How Devint measures, without turning metrics into judgment.
The DevScore combines automatic signals from the development workflow with leadership evaluations, comparing each person against stable reference bands — not against the top performer on the team. Weights and bands are configurable: your company defines what it values, position by position.
An absolute score, from 0 to 100.
The DevScore is not an internal ranking. It compares each person against healthy operating bands: when the whole team improves, the improvement shows up as collective — nobody ends up "in last place" artificially.
A single score scale, per person and per team. The dimensions explain where the number comes from.
Dimensions evaluated, each with its own weight and saturation cap.
Between closings: a trend reading, without day-to-day micromanagement.
Of full window in every closing, to reduce the noise of atypical days.
Illustrative example: mid-level profile with a DevScore of 84 at closing.
One snapshot per closing.
At every biweekly closing, the six dimensions form a hexagon: each vertex shows where the person stands relative to the healthy band for that dimension, on a scale of 0 to 100. The shape reveals the profile — where there is consistency and where there is room to grow — even before you look at the final number. Beyond the biweekly closing, you can also filter the data by other time periods, depending on what the analysis calls for.
AI insight. The AI analyzes performance shifts between closings and suggests what may have caused the change and what the manager can do about it. It is not a fixed rule — it is an assistant that helps interpret the data, keeping the decision with leadership.Six dimensions. One clear reading.
No single dimension defines productivity. Automatic signals show the period; leadership evaluations show maturity — together, they reduce unfair readings. The weight of each dimension is configurable, reflecting what your company values at each moment and in each position.
Delivery paceconfigurable weight
How often technical work turns into something reviewable. A healthy pace reduces the risk of surprises; smaller, more frequent deliveries make review and feedback easier.
A commit is the traceable record of a change to the code; a pull request is the request to review and integrate that change.
Contributionconfigurable weight
Technical contribution that leaves a material mark on the product: code that stays and structural changes, including useful deletions during refactoring.
Alive lines are the lines that remain active in the product over time — contribution that generated real value.
AI adoptionconfigurable weight
Real adoption of AI tools in the workflow. It measures usage, not value delivered — it should be read alongside the other dimensions.
Tokens are the units of text processed by AI tools — consumption indicates the real level of use at work.
Time trackingconfigurable weight
Completeness of time logging and predictability. Without reliable records, capacity becomes opinion. Logging above the ceiling does not raise the score.
Hours logged by the developer bring predictability, capacity visibility and support cost analysis.
Hard skillsconfigurable weight
Planning, execution and autonomy. It describes observed technical maturity — more stable than period metrics, it changes slowly.
Code quality, architecture, best practices and command of the technologies, assessed by those who lead technically.
Soft skillsconfigurable weight
Communication, accountability, predictability and collaboration. It measures behavioral impact on how the team works, not likeability.
Communication, collaboration, autonomy and organization — professional maturity and impact on the environment.
Weights that follow the position.
What you expect from a junior is not what you expect from a senior. That is why weights (and bands) can be configured by seniority level or role — always adding up to 100%. Here is one example configuration:
Junior
Mid-level
Senior
Companies at different stages configure it differently too: those accelerating AI adoption may raise the weight of that dimension; those who need predictability, the weight of time tracking.
Healthy zones, with a floor and a ceiling.
Every signal is read against a healthy operating band — with a floor and a ceiling. Like the weights, the bands are configurable by position: they are not blind targets, they are reading parameters tuned to the context of each company and level.
| Dimension | What it looks at | Example band* | Why it matters |
|---|---|---|---|
| Delivery pace | Commits per business day | 3 to 10 /day | Shows continuity of work |
| Delivery pace | Pull requests per week | 1 to 5 /week | Shows the cadence of reviewable delivery |
| Contribution | Alive lines | 1K to 8K /cycle | Shows contribution that is still present in the product |
| Contribution | Change entropy | 1K to 3.5K /cycle | Shows the breadth of the technical work |
| AI adoption | Tokens consumed | 50M to 500M /week | Shows operational adoption of modern tools |
| Time tracking | Hours logged | 30h to 40h /week | Shows completeness of time logging and predictability |
| Hard skills | Technical evaluation (manager + tech lead) | Scale 1 to 9 | Shows observed technical maturity |
| Soft skills | Behavioral evaluation (manager + tech lead) | Scale 1 to 6 | Shows observed collaborative maturity |
*The values above are a reference configuration. Each company adjusts the bands by seniority level, contract type, working hours and nature of the work — with governance: adjustments apply only to upcoming closings, preserving the history already closed.
See the DevScore applied to your context.
Book a demo and explore the dimensions, bands and levels with your team's data.