Week 1 · her first day

What she got: a job description of six lines, the OPERA export and her confidentiality rules. No map, no how-tos, no scripts, no checks.

What she did well:

• The occupancy of every night matches the reference.
• She was honest about the gaps: Sunday has no availability row in the export, and the reservations are a sample.

What went wrong (our feedback):

1. No threshold. Monday is at 60.4% and Saturday at 62.7%, both below our 63%. She never flagged them, because nobody told her where the line is.
2. She did the math herself. She multiplied by hand, so Friday came out at 63.6% instead of 63.5%. Small, but next week it can be a big one, and nobody can check how she got there.
3. No evidence. None of her five insights says which file and which row it comes from.
4. Two sources don't match, and she didn't say so. The export lists 25 group blocks, but the group grid has numbers for only one of them. She used a 401-room block from the list, a useful catch, without warning that the grid doesn't include it.
5. No fixed format. The report is free text, so nothing can check it before the owner reads it.

Cost of this run: about 0.54 dollars.