How to Estimate Complexity
Complexity Assessment Rubric – How to Score Effort on a 1‑5 Scale
Before you can compare initiatives apples‑to‑apples, convert the raw effort estimate from Engineering (story points, hours, or “sprints”) into a normalized “Effort” score (1 = smallest, 5 = largest).
That conversion has two parts:
-
Size – How long will it take? (Engineering estimate)
-
Complexity – How risky or uncertain is the work?
Key idea: Ten straightforward days of coding (low complexity) is not equal to ten days spent pioneering a new architecture (high complexity). The latter carries far more delivery risk and therefore should score higher.
Use the rubric below to rate complexity across five dimensions and use the highest single dimension as the Complexity score.
|
Dimension |
Low (1 – 2 pts) |
Medium (3 – 4 pts) |
High (5 pts) |
Illustrative Examples |
|---|---|---|---|---|
|
Technical implementation |
Re‑use existing patterns; isolated code changes; minimal dependencies |
Some new patterns or light refactor; affects several services |
New architecture or unproven tech; refactors core components |
Low: add a field to an existing screen. High: first release of AI‑driven autonomous dispatch. |
|
Integration impact |
No new integrations; tweak existing internal API |
1‑2 new standard REST integrations or major changes to several internal modules |
Multiple complex integrations (legacy EDI, streaming); overhaul of integration layer |
Low: expose an existing data point via REST. High: real‑time, bi‑directional link to a 3PL’s proprietary system. |
|
Requirement stability |
Well‑defined, stable, agreed by stakeholders |
Mostly clear but some ambiguity; minor scope shifts likely |
Vague, evolving, or contentious; high scope‑creep risk |
Low: clearly reproduced bug fix. High: “next‑gen customer portal” based on loose, shifting ideas. |
|
Data‑model change |
No schema change; existing structures used |
Minor schema tweak; simple migration |
New core entities or major table redesign; heavy migration; perf risk |
Low: show an existing field in a new report. High: introduce “Multi‑leg Shipment” entity affecting ordering, tracking, billing. |
|
New‑tech risk |
Familiar stack; team has deep experience |
Moderately new lib/framework; some learning curve |
First use of a new tech stack or ML technique; many unknowns |
Low: add feature with current Java/.NET libs. High: deploy first ML model with a new MLOps pipeline. |