Prescriptive vs. Framework Metrics
Know whether your metric is a recipe or a lens — and never misuse a recipe
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 3
- Confidence
- 90%
A mental model for classifying measurement tools into two types: prescriptive metrics (like DORA's four keys) that must be used ONLY for the specific purpose they were designed for, and frameworks (like SPACE) that tell you which dimensions to consider but leave the specific metric to you. Misapplying a prescriptive metric to answer a question it wasn't built for is a core measurement error, especially now that AI has changed feedback loops.
Origin
Forsgren, who created DORA (prescriptive) and co-created SPACE (framework), draws the distinction explicitly to explain why people misuse her own metrics — applying DORA's speed/stability metrics 'blindly' to questions AI has reshaped.
Core principles
- 01A prescriptive metric may only be used in the way it was prescribed
- 02DORA's four keys (deployment frequency, lead time, MTTR, change fail rate) assess pipeline speed and stability — nothing more
- 03A framework tells you dimensions, not numbers; you supply the fit-for-purpose metric
- 04AI moved feedback loops earlier and mid-pipeline, so old prescriptive uses now miss important phenomena
How to run it
- 1
Classify the metric
Determine whether the metric is prescriptive (a defined recipe like DORA) or a framework (a lens like SPACE).
- 2
For prescriptive metrics, honor the original purpose
Use DORA only to assess pipeline speed and stability. Do not stretch it to answer questions about AI code trust or new feedback loops.
Watch out Blindly reapplying prescriptive metrics under AI will make you miss super-important changes in how people work.
- 3
For frameworks, choose fit-for-purpose metrics
With SPACE, pick the specific metric that answers your actual question in your actual context.
Pro tip This is why frameworks survive context shifts — you re-select the metric each time.
In the wild
Forsgren says DORA's speed and stability metrics still work for assessing the pipeline overall, but because AI created much faster and earlier feedback loops, using DORA to understand feedback quality now misses signal that lives mid-pipeline.
→ Leaders keep DORA for pipeline health but stop over-reading it as a total-productivity verdict.
Common mistakes
Treating DORA as a universal productivity score
'If that's all you're looking at it's not going to be sufficient anymore' — DORA measures pipeline speed and stability, not AI-era code quality or trust.
Is it for you?
Best for
Metrics owners and eng leaders deciding which existing metric answers a specific question
Not ideal for
Teams with no existing metrics who first need to run a listening tour and surveys
From the transcript
“if it is a prescriptive metric it needs to be used only in the way it was prescribed”
“we can't just blindly apply the existing metrics we've used before because we'll miss super important phenomenon”
“space is a framework. It doesn't tell you what metric to use.”
From the episode
How to measure AI developer productivity in 2025
Nicole Forsgren