Build on evidence, not enthusiasm.

By Pragati Sinha | Work with Me | #EvidenceOverEnthusiasm

<aside> <img src="/icons/cursor-click_gray.svg" alt="/icons/cursor-click_gray.svg" width="40px" />

Use the Evidence Box when you're time-strapped, on the back foot, or at risk of knee-jerk solutions. Gather and organize proof before you pour time and budget into an idea. Separate facts from assumptions, capture how people actually work (not how slides say they work), and make calmer decisions about what to build, buy, pilot, or kill.

</aside>

Problem Evidence

User Behavior

Solution Tests

01 · SURFACE LAYER: STATED / PROPOSED

<aside> <img src="/icons/conversation_gray.svg" alt="/icons/conversation_gray.svg" width="40px" />

CLAIMS & HYPOTHESES

• What stakeholders say the problem is
• Assumptions driving the request
• Stated goals or desired outcomes </aside>

OBSERVED ACTIONS

• What users actually do, not what they say
• Patterns from observation or interviews
• Frequency and context of behaviors

PROPOSED FIXES

• Early-stage ideas and suggestions
• Vendor recommendations or quick wins
• Scope before validation

02 · HARD LAYER: VERIFIED / QUANTIFIED

HARD SIGNALS & DATA

• Quantitative data confirming the problem
• Error logs, support tickets, NPS scores
• Trend data showing direction and severity

ACUTE BREAKS

• Points where the workflow fails completely
• Where users abandon, escalate, or stop
• Highest-friction handoff moments

VETTED FIXES

• Solutions validated against real evidence
• Prototypes tested with actual users
• Pilots or proof-of-concept results

03 · SYSTEMIC LAYER: STRUCTURAL / INHERITED

INHERITED LIMITS

• Technical debt, legacy constraints
• Org structure or policy forcing functions
• Budget, headcount, or access limitations

HANDOFFS & ESCALATIONS

• Where ownership breaks between teams
• Escalation triggers and gap ownership
• Informal workarounds that fill gaps

GUARDRAILS & RULES

• Compliance, legal, or regulatory constraints
• Non-negotiables that bound the solution space
• Risk thresholds and approval gates

04 · STAKES LAYER: CONSEQUENCE / MEASUREMENT

COST OF INACTION

• What happens if nothing changes
• Revenue risk, churn risk, operational cost
• Reputational or regulatory exposure

COPING MECHANISMS

• Workarounds users have invented themselves
• Shadow systems, spreadsheets, manual steps
• The invisible work that keeps things moving

METRICS TO WATCH

• Leading indicators that signal improvement
• Outcome metrics tied to solution hypothesis
• Guardrail metrics that flag regression

Worked Example

Below is a worked example of this canvas filled in (AI chatbot implementation):

• Use it to see what "good" looks like.