Calculation model

The cloudtec Platform Configurator estimates a robust build budget from the properties of your project. The model is calibrated against eight real cloudtec reference projects; estimates should land within ±25 % of the eventual quote.

CapP — Capability Points

Every selected module, helper function, integration and migration stream contributes a fixed number of Capability Points (CapP). Example weights: module +4, helper function +2, custom core function +11, integration +0.05 per third-party system, migration stream +5. CapP is the volumetric base measure — it tells you how much "build work" the project carries.

CapP drives the base effort: effort in hours = CapP × 9. With 1 person-day = 8 hours the PD figure follows. Both numbers update live in the sidebar.

Complexity index (CI)

The CI weights ten categories into an index between 0.10 and 3.50:

• Functional scope (20 %)
• Security (13 %)
• Integration (13 %)
• AI functions (13 %)
• Data & migration (9 %)
• Compliance (8 %)
• Non-functional requirements (8 %)
• Maintenance & availability (6 %)
• User scale (5 %)
• Team skill / domain expertise (5 %)

The CI multiplies the base effort (Complexity Multiplier, CM), modulated by the uncertainty factor UF (greenfield, missing input, "don't know" answers). Final formula: effort × CM × UF = build effort.

Confidence interval

The point estimate carries a symmetric 90 % confidence interval. The more complete and concrete your input, the narrower the band. A discovery workshop typically reduces the band to ±15 %.

T-shirt sizing

The build effort is mapped to a T-shirt size (S, M, L, XL). It's a coarse comparison value — not a substitute for the point estimate in CHF.

Recurring costs

• Licence & maintenance: 10 % of build budget per year — stack updates, security patches, warranty
• Service management: 5 % per year — organisational steering, reviews, reporting, escalation
• Hosting: 3.5 % per year — Docker infrastructure, backup, WAF, monitoring, Swiss data residency
• Evolution: 15 to 50 % per year, decided by the customer

Calibration

Weights are tuned so all eight historical projects fall within the ±25 % tolerance. Last calibration: 14 May 2026. Every model change runs the pre-commit hook test-calc.js and verifies the tolerance.