What a fractional CPO does when the roadmap is guesswork

The roadmap changes on a Thursday afternoon, after one customer call that goes badly, and the only person with the standing to say no is on another call. A fractional CPO owns what gets built, in what order and why, part-time, and holds one gate: what will a customer do differently the week after this ships.

You need it when delivery is healthy and the business is not. Sprints close, releases go out, the demo passes, and the revenue line reads the same as it did in January. The roadmap has become a list of customer requests reordered by whoever escalated most recently, and you are the only person senior enough to arbitrate it.

What does a fractional CPO do?

The seat owns the what. Engineering owns the how.

Deciding what goes in, and what gets refused. Sequencing the survivors against commercial reality: the renewal in March, the deal that stalls without an integration. Sitting in the room where customers say what they actually want, rather than reading the summary a fortnight later. Then closing the loop, by checking whether what you shipped changed anyone’s behaviour.

In practice that is one standing session where the order gets set and the reason written beside each item, plus presence on the customer calls that would change that order. Two months in, you should be able to name three things that came off the roadmap and who was told.

The vacancy is the order of the work itself, and somebody senior has to decide it. Not a project manager: delivery predictability already has an owner. Not a business analyst, who documents decisions somebody else has already made, and not a second engineering manager.

Who is holding the gate on your roadmap?

The symptom is quiet. Nothing is broken. Every engineer, every account manager, every delivery lead is doing the job they were hired for. The only number that moved last quarter was the count of tickets closed.

Left alone for a quarter, a roadmap only accumulates. Every customer conversation adds an item. Every escalation moves one up. Nothing comes off, because taking something off means telling a named person their thing is not happening.

So refusal routes to the one person who can absorb the political cost. You, on a Friday, between two other jobs, in the gaps nobody scheduled. What survives that is not a roadmap. It is guesswork with dates on it.

Fractional CPO or fractional CTO: which seat is missing?

The CTO seat owns how the software gets built and whether the function can build it: architecture, delivery, hiring, the technical risk the board should hear about. The CPO seat owns what gets built, and in what order.

If the team ships late and nobody can tell whether the estimates are real, the missing seat is technical, and what the CTO seat covers, and when you need it is the page for that. If it ships reliably and nothing moves commercially, the missing seat is product.

In businesses of 10 to 50 people I usually find both wrong at once, which is exactly when hiring two makes it worse. What one part-time seat changes once you already have a team is the sizing question underneath. Earlier than that, the prior question is whether you need a technical seat yet at all.

How to tell the missing role is product, not engineering

1. Delivery is predictable and revenue is flat.
2. The roadmap changes after the loudest customer call of the week.
3. Nobody can say what the last three releases changed for a customer.
4. Engineering asks why, and the honest answer is that someone asked for it.
5. Nothing has come off the roadmap this quarter.

When several of those are true at once, I stop looking at engineering capacity. More developers move the queue faster in a direction nobody chose.

Before you hire anyone, take the next three items on the roadmap and write one sentence each on what a customer will do differently the week after it ships. The ones you cannot write are the ones to cut. That argument is the job the seat does every week.

Do you hire two people or one?

Between 10 and 50 people, one. A fractional CPO who has never held the engineering seat is half an answer here, and so is the reverse.

Two seats split the what from the how, which puts a handover between the decision and the delivery. At this size the handover is the expensive part. One person writes down what to build, another works out whether it can be built and what it costs the rest of the roadmap. The trade gets negotiated across meetings instead of settled in one head.

Hold both, and the sequencing call and the buildability call happen in one conversation, with one accountable answer. I have found the fastest route to a sellable product is putting whoever owns the roadmap where customers decide whether to buy.

One engagement ran exactly that shape. Joining the founder’s sales calls, distilling what came out of them into the roadmap, then shipping a sellable MVP inside six months. The same person decided what to build and wrote the code. Four years running my own product taught me the other half: an item’s real price is the build plus everything it displaces.

In the engagements I run, every roadmap call carries a name against it and a reason on the record. Nobody settles the same argument twice. Speed inside the team reaches the business only when somebody decides where it goes, which is one of the four decisions that turn faster engineering into business growth.

Once there are several product lines and more than one engineering team, usually past 50 or 60 people, product and engineering need separate owners with separate mandates. Below that, splitting them buys coordination where you wanted clarity.

Buying the seat changes nothing if you keep reopening the decisions it makes, because that turns it into one more approval step. The seat works only when the refusals stick.

Where to start when the seat is unnamed

You are not choosing whether to hire a product person. You are choosing which of the two seats is empty, and whether one person can hold both. Nothing breaks while you decide. Which is why it waits.

What it costs is another quarter of shipping on time into no commercial movement, with the arbitration landing on your desk every week, unpaid and unscheduled.

Naming the seat is the job of a two-week written read: how a two-week read names the seat you are missing, the Technical/Strategic Review. It reads the ordering on your roadmap, what your recent releases changed for a customer, and the decisions that finished on your desk. When the judgement is already in the building and needs backing rather than replacement, it says so. You keep the written verdict either way.

If the roadmap keeps getting re-decided on your desk, book the exploratory call.