Say/do ratio: measuring roadmap predictability from Jira epics

Predictability is the quiet superpower of a mature engineering organization. A team that ships 80% of what it commits to, reliably, is far more valuable to a business than a team that occasionally ships 120% but is impossible to plan around. The say/do ratio is the simplest honest measure of that reliability — and you can compute it directly from your Jira epics.

What the say/do ratio measures

The say/do ratio compares what you committed to against what you actually delivered. At the roadmap level, the cleanest definition uses epics:

The metric is deliberately blunt. It does not care how hard people worked or how much got done off-plan. It asks one question the business cares about: when you said you would deliver something by a date, did you?

Why epics, and why due dates

Stories and tasks are too granular and too volatile to measure commitment — they churn constantly within a sprint. Epics sit at the altitude where the business actually makes promises: a feature, an initiative, a deliverable with a target date. By using the epic's due date as the commitment and its done status as the outcome, you measure predictability at the level stakeholders experience it.

Delimetrics computes this from your Jira epics and their due dates. Because it reads the data you already maintain, there is no separate commitment-tracking process to keep in sync — the roadmap in Jira is the source of truth.

Beyond the headline ratio

The say/do ratio is the headline, but predictability has more texture than a single number. The roadmap predictability view rounds it out with:

• On-time delivery — of the epics that were due, what share landed by their due date rather than slipping past it.
• Schedule slip — how far late deliveries actually ran, so you can distinguish a few days from a few months.
• At-risk epics — commitments approaching their due date that are not yet on track, so you can intervene before they miss rather than reporting the miss afterward.

Together these answer not just "are we predictable?" but "where and how are we becoming unpredictable, and what can we do about it now?"

How to read a low say/do ratio

A consistently low ratio is rarely a people problem. The usual causes are structural:

• Over-commitment. Plans are built on optimistic capacity, ignoring KTLO and interruptions.
• Unmanaged scope. Epics grow after commitment without the date moving with them.
• Hidden dependencies. Work stalls waiting on other teams or external inputs.
• No date discipline. Due dates are aspirational rather than negotiated, so they were never realistic.

The fix is usually to commit to less and deliver it reliably. A ratio trending toward 1.0 over a few quarters is one of the strongest signals of organizational maturity you can show a board.

A note on gaming

Any metric can be gamed, and say/do is no exception — a team can hit 1.0 by committing to almost nothing. That is why the ratio should always be read alongside throughput and the investment mix. Predictability without meaningful output is not the goal; predictable delivery of meaningful work is. Looking at all three together keeps the incentive honest.

Reporting it upward

Say/do is one of the most board-friendly engineering metrics because it requires no translation. Executives intuitively understand "we committed to ten things this quarter and delivered eight." It pairs naturally with delivery metrics in a leadership update — see our guidance on CTO board reporting for how to frame it alongside throughput and cycle time.