RACE Programming
Practice Guide · Measuring a RACE rollout

If your team measures with DX Core 4

A measurement playbook. For each of the four DX Core 4 dimensions: the primary DX metric, what to measure in its place under RACE Programming, where to pull it from, and what to target. DX Core 4 already balances speed, which is exactly what the RACE triangle does, so the mapping is close to one to one.

The four dimensions, instrumented

Speed

  • DX primary metric: Diffs (or PRs) per Engineer.
  • Measure instead: Executable User Stories accepted per week per team, plus Pit Stops per week.
  • Source: Executable Product Backlog and CD pipeline.
  • Target: EUS per week trending up; Pit Stops weekly or faster.
  • Watch: do not track Diffs per Engineer. The agent inflates it, and it rewards volume over accepted value.

Effectiveness

  • DX primary metric: the Developer Experience Index (DXI).
  • Measure instead: a small index built from delegation fluency, Inner-Cycle friction, and cognitive load (the DevEx measures: time-to-gate, clarification rate, focus blocks).
  • Source: a per-Stint pulse plus Inner Cycle and board logs.
  • Target: the index rises as Speed rises. This is the counterbalance.
  • Watch: Speed up while Effectiveness falls means you are burning the team to hit throughput. Stop and fix the experience.

Quality

  • DX primary metric: Change Failure Rate.
  • Measure: four-gate Definition-of-Done first-pass rate, Change Failure Rate, and Deployment Rework Rate.
  • Source: Definition-of-Done gate logs joined to incidents.
  • Target: high first-pass rate, low and stable CFR and rework as Speed rises.
  • Watch: always plot Quality on the same chart as Speed. A speed gain with falling quality is not progress.

Impact

  • DX primary metric: percentage of time on new capabilities.
  • Measure: share of Stint capacity on new capability versus maintenance, and the one triangle outcome you took (scope, cost, or time at held quality).
  • Source: Executable Product Backlog tagging (new versus maintenance) plus delivery data.
  • Target: state one triangle number per engagement, next to the new-capability share.
  • Watch: high Speed on low-Impact work is motion, not impact.
The one addition

Add the ceiling DX Core 4 does not name

  • Measure: Team Principal absorption capacity, EUS the client can specify and accept per week divided by EUS the Pit Crew can ship per week.
  • Source: Executable Product Backlog (accepted-by-client timestamps) versus Pit Crew throughput.
  • Target: ratio approaching 1. Below 1, the client is the bottleneck; add Pit Wall capacity (see the staffing guide).

A "RACE Core 4" dashboard reads Speed and Quality off the Stint and the Definition of Done, Effectiveness off the Inner Cycle, and Impact off the triangle, with absorption capacity as the fifth reading that sets the pace once execution is free.

DX Core 4 is the unified scorecard. RACE Programming is the operating model it scores.

Read the framework → · All practice guides →

FAQ

Frequently asked questions

How does DX Core 4 apply to RACE Programming?
DX Core 4 unifies DORA, SPACE, and DevEx into four dimensions, Speed, Effectiveness, Quality, and Impact, each with a primary metric. RACE Programming maps almost one to one: Speed becomes Executable User Stories accepted per week and Pit Stop cadence, Effectiveness a delegation-fluency index off the Inner Cycle, Quality the four-gate Definition of Done and change failure rate, and Impact the share of Stint capacity on new capability plus the triangle outcome.
Why is DX Core 4 the best framework for measuring a RACE rollout?
Because it is the only one of the four that already counterbalances speed with effectiveness and impact in a fixed structure, which is exactly the balance the RACE Programming triangle asserts: you do not spend the AI dividend on raw speed alone, you choose scope, cost, or time at held quality. A RACE measurement layer maps naturally onto DX Core 4, with one addition it lacks: Team Principal absorption capacity.
What replaces Diffs per Engineer as the Speed metric under RACE Programming?
Executable User Stories accepted per week. DX Core 4 measures Speed with Diffs or PRs per engineer, a human-output proxy AI breaks because the Silicon Software Engineer writes the code. RACE measures the unit of accepted, specified value instead, alongside Pit Stop cadence.
How does RACE Programming measure the Impact dimension?
Measure the share of Stint capacity spent on new capability versus maintenance, tagged in the Executable Product Backlog, and state the one triangle outcome you took (scope, cost, or time at held quality). DX Core 4 measures Impact as the percentage of time on new capabilities; RACE quantifies the business result on top of it with the triangle.