Modernization assessment benchmarks (illustrative)
Across a 50-workload illustrative corpus spanning Mainframe, Java EE, .NET, Database, and VMware tiers, we measured how saasups scores compare to a manual consulting assessment, generic migration tooling, and in-house scripting approaches. These numbers are a methodology preview — not audited results.
Four numbers that matter to engineering leaders
Illustrative point-in-time figures across the 50-workload corpus. All values compare saasups to the next-best generic approach category.
Illustrative mean saasups score across the 50-workload corpus. Next-best approach scored 67/100.
Illustrative percentage of workloads where saasups matched the expert-adjudicated correct 7-R decision. Next-best: 54%.
Illustrative time from estate ingestion to a complete scored output. Manual consulting assessment: 47 days.
Illustrative median absolute percentage error on post-migration cost projections. Generic tooling: ±31%.
↑ All figures illustrative. See the methodology section below for corpus composition and scoring protocol.
saasups against generic assessment categories
Illustrative scores across four assessment dimensions. saasups is compared to generic approach categories — not any named vendor product.
How the illustrative corpus was constructed
This section describes the methodology behind the benchmark numbers above. Every figure is illustrative — designed to preview the scoring protocol, not to assert audited results.
Corpus composition (illustrative — 50 workloads)
| Workload class | Count | Notes |
|---|---|---|
| Mainframe (COBOL/PL1) | 8 | Batch-heavy, high compliance constraints |
| Java EE monoliths | 12 | High coupling, variable change cadence |
| .NET Web Forms | 10 | Mix of commodity and differentiated capability |
| Relational database tiers | 11 | Oracle/SQL Server, schema complexity varies |
| VMware-hosted VMs | 9 | Infrastructure-layer only, no app-layer data |
| Total | 50 | Illustrative — not a public dataset |
Scoring protocol
- 01Each workload was independently adjudicated by a three-person expert panel to establish the ground-truth 7-R decision.
- 02Each approach category (saasups, manual consulting, generic tooling, in-house scripts, rules-only) was then applied to the same corpus to produce its own recommendation for each workload.
- 03The recommendation was compared to the expert panel's adjudicated decision to compute the 7-R match percentage.
- 04FinOps prediction error was measured as median absolute percentage error against post-migration actual costs on a subset of 18 workloads with available actuals.
- 05Risk recall measures the fraction of dependency and compliance risks flagged by the expert panel that each approach independently identified.
- 06Coverage measures the fraction of the 50-workload corpus that each approach produced a complete, actionable recommendation for — partial or incomplete outputs counted as not covered.
This methodology is published for transparency, not as a reproducibility guarantee. Because the corpus is illustrative rather than a public dataset, external replication is not possible from this page alone. Engineering leaders who want to run the same protocol against their own estate should contact the saasups team for the full methodology specification.
A representative slice of the illustrative corpus
These 12 rows are illustrative. The ✓ / × columns show whether saasups matched the expert-adjudicated decision — not whether the expert was correct.
Illustrative match rate on this slice: 10/12 (83%) — all figures illustrative.
Run this methodology against your own estate.
The protocol above is designed to be applied to any workload corpus. Engineering leaders who want to apply the same scoring framework to their estate — and get back real numbers, not illustrative ones — can request the full methodology specification from the saasups team.
Book a technical walkthrough