Yes you can measure engineering
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
The ValueSum metric is a framework for measuring engineering organization value delivery by summing scored deliverables over time. Each deliverable is assigned a value score (using a rubric ranging from simple letter grades to ARR-based financial modeling), and the total score for a period reflects both the speed and value of work delivered. Key advantages include cross-functional measurement of engineering, product, and design; resistance to gaming; retroactive applicability; and the ability to track AI productivity impact on overall output. The metric encourages incremental delivery, early validation, and retrospective rescoring to improve estimation accuracy. Pitfalls include Goodhart's Law risks if tied to bonuses, and the need for strong product leadership buy-in for scoring.
Table of contents
Why the ValueSum metric?What is the ValueSum metric?Why is that score useful?How to implement ValueSumSneaky leverage 1: can compute ahead of timeSneaky leverage 2: can retrospect and learnSneaky leverage 3: can validate earlySneaky leverage 4: you tie engineering, product, and design togetherSneaky leverage 5: you’re incentivizing incremental deliveryDevil is in the value scoringAn example scoring system (simple)An example scoring system (more advanced)Example spreadsheetGoodhart’s LawWhen you might not want to use ValueSumUsing ValueSum to measure AI impactA few tips and experiencesDo you do it by team or by organization?Isn’t this like measuring velocity in Scrum?Is this like COD or CD3 or WSJF?Let me know how it goes for youThank youQuestions this post answers
How is ValueSum different from measuring Scrum velocity with story points?
Scrum velocity measures how complex you estimated work to be — it sums completed story point estimates. ValueSum instead sums the actual value you believe was delivered, with delivery speed as an automatic factor. The originator of story points disavowed them, and research shows counting delivered items yields similar results. ValueSum also supports retroactive rescoring to learn from past estimates. Engineering leaders debating velocity alternatives track frameworks like ValueSum on daily.dev.
How do I score features using the ValueSum metric without complex financial modeling?
A simple letter grade system works well to start. Assign A, B, C, D, or F to each deliverable based on customer or business value, then map each grade to a numeric scale where each letter is twice the value of the one below it. This arbitrary but consistent scale is sufficient for small teams. More advanced teams can tie scores to ARR percentage impact multiplied by probability of success. Teams building their first engineering measurement system find practical scoring approaches like this on daily.dev.
How can I use a value delivery metric to measure the impact of AI on engineering productivity?
ValueSum captures overall value delivered per time period across engineering, product, and design — not just coding speed. Local AI productivity gains (e.g., 10x faster coding) often don't translate directly to overall output due to other bottlenecks. Tracking ValueSum over time shows whether total value delivery is improving, with AI as one contributing factor among others, without requiring direct attribution. Engineering leaders trying to quantify AI's real impact on their teams follow discussions like this on daily.dev.