lesleymaore.

See the Demand. Test the Decision.

Experience and purpose

Inspired by my Commercial Data Analyst & Sales Manager role at Unified Meat Packers (September 2019–January 2023). My CV records that Excel tracking tools and sales forecasting models supported a 35% increase in total revenue. This case is a separate, fictional learning exercise. The CV figure is not a simulation target, an independently verified result or an outcome of this portfolio work; its comparison period is unspecified.

For sales operations, purchasing and finance at a fictional distributor: decide whether to trial an unfilled-request log before changing replenishment. These stakeholders have not been interviewed.

The problem in one example

A 10 kg request with 6 kg supplied leaves 4 kg invisible in sales-only reporting. Logging the shortfall reveals demand, but automatically buying more can increase perishable waste.

What the experiment establishes

Under the unchanged buying rule, fuller logging slightly improves fulfilment but not contribution reliably. Buffer and shelf-life assumptions matter. Test the log and purchasing rule separately; no universal gain is established.

Proposed workflow

Sales captures a request → record requested and fulfilled kg → identify the unfilled reason → purchasing reviews shortages alongside stock age and outstanding replenishment → finance checks fulfilment, waste and contribution together → approve or reject a bounded policy trial against an agreed service floor.

Requirements and acceptance criteria

  1. One request-product line per identifier. Store requested kg, fulfilled kg, unfilled kg, date, product and reason. Acceptance: 10 requested and 6 fulfilled reconciles to 4 unfilled. A later delivery must not be counted again as new demand; distinguish cancellation, substitution and backlog explicitly if used.
  2. Missing is unknown. Acceptance: unanswered quantity fields do not become zero; unsupported reasons remain “unclassified”. Report completion coverage and audit a sample of entries.
  3. Preserve the existing sales record. Acceptance: invoice totals remain unchanged when an unfilled request is logged. No fictional requested amount becomes recognised revenue.
  4. Separate visibility from action. Acceptance: the request log creates a review item, not an automatic purchase order. Buyers see stock age, open purchase orders and the applicable service/waste comparison before deciding.
  5. Keep a balanced scorecard. Acceptance: show requested and fulfilled kg, expired kg with denominator, average stock and contribution definition. No single KPI determines success. Logging effort and costs must be measured before claiming a positive business case.
  6. Evaluate fairly. Acceptance: define the comparison period and service floor before the trial, preserve product and weekday mix, document supplier disruptions, and report adverse as well as favourable outcomes. Agree any service threshold with stakeholders rather than inventing one from this simulation.

Open questions for a real business

Are requests logged before checking stock? What happens to unfilled demand? Can customers substitute? Can shelf age be tracked reliably? Which costs change with stock? What service level is commercially acceptable? How much staff time would logging take? Do account prices and delivery promises affect demand?

Methods

The experiment uses fictional inventory records and paired simulation runs to compare sales-only, partial-logging and complete-logging policies under the same demand and supplier-delay sequences. Read the simulation methodology for the stock-flow rules, scenario assumptions, cost definitions and limitations, and the research and attribution record for the work informing the question.