Design Sprint Prototype Plan: Workbench One-Screen UX Validation

Scenario

Thursday morning 2026-08-06, 09:30 PT. Wednesday closed with Sketch A supervoted all-in-one and an 8-panel storyboard locked. The team invokes tool-design-sprint-prototype-plan before parallel build work begins.

Prototype Role Plan

Role	Owner	Day allocation	Notes
Maker	Ari	09:30-16:30 PT	Owns 8 Figma frames; phone + laptop variants for frames 3-7
Stitcher	Ari (paired with Marcus)	13:00-16:30 PT	Inter-frame interactions; deeplink flow Panel 2->3
Writer	Priya	09:30-12:30 PT	Callout copy (“user-service is red; spreading to checkout”), top-action language, severity wording
Asset Collector	Marcus	09:30-15:00 PT	Anonymized incident-data from 3 real pilot incidents; service-graph for fintech use case
Interviewer	Jin	09:30-12:30 script + 14:00-15:00 mock-run + 15:30-17:00 trial run	Simulated-incident script

Doubling-up: 4-person team; Ari covers Maker + Stitcher.

Prototype Brief

What to build: All 8 storyboard panels as Figma frames; full interaction Panels 1-7; Panel 8 (rollback) is a non-interactive end state. Phone + laptop variants for Panels 3-7.

Fidelity bar: SRE-realistic data (anonymized pilot incidents); on-brand visual design; copy final; interactions trigger within 200ms.

Time: Ari 5 hrs for Figma build; Marcus 3 hrs for data + 2 hrs for interaction-logic edge cases.

NOT being built: Real API aggregation backend; real rollback action (simulated). Onboarding flow before Panel 1 (interviewer manually opens Workbench to Panel 2 state).

Interview Script (Five-Act)

Welcome (5 min)

Hi [name], thanks for joining. I’m Jin. The next 60 min is a conversation; no wrong answers. The prototype you’ll see is an SRE incident-response tool; some things won’t work. Recording today per the consent form. Sound good?

Context (8 min)

1. Tell me about your most recent on-call shift. What incidents did you handle?
2. When PagerDuty fires, what is the first tool you open and why?
3. How long does it typically take you to figure out “what is happening” after the alert?
4. When do you revert to a different tool, and why?

Intro (4 min)

I’m going to put you in a simulated incident scenario. You just got paged at 2:34 AM. You’re going to open Workbench. Tell me what you see, what you think, and what you’d do next. Think out loud.

Tasks (30 min; one simulated incident)

Task 1 (15 min): “It’s 2:34 AM. PagerDuty just fired. Open Workbench. Walk me through what you see and what you’d do.” Probe: “What’s broken? How did you know? How long did that take? Would you stay here or open another tool? Why?”

Task 2 (10 min): “Imagine the incident has been going for 8 minutes. You think you know what’s wrong. Show me how you’d verify your hypothesis using Workbench. Would you need anything else?”

Task 3 (5 min): “Imagine this same Workbench but for a different type of incident: a slow degradation over 3 hours rather than a sudden break. Would the one-screen still work? What would you want different?”

Debrief (10 min)

1. Overall reaction?
2. Would you actually use this at incident time vs reverting to Datadog/Honeycomb/etc.?
3. What would you pay per-seat per month?
4. What would you change?

Trial-Run Checklist

Run with Ari as fake customer 15:30-17:00 PT.

• Prototype reaches all 8 panels
• Phone + laptop variants both work for Panels 3-7
• Anonymized incident data is SRE-realistic; service names + metrics believable
• Inline drill-down (Panel 6) works without source-tool revert
• Jin can complete full script in 55-65 min
• Zoom recording + Otter transcription armed
• Remote observer setup (Zoom breakout)
• Calendar invites confirmed for all 5 + buffer

Participant Confirmation Tracker

All 5 + buffer confirmed Thursday morning.

Decider Checkpoint

• Priya approves role plan
• Priya approves fidelity bar
• Priya approves Five-Act script with simulated-incident Tasks
• Priya approves trial-run gate
• Priya confirms Friday 14:00-18:00 PT Decider window

Signed: Priya (founder, PM), 2026-08-06 09:55 PT.

Build authorized. Ari begins Figma build 10:00 PT.