What to monitor when a supplier handles prepaid fulfilment
If the pending queue grows, the operator needs to see which orders are in it, how long they’ve been waiting and where to check the original references. Put funding and statement discrepancies in front of finance, and send connection or login failures to the team that can fix them.
| Measure | Review action |
|---|---|
| Pending count and age | Check supplier status by original reference |
| Definite failures | Group by documented reason |
| Unknown outcomes | Reconcile before any repeat action |
| Available funding | Follow agreed funding threshold |
| Statement variance | Assign finance investigation |
By Freepaid · Published
Choose alerts that lead to action
If 3 of 120 orders are still pending, show the age and reference for each. Count a failure only when the supplier has confirmed it.
Set the first alert thresholds using the pilot and the supplier’s written terms. Show staff the references and timestamps behind each alert, with a link to the steps they should follow and the person who can help. Limit access to the dashboard and logs under your data policy.
Make alerts actionable
Show validation errors, confirmed failures, pending orders and unknown outcomes separately. Give staff the internal and supplier references so they can look up the order already in progress.
When the pending queue grows, have operations check a sample of original references. Send connection and authentication errors to engineering. Give finance the funding and statement alerts, with a link to the account evidence.
Use the pilot to learn normal pending times and volumes. Keep full recipient files and credentials out of alerts. Review how often each alert fired before changing its threshold.
Give each alert a threshold and owner
For each alert, record the source, how often it refreshes, the threshold, who gets it and whom they should call if they cannot resolve it. On the dashboard, link totals for submitted, completed, failed, pending and unknown orders to the records behind them.
If 80 orders are complete, 4 are rejected and 2 are pending, show all three counts. Staff looking at 86 total orders still need the two open references and the times they were last checked.
Check whether a spike is limited to one product or time of day. Test that the alert reaches the named operator with enough information to start an investigation.
Check which alerts led to action
Pick a few alerts from the pilot and ask the people who received them what they could do next. If they could not find the original order or did not know who owned it, fix the message or send it to a different person. Date threshold changes so a later reviewer can explain changes in alert volume.