Building an approval workflow drivers (and auditors) actually love
Fast, low-friction approval without losing control. The workflow design that cuts friction by 60%.
The two failure modes
Approval workflows for mileage tend to fail in one of two ways:
1. **Bottleneck**: every trip queues for the manager's approval. Managers approve in batches once a week or once a month, drivers wait, the cycle stretches, late submissions accumulate. 2. **Rubber-stamp**: managers approve everything in one click without reading. The audit trail looks complete but no real review happened — and audit findings start showing up.
A workflow that drivers love and auditors trust avoids both. The trick is to differentiate between trips that need real review and trips that don't.
The 80/20 of trips
Across the Quilometragem customer base, ~80% of trips fall into a small number of patterns: regular client visits, recurring office-to-site routes, recurring inter-office travel. These trips don't need scrutiny — they're routine, the route is known, the purpose is the same week after week.
The other ~20% are non-routine: new clients, exceptional routes, longer-than-typical trips, trips outside business hours. These deserve a real look.
A good workflow auto-approves the 80% (with audit-quality recording) and routes the 20% to the manager. This is the *exception-based approval* pattern.
How auto-approval works without losing control
The auto-approval rule fires when a trip matches *all* of these:
- Origin and destination are in the driver's recurring-routes list. - Distance is within ±15% of the historical mean for that route. - The driver has no flagged trips in the last 90 days. - The trip duration is consistent with the route (not 4 hours for a 30-min drive). - Within working hours.
If any condition fails, the trip queues for manual approval. The manager sees only the queue, not the auto-approved 80%.
What the manager sees
The manager's queue should show, per pending trip, in one row:
- Driver, date, route map (with the proposed track). - Distance and reimbursement amount. - Why it's in the queue (which auto-approval condition failed). - A note field for any explanation the driver added. - One-tap approve / reject / request-info.
A queue with 5-15 trips a week per manager is the sweet spot. More than that suggests auto-approval rules are too tight.
What auditors see
The audit log should show, per trip:
- Driver, date, route, distance, amount. - Auto-approval status (auto vs manual). - If manual: approver identity and timestamp. - If auto: the rule that triggered.
Auditors care less about the human-in-the-loop status and more about the documentation completeness. As long as every trip has an auditable approval (auto or manual), audit findings stay clean.
Three patterns to reject
**'Approve all' button**: don't have one. It signals the manager isn't reading. Even if technically the same as approving each, it has different audit appearance.
**Per-day approval**: approving a day's worth of trips at once, with no per-trip review. Same problem as 'approve all'.
**Approval after payroll**: paying first and approving later. Even if the policy says approval is post-pay, the approval shows in the audit log with a timestamp after the payment, looking like ratification rather than review.
Three patterns to add
**Driver-side flagging**: drivers can flag their own trip with a note ('long lunch break, please review'). Flagged trips skip auto-approval and route to the manager with the note. This builds trust and surfaces issues early.
**Approver delegation**: when the primary approver is on leave, the workflow auto-routes to a delegate with one-day setup. Without this, vacations create approval backlogs.
**Weekly digest for managers**: even with exception-based approval, send the manager a Friday summary of the week — auto-approved count, manual approval count, total reimbursed. It keeps the manager in the loop without requiring per-trip attention.
Cadence interaction
The approval workflow plays with the submission cadence (post 105). Best combinations:
- **On-the-fly submission + exception-based approval**: lowest friction overall. Driver submits each trip, system auto-approves 80%, manager handles 20% within 24 hours. Reimbursement on the next pay run. - **Weekly submission + exception-based approval**: works for teams with smaller volumes. Manager spends 15-30 min weekly on the queue. - **Monthly submission + exception-based approval**: not recommended. Auto-approval at month end loses the immediacy benefit; manager queue spikes.
What to measure
- **Auto-approval rate**: target 70-85%. Above 85% suggests rules too loose; below 70% suggests too tight. - **Manager approval lag**: target <24h for the manual queue. - **Driver satisfaction (NPS)**: target +50. - **Audit findings**: target zero per audit cycle.
Migration playbook
1. Audit the last 3 months of trips. Identify the recurring routes per driver. 2. Configure auto-approval rules using the 80/20 of those routes. 3. Pilot with one team for 30 days. Measure all four metrics. 4. Adjust rules to hit the auto-approval target band. 5. Roll out to remaining teams in 2-week waves. 6. Send a Friday digest to managers from week 1. 7. Communicate to drivers: 'Most trips approve instantly; the manager will look at the unusual ones.'
What changes for the manager
A manager who previously spent 90 minutes a week clicking through approvals now spends 15-20 minutes reviewing 5-15 unusual trips, with a richer view per trip. The same total time, much higher review quality, and the manager isn't a bottleneck for routine trips.
Drivers feel the difference too: most reimbursements clear in 24 hours instead of waiting for the weekly batch.
Bottom line
The workflow that drivers and auditors both love is exception-based: auto-approve the 80% routine trips with audit-quality recording, route the 20% non-routine to the manager with rich context. Friday digest keeps the manager informed without per-trip attention. NPS and audit findings move in the right direction together.