Payment behavior drift skill
You can use the Payment behavior drift skill to find customers with a changed payment behavior, such as paying slower or with more overdue invoices.
The skill compares customer payment behavior between two periods (for example the last 6 months compared to the prior 6 months) and ranks all customers by how much their behavior has shifted.
The skill returns:
- A watch list with a few customers drifting, that may be worth acting on
- Full ranking of all qualifying customers
- A methodology footer with the exact numbers used
The skill is read-only, and doesn’t modify anything in Business NXT.
Typical use cases:
- Quarterly AR review before a meeting with the leadership team or the finance director
- Before extending new credit to an existing customer
- After a reminder round to assess whether reminders hit the right customers
- When a controller or managing director asks "Which of our customers are a risk right now?"
The skill isn’t intended for single-customer analysis such as "How does customer X pay?". For that, ask Claude directly without invoking the skill.
Prerequisites
- You need to know the
companyNofor the client to be analysed.
Using the skill
You can type your prompts in English or Norwegian - the reply comes back in the same language.
companyNo.
| Desired outcome | Prompt |
|---|---|
| Standard analysis | which customers are paying slower at [companyNo]? |
| Stricter threshold | watch list over 0.5 for [companyNo] |
| Shorter period (quarterly) | run drift on [companyNo] with 3-month windows |
| Historical period | use 2023-09-07 as anchor for [companyNo] |
| High-volume customers only | [companyNo] with minimum 25 invoices per window |
| Anonymised (for demos or external reports) | ... anonymize customer names |
Example: Which customers are paying slower at 7844257 - use 2023-09-07 as
anchor
Understanding the result
The watch list at the top is the most important - the few customers drifting most. The rest of the table shows the full picture, if you’re interested in the details.
| Parameter | Description | Comment |
|---|---|---|
| Drift score | Composite metric | Above +0.30 = worth looking at; above +0.70 = call or meet the customer. |
| Δ DSO (days) | Change in average days from invoice date to payment date | +10 means the customer takes 10 days longer to pay than in the prior period. |
| Δ Late% (pp) | Change in share of invoices paid after the due date | +25 pp means 25 percentage points more often late. |
| Δ Volume | Change in invoice count | A volume drop often means the customer is winding down activity. |
Specific signals to look for:
- 100% late-rate in the recent window - state change, call directly
- Volume drop PLUS rising DSO - double signal; the customer is likely on their way out
- Rising credit-note ratio - drift score may be underestimated; actual behavior could be worse
- Improvers (negative drift score) can also be worth a quick look. If a customer suddenly pays 22 days faster, someone in their accounting probably switched to OCR or Visma AutoCollect - worth telling sales about.
The result is delivered in a markdown table - if you want a file to bring to a meeting, you can ask for "...and deliver as Excel".