No Process Is Still a Process: Why Operating Models Break on Ownership, Not Design

No process is still a process. No decision is a decision. That’s the reason so many product organizations end up running on an operating model nobody actually designed, just the accumulated residue of every call that got deferred instead of made.

Courtney Cunningham sees that residue up close. As a Partner at Fractional CPO, he embeds inside established companies mid-transition, where the operating system that got them this far has quietly stopped working, and nobody owns fixing it, because nobody ever decided who should.

In this episode of Accelerate Podcast, host Becky Flint, Founder and CEO of Dragonboat, talks with Courtney about what leadership actually means when it asks for a new PDLC, why the operating model is a partnership product ops can’t complete alone, and where AI is quietly raising the stakes on data quality and decision speed.

What you’ll hear:

• “We need a new PDLC” is a request to investigate, not a brief to execute: The instinct is to go build one. Courtney pushes back first — what are you actually observing? — then maps the current state with a simplified event-storming workshop: stakeholders plot key events on a whiteboard, and the real gaps surface on their own.
• The operating model is a partnership, and product ops can’t supply the half that isn’t design: Product ops can build and maintain the system, but it has no authority over product managers and can’t manufacture the will to use it. That intent has to come from product leadership itself. Skip the partnership, and ops ends up owning a behavior-change problem that was never theirs to own.
• Consistency is a design choice at the seams, not a mandate everywhere: Teams should run differently, but what has to be consistent is the handful of touchpoints where they depend on each other. Courtney’s example: no shared way to signal a feature’s real stage i.e alpha, experiment, GA so sales sells things that aren’t ready, generating churn that has nothing to do with the product.
• Shipping speed and customer-absorption speed are different constraints, and only one moves with better tooling: CI/CD can push how fast engineering ships, but B2B businesses still gate releases to a cadence sales can sell against, and customers still have a ceiling for how much change they’ll tolerate. Speeding up the build side doesn’t move that ceiling.
• AI didn’t just speed up decisions, it removed the person who used to catch bad ones: When PMs query the warehouse directly instead of routing through an analyst, decisions get faster,  but the analyst was also the checkpoint catching bad data first. Remove that checkpoint, and ops’ job shifts from producing reports on request to owning the data itself as a real source of truth.

Becky closes with the connecting thread: the harder AI pushes every team to move faster, the more it exposes whether product, GTM, and customer teams are actually working from the same context or from scattered, quietly drifting versions of the truth.

That’s the coordination tax Dragonboat is built to remove: one connected operating model and source of truth for product, portfolio, and outcomes, built for the partnership between product leadership and product ops that this whole conversation comes back to — for the humans running it, and the agents increasingly running alongside them.

Reference

• Courtney Cunningham on LinkedIn: https://www.linkedin.com/in/courtneycunningham13/
• Becky Flint on LinkedIn: https://www.linkedin.com/in/beckyflint/