A workforce scheduling platform was losing manager visibility over schedule changes — not because of a bad feature, but because of a skipped onboarding step. Here's how I diagnosed the real problem and designed a better fix.
When managers onboard to the platform they move through a five-step setup flow. Step 4 — defining an approval workflow — determines whether schedule changes made by supervisors go live immediately or wait for manager review.
Without it configured, supervisors could silently edit any shift. Managers had no visibility. The result: payroll errors, compliance violations, and significant customer support overhead cleaning up the fallout.
The product team's proposed fix was straightforward: make Step 4 mandatory. Remove the skip option and block progression until the step is complete.
Before designing anything, I needed to understand why managers were skipping. If the fix doesn't address the root cause, the problem persists — just in a different form.
I identified three causes:
At Step 4, managers have just created their location and imported employees. They haven't thought about supervisors, approval hierarchies, or autonomy levels yet. The context to make this decision doesn't exist yet.
"Define approval workflow" means nothing to a manager. They think in terms of: "Do I want to know when someone changes a shift?" The label creates friction before the user even reads the options.
There's no signal about what happens without this configured. No warning, no example, no mention of payroll or compliance risk. From the manager's perspective, skipping is completely safe.
Mandatory gates work when users have the information needed to complete the step. Here, they often don't. The result is one of two failure modes: managers abandon the onboarding flow entirely, or they pick any option just to unblock themselves — creating a misconfigured workflow that's harder to fix than no configuration at all.
The goal is to ensure every location has a working approval configuration — without forcing managers to make decisions they're not ready to make. The solution has two parts:
Step 4 arrives pre-filled with the safest option: "All changes require your approval." The question is rewritten as plain language. Confirming takes one click. Completion goes from 45 seconds to 3 seconds for most users.
For anyone who still skips, the dashboard shows a persistent warning with a live count of unreviewed changes. The first time a supervisor makes a change, a contextual alert fires — showing the exact change that just went live without review.
Three screens. Each solves a specific part of the problem:
Each choice below was made to solve one of the three root causes — not just to stop the skip.
| Decision | Root cause addressed | Tradeoff considered |
|---|---|---|
| Smart default pre-fills Step 4 | Premature timing + cognitive friction | Users may accept defaults without reading — mitigated by plain-language rewrite |
| Plain-language question replaces jargon | Abstract product language | None significant — strictly an improvement |
| Skip option stays (not mandatory) | Avoids abandonment + misconfiguration | Some locations may go unconfigured — mitigated by safe fallback default |
| Dashboard banner with unreviewed count | Consequence was invisible | Banner fatigue — mitigated by allowing dismissal once step is complete |
| Contextual alert on first supervisor change | Consequence was invisible | Could feel intrusive — framed as informational, not an error |
| Safe fallback: all changes require approval | Protects against worst-case scenario | May frustrate supervisors — surfaced as a setting they can change |