Nexus / Built around retail decisions
A recommendation
should stand up to questions.
Why this option? Why now? What would make us change our mind?
In Retail Pro, Nexus is being developed to weigh stock, margin, cash, and store priorities together. Ava helps you question the recommendation, understand its limits, and decide what to try.
Retail decision record / Planned experience
The recommendation is only
the start of the record.
- Before the test
Record the evidence and the limit.
Keep the stock snapshot, exchange holds, cost assumptions, and spending ceiling with the recommendation. Missing size-level counts mean the evidence is incomplete.
- Owner review
Approve a bounded experiment.
Compare the bundle with full-price merchandising and markdowns. Record what the owner chose, what changed, and which actions were approved.
- After the test
Compare the prediction with the result.
Review units sold, contribution after campaign cost, remaining sizes, and the review period. Log surprises and outside factors before drawing a conclusion.
Inside the evaluation
A good-looking campaign
can still be the wrong move.
Suppose a seasonal promotion looks promising. The right answer changes when the available sizes, cash reserve, or evidence changes.
Evidence Is the stock count reliable?
A product total can hide missing sizes, exchange holds, and already-committed units. A stale count should trigger verification before the campaign, not a more confident recommendation.
Alternatives What are we giving up?
Compare a targeted bundle with broad markdowns, full-price merchandising, and holding the supplier order. Include margin lost on customers who would have bought anyway.
Constraints Does it fit the cash limit?
A positive expected return doesn’t pay a bill due tomorrow. Rule out options that breach the owner’s reserve before ranking the remaining choices.
Uncertainty What breaks the recommendation?
Test a lower response, fewer size matches, and changing costs. Show the point where another approach becomes preferable rather than hiding the decision behind a score.
Authority What can happen next?
A proposed test should name the spending limit, stock checks, and review date. The owner still reviews consequential actions before any campaign, price change, or purchase.
Confidence & outcome review / In development
A confidence score
isn’t an accuracy rating.
Confidence explains the evidence.
A useful confidence assessment should show data freshness, missing inputs, source agreement, and sensitivity to assumptions. A percentage without that explanation tells the owner very little.
Accuracy needs a testable prediction.
Outcome review needs a stated prediction, a time window, and an agreed way to judge it. Sales forecasts, margin estimates, and operational decisions need different measures; they should not be collapsed into one success score.
What should a decision history preserve?
The original question; dated source information; alternatives and assumptions; the recommendation and confidence explanation; the owner’s decision; the action actually taken; and the observed outcome with any unresolved gaps.
Decision history and outcome scoring are planned capabilities. No validated Nexus accuracy percentage is being claimed here.
Review availability and planned capabilities ↗Bring a decision to examine
Which assumption could change your answer?
Tell us about a stock, margin, or cash question you want to work through. We’ll discuss the available evidence and whether it fits the pilot scope.