| Benchmark variable | Reference range | Interpretation boundary |
|---|---|---|
| Duplicate-rate benchmark | 8-18% of active MRO catalog records | Planning assumption until replaced by uploaded catalog evidence. |
| Average affected item value | USD 850 base reference | Used only for benchmark sizing where actual unit costs are unavailable. |
| Carrying-cost drag | 18-25% annualized reference range | Represents working-capital cost, storage, obsolescence, and management overhead. |
| Recoverable capital range | 25-55% of duplicate exposure | Depends on stock status, criticality, interchangeability, and owner approval. |
| Emergency-buy premium | 15-25% planning range | Applied where duplicate records obscure preferred supplier or available-stock logic. |
| FX conversion | USD base values converted to local display currency | South Asian currencies use Lakh/Crore notation; other currencies use K/M/B notation. |
Evidence discipline
Benchmark first. Actual diagnostic second. ERP action only after owner review.
AI2COE uses benchmark variables to help leaders size whether a problem is worth examining. PartsCleanse AI then uses uploaded data to create actual findings. The product never treats a planning benchmark as a command to retire, merge, or delete ERP records.
Force Team control: Any public value statement should say whether it is a benchmark assumption, a calculated estimate, or diagnostic evidence from uploaded data.
FAQ
How to interpret AI2COE benchmark numbers.
Are AI2COE benchmark numbers final ROI claims?
No. Benchmarks are planning assumptions. Final ROI depends on uploaded data, actual quantities, unit costs, confidence tiers, and owner-approved remediation.
Why does AI2COE use USD as a base value?
USD is used as a neutral calculation base for benchmark comparability. The portal converts display values into the user's country currency or authenticated profile currency.
Why use Lakh and Crore for India-style currencies?
Large values in INR, PKR, BDT, LKR, and NPR are easier for local executives to read in Lakh and Crore notation than in generic M/B notation.
What makes the diagnostic different from the benchmark?
The diagnostic uses the actual uploaded catalog to produce duplicate-family findings, confidence tiers, and report artifacts. The benchmark only estimates the value signal.
