Running everything in parallel is not the same as moving fast.

The problem was never effort.

When I took over the area I spent the first stretch listening to the team and following the flow of work rather than changing it. Nobody was idle. Initiatives were running side by side, each one defensible on its own.

What was missing was a shared way to decide which of them deserved the team’s capacity. Without that, every new request became another parallel track, and the area was always one commitment away from overload. A team in that state does not fail loudly. It just gets slower at everything at once.

I institutionalized RICE so priority stopped being an argument.

Reach, Impact, Confidence, Effort. Four numbers, scored openly, on every initiative.

The point was not arithmetic precision. It was to move the decision off the loudest voice in the room and onto something written down that anyone could challenge. When a score is visible, disagreement becomes useful: people argue about the confidence number instead of about each other.

It also made refusal survivable. Saying no to a request is political; showing where it landed against everything else is a conversation.

Then I read the client, not only the idea.

A good idea still fails if it lands on a client who cannot carry it. So I mapped a funnel of approaches against how ready each client actually was, on four readings:

• the type of challenge in front of them
• how open they are to innovating at all
• who inside the client is accountable for it
• how far they can back a proposal on their own

Each reading narrows what is worth proposing. A client wide open to an idea and unable to sign for it needs a different move than one holding the budget with no appetite. The funnel decides how far a proposal can travel before it needs someone else’s signature.

Both feed one portfolio, not one queue.

RICE ranks. The funnel qualifies. Neither is useful alone, and a ranked list is still just a queue.

So the two feed a single innovation portfolio, held the way an investment portfolio is: many small, cheap, mostly-wrong explorations on one side, and a few larger commitments on the other where the answer is already known. That balance is the point. An area that is all exploration ships nothing; one that is all delivery stops finding anything.

The portfolio frame comes from Osterwalder’s work on explore and exploit. What I built was the way into it: the scoring that decides what enters, and the funnel that decides how far it goes.

A team that can say why it is not doing something is a team that can finish what it started.

We generated five times more AI assets than in the same period the year before. The number matters less than the reason behind it: every choice now had something written under it, so the team stopped re-litigating and started finishing.