Skip to content

Requirement Verification Checkpoint (CQ)

Owner: Vi | Updated: 2026-02-05

Purpose

Before building, devs must prove they understand what they're building—in their own words. This catches misunderstandings BEFORE code is written, not after.

The problem: Devs interpret requirements through their own lens, not the intended lens. They build what they understood, not what was meant.


Why This Matters

Without Verification With Verification
Dev assumes they understand Dev proves they understand
Misunderstanding found at review Misunderstanding found before coding
Rework after building Get it right first time
"That's not what I meant" "Yes, that's exactly right"

The lens problem: Devs see requirements through their own experience and assumptions. Verification forces them to articulate through the RIGHT lens.


The Verification Process

Step 1: Read the Task Completely

Before doing anything: - Read the full task description - Read acceptance criteria - Look at reference links - Check related tasks

Don't skim. Actually read.

Step 2: Write a Summary in Your Own Words

In the task comments, write:

## My Understanding

**What I'm building:**
[Explain in your own words—not copy-pasting the task]

**The user problem this solves:**
[Why does the user need this? What are they trying to do?]

**What "done" looks like:**
[How will I know it's complete? What can the user do?]

**What I'm NOT building:**
[Scope boundaries—what's out of scope]

**Questions I have:**
[Anything unclear]

This is not optional. Writing forces clarity.

Step 3: Get Verification

Share your summary and get confirmation:

What to Verify Who Verifies
Product understanding (what and why) Ryan or Erik
Technical approach Senior dev or Vi
Scope and deadline Vi

Why Ryan or Erik for product? Vi sometimes doesn't have full product understanding. Ryan and Erik know what needs to be done. Go to the source.

Step 4: Adjust If Needed

If your understanding was wrong: - Update your summary - Clarify what changed - Get re-verified

Only then: Start building.


Verification Checklist

Before starting Alpha:

  • Read task completely (not just title)
  • Wrote summary in own words
  • Got product understanding verified (Ryan/Erik)
  • Got scope confirmed (Vi)
  • All questions answered
  • Pre-build research complete (see Pre-Build Research SOP)

If any box is unchecked: Don't start coding.


What Good Verification Looks Like

Good Summary

## My Understanding

**What I'm building:**
An email sequence builder where users can create a series of
automated follow-up emails with delays between them.

**The user problem this solves:**
Sales reps want to send follow-up emails without manually
remembering to do it. They set it up once, and it runs automatically.

**What "done" looks like:**
User can: create a sequence, add 3+ emails with delays (1 day, 3 days, etc.),
preview each email, and start the sequence. Emails send automatically.

**What I'm NOT building:**
- Branching logic (if user replies, do X) — that's a future feature
- Email templates library — using plain text for now
- Analytics/tracking — separate task

**Questions:**
1. Should delays be in days only, or also hours?
2. Can user edit a sequence after it starts?

Bad Summary

I'll build the email feature as described.

This proves nothing. Rewrite it.


The Lens Problem

Wrong lens (dev's own perspective): - "I would want to set exact times, not just delays" - "In my experience, branching is essential" - "I think 3 emails is enough"

Right lens (user's perspective, verified): - "Ryan said users just need simple delays for now" - "Erik confirmed branching is phase 2" - "Users in the interview said they typically send 5-7 emails"

How to fix: Always trace your understanding back to a source (Ryan, Erik, user feedback, research). Not your opinion.


When Understanding Doesn't Match

If your summary reveals a mismatch:

Mismatch Type What to Do
Scope larger than deadline allows Discuss with Vi, adjust scope or deadline
Requirements are unclear Ask Ryan/Erik to clarify
You disagree with approach Raise it, propose alternative, but defer to decision
Task seems wrong Speak up, but verify you're not missing context

Speaking up is good. But do it BEFORE building, not after.


Who Verifies What

Type of Verification Go To
"Is this what we want to build?" Ryan or Erik
"Is this how it should work for users?" Ryan or Erik
"Is this the right technical approach?" Senior dev
"Is this scope realistic for deadline?" Vi
"Am I understanding the task correctly?" Whoever wrote the task + Ryan/Erik

Note: Vi is the gatekeeper for process, but for product understanding, Ryan and Erik are the authority.


Consequences for Skipping

If a dev skips verification and builds the wrong thing:

Situation Consequence
First time Rework. Conversation about why verification matters.
Second time Escalate to Ryan. Formal discussion.
Pattern Performance issue.

This wastes everyone's time. Verification takes 30 minutes. Rework takes days.


Integration with Other SOPs

SOP Connection
Pre-Build Research Research informs understanding
User Context Briefing User context helps verify "why"
Task Creation Standards Good tasks make verification easier
Alpha-Beta-Final Verification happens before Alpha

FAQ

Q: What if Vi says it's fine but Ryan disagrees? A: Ryan's understanding of the product takes precedence. Update your approach.

Q: What if no one is available to verify? A: Wait. Don't start building with unverified understanding. Flag it as a blocker.

Q: What if the task is really simple? A: Still write a brief summary. "I'm adding a button that does X. Clicking it will Y." Takes 2 minutes.

Q: What if requirements change after verification? A: That's a scope change. See Scope Change Process. Re-verify if significant.

Q: What if I disagree with the requirements? A: Raise it during verification. Propose alternatives. But once decided, build what was agreed.


Summary

Principle Why
Write summary in own words Forces clarity
Get verified by Ryan/Erik They know what's needed
Prove understanding before building Catch errors early
Use right lens, not own lens Build what's intended
No verification = no coding This is not optional

Prove you understand before you build. Don't assume.