Try Guided Weight Discovery
Enter a Decision Catalog ID. The shared capability loads its authoritative criteria, then the server selects one useful comparison at a time. Preliminary priorities and the stopping decision are server-authoritative—not a documentation-only calculation.
When Should I Use This?
Criterion weights represent how much each factor matters in a decision. Choose the path that matches what your organization already knows.
Use Decision Defaults
“I am comfortable with the standard priorities.” Use the weights from the selected Decision Catalog decision or Profile.
Provide My Own Weights
“My organization already knows its criterion weights.” Submit the established values and let DecisioQ validate and normalize them.
Help Me Determine Weights
“I know what matters, but I do not know what percentages to assign.” Use Interactive Weight Discovery.
Interactive Weight Discovery
Interactive Weight Discovery uses AHP (Analytic Hierarchy Process) to translate business priorities into criterion weights. Instead of asking you to invent percentages, DecisioQ asks which decision factors matter more and derives normalized weights from those judgments.
- Business priorities
- Business-friendly comparisons
- AHP and consistency check
- Review, then execute
You express preferences; the system performs the numerical work. The result reflects your stated priorities rather than an objectively universal “correct” weighting.
How the Questions Work
Suppose a vehicle acquisition decision uses Expected Profit Margin, Acquisition Cost, Vehicle Condition, and Days-to-Sell Risk. Instead of asking for four percentages, the Studio can ask: “When deciding whether to acquire this vehicle, which matters more: Expected Profit Margin or Vehicle Condition?”
- About equally important
- Expected Profit Margin is somewhat more important
- Expected Profit Margin is much more important
- Vehicle Condition is somewhat more important
- Vehicle Condition is much more important
A second question might compare Acquisition Cost with Days-to-Sell Risk. The business user never needs to enter a matrix value; DecisioQ maps the selected statement to an AHP intensity.
What AHP Does
AHP converts the complete set of relative-importance judgments into normalized weights.
| Criterion | Illustrative derived weight | Meaning |
|---|---|---|
| Expected Profit Margin | 34% | Strongest priority. |
| Acquisition Cost | 28% | Nearly as important. |
| Vehicle Condition | 23% | Remains significant. |
| Days-to-Sell Risk | 15% | Lower relative importance for this preference set. |
These numbers are illustrative, not authoritative weights for a Decision Catalog decision. A margin-focused dealership and a dealership managing aging inventory can legitimately derive different weight sets.
Why You May Not Need to Compare Everything
A complete AHP matrix contains n(n - 1) / 2 unique comparisons—15 for the six criteria in AUTO-AUCT-006. Guided mode first builds a connected preference graph, calculates preliminary log-least-squares priorities, and asks high-value validation questions. It stops when the graph is connected, the configured validation evidence is present, and no material conflict needs clarification. The exact number adapts to the answers; Full Pairwise remains available as an advanced option.
What Happens If My Answers Conflict?
Human judgments can conflict. For example, Cost > Quality, Quality > Risk, and Risk > Cost creates a preference cycle. AHP calculates a Consistency Ratio (CR) to detect this pattern. The Studio explains whether comparisons are reasonably consistent and asks the user to refine answers when they are not.
0.10 is the current DecisioQ acceptance policy for three or more criteria. It is a commonly used guideline, not a universal law. The numerical CR remains available as advanced information.Review Your Weights
Before execution, review the derived percentages in business terms. You can conceptually:
- Use These Weights and proceed with the AHP-derived values.
- Refine My Answers by revisiting comparisons.
- Use Decision Defaults and discard the derived values.
The current Studio uses the accepted weights for the current request. It does not provide persistence for saving a discovered set for later reuse.
AHP Determines Weights—It Does Not Choose the Winner
- Business priorities
- AHP
- Criterion weights
- TOPSIS/WSM recommendation
Interactive Weight Discovery does not change criterion values, units, directions, validation, eligibility, or the authoritative criterion set. It does not rerank a finalized result, replace TOPSIS or WSM, or directly select a winning alternative.
A selected Profile can supply default weights. Consistent AHP-derived weights explicitly supplied for execution replace those Profile weights. A Scenario affects weights only when its authoritative decision configuration declares that behavior. Check configurationUsed.effectiveWeights.source for the final provenance.
Advanced: How AHP Works
Pairwise comparison matrix
For each comparison, the preferred criterion receives an importance value and the reverse relationship receives its reciprocal. Diagonal values are 1. The current Studio exposes business statements that map to 1 (equal), 3 (somewhat/slightly more important), or 5 (strongly/much more important), including the reciprocal direction. It does not expose the full traditional 1–9 scale in the business interface.
Deriving the weight vector
The current Studio initializes an equal weight vector, repeatedly multiplies it by the comparison matrix, and normalizes the result for 1,000 iterations. This power-iteration method approximates the matrix's principal eigenvector, which becomes the normalized criterion weight vector.
Consistency Index and Consistency Ratio
The implementation calculates the principal eigenvalue estimate, then CI = (lambdaMax - n) / (n - 1). It calculates CR = CI / RI using RI values 0.58, 0.90, 1.12, 1.24, 1.32, 1.41, 1.45, and 1.49 for matrices of size 3 through 10. For two criteria, CR is zero. Studio blocks execution above 0.10 for three or more criteria.
Current Implementation and API Boundary
| Capability | Status |
|---|---|
| Guided, server-selected comparisons | Available; recommended default |
| Sparse preliminary priorities and preference-graph reasoning | Available |
| Early conflict detection and targeted clarification | Available |
| Full Pairwise comparison | Available; advanced option |
| Authoritative final AHP derivation | Available |
| Reusable/persisted discovered weights | Not currently supported |
POST /api/v1/weight-discovery/evaluate is stateless: clients send the criteria, orientation, and preserved direct judgments on each call; the server returns preliminary or final weights and the next comparison. Full matrices may still be submitted to /api/v1/ahp/generate. See the API Guide.