Week 5 dataset notes, Summit Gear accounts receivable

How the Week 5 receivables data was built and how aging is defined. The lab, the homework, the dashboard and the grader all follow these rules, so if a number is ever in question, this page decides it.

Files

• Lab 5: invoices.csv and customers.csv (the same data is already loaded in lab_05_template.xlsx).
• Homework 5: invoices.csv and customers.csv (the data behind the Homework 5 dashboard).

invoices.csv has one row per invoice: invoice_id, customer_id, invoice_date, amount, paid_date. customers.csv has customer_id, customer_name, segment, net_terms. You attach the customer name and terms to each invoice with XLOOKUP.

Same company as Week 3, not the same data

Week 5 reuses the company, Summit Gear Co, so the story carries over from Week 3. It does not reuse the Week 3 data. The Week 5 set was generated fresh for this week, with 100 customers, specific net terms and a Q1 2025 to Q1 2026 window, so the numbers do not tie to Week 3 and are not meant to.

The dataset

• Synthetic and repeatable. The data comes from a seeded generator, so the same seed always gives the same files. There are no real people or companies; every name is a fictional business account.
• Two separate sets. The lab and the homework are generated independently from different seeds, so a lab answer cannot be carried into the homework.

Aging: days past due

Net terms are mixed (Commercial 60, 90 or 180 days, Consumer 30), so aging by days since the invoice date would not be comparable across customers. A net-180 invoice that has been outstanding for 130 days is still current, while a net-30 invoice at 130 days is 100 days late. Aging is therefore measured in days past due, against each invoice's own due date. That is why the lab attaches net_terms: you need it to find the due date.

• Snapshot (as-of) date: 2026-03-31. Aging is measured against this date unless you choose a different one on the dashboard.
• Due date = invoice_date + net_terms.
• Days past due = (paid_date if paid, otherwise the as-of date) minus the due date. Zero or less means the invoice is not yet due (within terms).
• Aging buckets, with these exact labels: Not due, then 0-30, 31-60, 61-90, 91-120, 121-150, 151-180, 181+, all in days past due. Paid and open invoices are both bucketed; a paid invoice's bucket is how late it was when it was paid.
• Open or paid. A blank paid_date means the invoice is open as of the snapshot. A paid_date is recorded only if the payment landed on or before the snapshot.
• Paid in full. Every invoice with a paid_date is paid in full. There are no partial payments.
• Voids: none. Every invoice is a live receivable, open or paid in full.
• Late payments trickle in. Most slow payers settle within about 90 days past due, but a small share of delinquent accounts pay far later. Well under 1% of payments land 181 or more days past due (the latest land roughly 10 to 11 months past due), so the high buckets see some collections rather than a hard cliff.
• Days to pay is a separate metric, used for the scatter and the histogram. It is measured from the invoice date: paid_date minus invoice_date. It answers how long a customer took, which is a different question from how late they were.

Invoice cohorts on the dashboard

The dashboard's cohort table groups invoices by due month, not billing month, so every invoice in a cohort comes due at the same time and ages in lockstep regardless of its net terms. Following a fully aged cohort, the share that reaches 181 or more days past due and stays uncollected is a steady-state bad-debt loss rate. Recent cohorts have not aged yet and understate the loss, so a reserve estimate usually averages several mature cohorts.

Segments and terms

• Commercial: net 60, 90 or 180. Larger invoices, fewer of them, slower payers.
• Consumer: net 30. Smaller invoices, more of them, a mix of fast and slow payers.

Filters

• The lab and the dashboard include every invoice in the window unless a chart says otherwise. Nothing is excluded by default.
• The one explicit subset is the lab's January 2026 column chart: only invoices with an invoice_date in 2026-01.
• The dashboard states its active as-of date, the days-past-due basis, and that there are no voids. Changing the date re-ages every invoice.