.1m in the bank. A raise is "expected in Q1." The Q3 plan proposes two major initiatives, a second product line and a launch in Germany, which together would take costs to about £465k a month. Everyone who will be in the planning meeting thinks both initiatives are good.
Check yourself
What should Tomas bring into the planning meeting first?
Example 2 of 3
The costless tradeoff, made honest
A platform team proposes migrating to a new job-processing system. The memo is strong: more reliable, cheaper to run, better observability, easier hiring, and the migration is "mostly mechanical." Every consequence in it is a benefit. The migration may well be the right call. The reviewer gets one question before signing off.
Check yourself
Which question should the reviewer ask?
Example 3 of 3
The third option nobody proposed
Back to the two bets from the start of the chapter: the onboarding rebuild versus the integrations for two renewing customers. Each customer's request lists the full integration suite, and each email also says which one data flow their team relies on most. The room has been arguing for forty minutes. Each side is right about its own half, and nobody has moved.
Check yourself
Which question is most likely to end the argument well?
Case studies
Real situations with more than one defensible reading.
Case 1 of 2
A resourcing decision under a closing runway
An invented composite, built from the pattern rather than any real company.
A B2B analytics company: 31 people, four years old. Revenue is £480k a month, growing 3% a month; costs are £610k a month; there is £0.8m in the bank. A Series B has been "in early conversations" for two months. Three demands compete for the next two quarters, each with a competent advocate who is right about their own area. The platform rebuild (the CTO): the data pipeline is at its limit, and every month of features built on top makes the rebuild more expensive; nine months, most of engineering. The enterprise tier (the CRO): four prospects need single sign-on, audit logs and a security review, worth about £70k a month combined; three months. The retention problem (the head of product): mid-market cohorts (customers grouped by the month they joined) lose customers steadily from month four, nobody knows why, and discovery would take a quarter. On today's numbers the company is default alive, just: at 3% growth revenue crosses costs in about nine months, using about £610k of the £0.8m.
Check yourself
What should the company decide first?
Case 2 of 2
The large customer's roadmap demand
An invented composite.
A workflow product with £8m annual revenue. The largest customer, at £900k a year and up for renewal in five months, has asked for a capability that would take a quarter: a permissions model matching their internal org structure, which is unusual and, as far as anyone can tell, unique to them. The account team says the renewal is at risk; that view comes from one conversation with one champion. Product says it is bespoke work that will benefit nobody else and complicate the permissions system permanently. Both are right. The company is modestly profitable, with about 14 months of costs in the bank, so this is not existential either way.
Check yourself
What should happen before a quarter is committed?
What great operators do
The rules this chapter's scenarios test for, and how to tell when someone follows them.
Rules to keep
Before you open each card, say what you'd look for.
Common failure patterns
How smart people get this wrong. Find each one in the wild.
Trap 1 of 3
The single-option proposal
What it looks like. A memo recommending one course of action, with a risks section listing only execution risks. Alternatives are absent, or present as two obviously worse options flanking the intended answer.
Check yourself
Tap the lines that show the decision was made before the memo.
Proposal — rebuild the reporting engine
Found 0 of 2.
Why smart people do it. Presenting options reads as indecision, and the author usually did decide, weeks ago, on instinct that may well be right. The document is written to transmit a decision, not to make one.
The correction. Require the real alternatives with their costs, including doing nothing. If the author is confident, the comparison costs them nothing and gives readers something specific to disagree with.
Trap 2 of 3
The costless tradeoff
What it looks like. A recommendation where every consequence is a benefit. The new approach is faster and cheaper and better for users and strategically superior. Nothing is given up.
Check yourself
Tap the lines that claim the change costs nothing.
Memo — move to usage-based pricing
Found 0 of 2.
Why smart people do it. Naming a cost invites attack on that cost. Advocacy culture punishes candour about downsides, especially in writing that will be read by people who fund things, so the incentive runs against the honest version at exactly the moment it matters most.
The correction. Ask what the option sacrifices and who bears it. A response that cannot name the cost has not understood the option well enough to choose it.
Trap 3 of 3
Runway denial
What it looks like. Ambitious multi-quarter plans, new headcount and long payback bets at a company whose default-alive calculation has not been done, or was done and quietly ignored.
Check yourself
Tap the lines where a story stands in for the arithmetic.
2027 plan — headline slide
Found 0 of 2.
Why smart people do it. The calculation forecloses options the team is emotionally committed to, and there is always a plausible story about the next raise or the next big customer. Both of those things are usually true, which is what makes the story so easy to keep believing.
The correction. Compute it, state it at the top of the plan, and scope to it. The plan that respects the constraint is usually smaller and much more likely to happen.
Cheat sheet
The whole chapter on one card. Save it or share it.
Prioritisation Under Real Constraints
- 1Work out default alive or dead before you debate.Revenue, growth, costs and cash. Assume no new money. Then argue.
- 2Runway decides which arguments are admissible.A bet that pays back after the money runs out can't be weighed at all.
- 3Three real options, including doing nothing.Each with a cost someone could choose to live with. No strawmen.
- 4Write down what the chosen option costs, and who pays.Deferred work, segment not served, complexity, reversibility lost.
- 5Find the assumption the plan stands on. Test it first.The smallest test that could change your mind, before the quarter starts.
- 6Trace both options up to the need behind them.That is where the third option, the one nobody proposed, shows up.
Memes
The failure patterns, one picture each. For the group chat.
Your turn
Make the call
Each checkpoint below hands you a real plan or proposal and asks you to commit: which bet gets the team, how much of a plan to sign, or what a memo must say before anyone can weigh it. Do the runway sum where the numbers allow it, name what your choice gives up, and look for the line the whole plan stands on.
The trap. Splitting one team across two bets that each need all of it feels like the grown-up answer. It is usually the costless tradeoff in disguise: neither half finishes, and the cost of choosing nothing goes unnamed. Cutting a plan down to what the runway supports is different. That is a choice, and it names its cost.


