Framing a big initiative as a hypothesis instead of a project is a genuine advance. But a hypothesis is a contract that requires a counterparty: someone to read the evidence and decide. Write rigorous hypotheses into an organization with no review gate, and you have produced the paperwork of agility with the behaviour of waterfall.

There is a real and valuable shift in how mature organizations frame large investments. Instead of defining a big initiative as a project, “we will build X, by date Y, for cost Z”, they define it as a hypothesis: “we believe X will produce outcome Y for these customers; here is the early signal that will tell us whether we’re right, and here is the outcome we’ll measure to confirm it.” Written well, an epic hypothesis is one of the most honest artefacts in the whole discipline, because it turns a confident bet into something falsifiable and measurable. But it comes with a catch that almost every organization hits and few name, and that catch, not the writing of the hypothesis is the actual subject of this piece.

What an epic hypothesis actually is

The artifact has a recognizable shape. It states, for a defined group of customers who have a particular need, what the solution is and what value it delivers, and how it differs from the current state. To that it adds a benefit hypothesis, the measurable outcome the organization believes the epic will produce. And it adds indicators: leading indicators, the early, in-flight signals that the benefit hypothesis is proving true while the work is still underway; and lagging indicators, the outcome measures that confirm, after the fact, whether the bet paid off.

Done properly, this is a serious act of intellectual honesty. A project plan asserts that a thing will be built. A hypothesis admits that a thing is believed to be worth building, states what would prove or disprove that belief, and commits in advance to the evidence. It replaces “we will deliver” with “we will find out.” That is exactly the posture a transformation is supposed to instill.

The catch: a hypothesis requires a reviewer

Here is the part that gets missed. The entire logic of a hypothesis depends on someone evaluating it. This is not decoration, it is the mechanism.

Leading indicators exist for one reason: so that partway through, someone looks at the early signal and makes a decision: persevere if the benefit is proving out, pivot or stop if it isn’t. Lagging indicators exist so that afterward, someone judges whether the outcome the epic promised actually arrived, and carries that lesson into the next bet. The benefit hypothesis exists so that someone can say, before funding it, “yes, that outcome is worth this investment”, or “no, it isn’t.” Every element of the artifact presumes an accountable counterparty: a forum that receives the hypothesis, funds it or declines it, watches its indicators, and makes the go, no-go, pivot, and persevere calls that the hypothesis was written to enable.

In a scaled framework this counterparty has a name and a shape, an accountable epic owner, a portfolio-level review with real gates, a lean-portfolio forum that decides against the business case, and a minimum viable product built specifically to test the hypothesis before committing to the rest. But the specific mechanism matters less than the principle: a hypothesis without a reviewer is not a hypothesis. It is a wish with a spreadsheet attached.

The failure mode: rigor into a void

Now picture what happens when the artifact exists and the reviewer does not. Someone does the disciplined thing. They write epic hypothesis statements with genuine benefit hypotheses, real leading indicators, real lagging indicators: the textbook-correct, falsifiable, measurable version of the work. And there is no epic owner accountable for any of them, no portfolio review that gates them, no forum that says go or no-go, no one who ever looks at a leading indicator or owns a lagging one.

The hypotheses are orphaned. And the specific pain of this is worth being precise about: they are not rejected. Rejection would be engagement, someone read the bet and pushed back. This is worse. They are never received. The author produced a measurable commitment and offered the organization the other side of it, and no one took the other side. The leading indicators, which only have value if someone is watching them to decide whether to keep going, become inert decoration. The lagging indicators are never measured, because no one owns the outcome. The whole measurement apparatus, the thing that made the hypothesis better than a project plan, sits switched off.

This is demoralizing in a way that is hard to explain to someone who hasn’t lived it. You did the most accountable version of the work, and the accountability had nowhere to land. It is the ownership vacuum showing up at the level of investment: you can author hypotheses all day, but if no one is empowered to review, fund, and judge them, the authoring is a message sent into a room with no one in it.

Why it is so common, and so easy to miss

The reason this happens so often is that the artifacts are convincing. When an organization has a folder full of epic hypothesis statements, complete with benefit hypotheses and indicators, it looks like hypothesis-driven investment is happening. The paperwork of an evidence-based portfolio is all present. What is missing: the review gate, the accountable owner, the willingness to actually decide based on the indicators, is invisible, because absence always is. So the organization believes it has become evidence-driven when in fact it has only added a document type. It writes hypotheses and behaves exactly as it did before: funding what it was always going to fund, killing nothing on the evidence, measuring nothing after the fact.

That is the waterfall-in-disguise version of portfolio management. The hypothesis format is agile; the behavior, decide upfront, never revisit, never measure, is unchanged. The document changed. The decision-making didn’t.

What the review actually is

The fix is not better hypothesis-writing. It is installing the counterparty the hypothesis was always addressed to.

That means naming an accountable owner for each epic, a person who shepherds it, is answerable for its benefit hypothesis, and has standing to bring it for decision. It means a real review gate where epics are funded or declined against their business case, rather than drifting into implementation because no one said no. It means building the minimum needed to test the hypothesis first, and then actually making a decision on the result, rather than treating the MVP as merely the first installment of a foregone conclusion. It means someone actively watching the leading indicators during implementation, with the authority to pivot or stop when they disprove the benefit. And it means someone owning the lagging indicators afterward, so the organization learns whether its bet paid and gets better at betting.

None of this is bureaucracy, and it is worth insisting on that, because it will be called bureaucracy. It is the entire reason to frame an epic as a hypothesis rather than a project in the first place. Without the review, the hypothesis format is pure overhead: more work than a project plan, with none of the benefit.

The real point

Framing epics as hypotheses only pays off if the organization is willing to decide based on them, to kill an attractive epic whose leading indicators are disproving its benefit, and to double down on one that is proving out. That willingness to act on evidence is the actual transformation. The statement format is just the vehicle for it, and a vehicle with no driver goes nowhere.

So the lesson is not “write better epic hypotheses.” It is that a hypothesis is a contract offered to the organization: I will make my bet measurable and falsifiable, if you will commit to reading the evidence and deciding. If no one is empowered to take the other side of that contract, the rigor is wasted and the person who wrote it is left measuring nothing, for no one. Before investing in better hypotheses, make sure there is a reviewer, because the review is where the whole method actually lives.