The numbers. The people.
Worked investigations for a calmer payday. Commission, source records and the details worth getting right.
Worked close · 4 min read
Ten rows. Four questions. One explainable close.
Follow ten fictional sales rows through refunds, overlap, a rate difference and an excluded voucher to a verified $255 commission close.
Read the storyTen rows. Four questions. One explainable close.
Follow ten fictional sales rows through refunds, overlap, a rate difference and an excluded voucher to a verified $255 commission close.
Worked close · 4 min readWhere the extra cent goes
Three five-cent bases reveal a one-cent difference between line rounding and total rounding. Reproduce both methods and their negative mirrors.
Calculation notebook · 4 min readThe column we almost threw away
An original-column investigation shows how keeping list price, discount and staff attribution explains a $20 commission difference.
Product decisions · 4 min readThree reports, one week, three honest totals
Reconcile fictional itemised sales, payments received and a staff commission statement through a precise $990, $880 and $360 bridge.
Report investigations · 5 min readWhen Mia Chen becomes Mia Patel in an export
A fictional rename case separates confirmed identity from a similar name and explains legacy matching, dated staff links and duplicate checks.
Source identity · 5 min readThe retail return that arrived next week
Trace a fictional $33 partial retail return to a prior $66 sale while keeping the original $6 commission record unchanged.
Refund investigations · 4 min readOne service. Two staff. Start with the evidence.
A fictional shared service compares a confirmed 60/40 attribution with two incorrect shortcuts, and explains the one-earner-per-row product limit.
Attribution investigations · 4 min readThe blank tax cell that changed the close
Compare supplied tax, absent tax and explicit zero in three fictional $110 rows, with verified $40, $40 and $44 commission outputs.
Calculation notebook · 4 min readA review reason should survive the next question
Compare weak and useful notes for a refund, repeated sale and reported-rate difference, then test whether another reviewer can reconstruct the decision.
Product decisions · 4 min readHow we tested the commission calculators
Actual fixtures and results from 31 public-tool tests: signed half cents, tax reversal, malformed files, incomplete comparisons and statement downloads.
Engineering notebook · 4 min readRebuilding a handover from two unexplained totals
Turn an incomplete fictional Mia $123 / Noah $132 spreadsheet into a traceable handover with source, decisions, assumptions and a reconciled summary.
Handover investigations · 5 min readWhat a finalised snapshot actually keeps
Inspect the real fields saved at close and a controlled experiment showing later draft changes do not rewrite the fictional $255 historical result.
Product decisions · 4 min read