The memo that stopped us building the wrong thing
How a one-page, press-release + FAQ memo forced the real decision before we wrote the PRD, with a worked example and the mistakes to avoid.

We were about to build an invite-flow email verification. The spec was long. The decision was not. People started coding while the choice was still fuzzy. I stopped the work and wrote a decision memo first.
In brief
- Separate the decision memo from the PRD.
- Write a press-release intro and a tight FAQ.
- Force trade-offs and rollback before any code.
What happened
Security asked for verified emails on invites. I opened a doc. It grew. Edge cases. Screens. API notes. The team read parts and formed their own picture. Two sprints in, design was polishing states. Backend had a branch that changed our token model. Support asked when to update help pages.
In standup I heard three different answers to a simple question: who must verify and when. That was on me. The real decision was buried. I paused the work for 48 hours and said we would decide first, then spec.
The steps I took
I drafted a one-page, working-backwards memo: a press-release intro and six FAQ. I used our One-pager outline to keep it short.
I wrote down benefits, the key trade-offs, two success metrics, and a clear rollback trigger. No UI. No API.
I sent it to engineering, design, security, support, and sales. We marked assumptions in the FAQ and resolved owners.
Only after sign-off did I write the technical PRD that mapped scope to tests, APIs, and milestones.
Decision memo vs PRD
| Decision memo | Technical PRD |
|---|---|
| Press-release intro + FAQ | Detailed requirements, flows, APIs |
| Audience: cross-functional stakeholders | Audience: engineering and QA |
| States who, what, why now | Specifies how |
| Includes trade-offs, metrics, rollback | Includes tests, milestones, instrumentation |
| Approved before any build | Written after the decision is signed |
Worked example: invite email verification
Press-release style intro
Headline: Teams can trust that invited users own the email they use to join.
- Reduce account takeovers from mistyped or shared emails.
- Give admins a clear signal that members are verified.
- Keep invites simple for legitimate users.
Trade-offs: We add one step in the invite accept flow. We delay some onboarding cases while email is confirmed.
Success metrics: Stable invite acceptance with a rise in verified-member rate.
Rollback trigger: If acceptance meaningfully drops or critical customers cannot onboard due to verification friction, we revert and review.
FAQ (6)
- Who is impacted?
- All new users who join by invite. Existing members are not forced to re-verify.
- How strict is enforcement?
- Access to the workspace is granted only after the user clicks a link sent to the invited email. Until then, the invite remains pending.
- What data do we need?
- Email address captured at invite and a verification token tied to that email. We log verification status and the time it happened.
- What is the rollout plan?
- Start with new workspaces. Then enable for all new invites. Communicate to admins in product copy and release notes. Support has a short guide.
- What is the dev cost?
- Changes to invite accept flow, token generation and validation, and a flag on the member record. QA covers expired/resent links and duplicate invites.
- What about privacy?
- We send only verification emails triggered by an admin invite. We store minimal metadata on status and timestamps and follow existing retention rules.
What changed
With the memo, the team made the real choices on one page. Security got the assurance they needed. Design cut a whole set of states that the decision made moot. Engineering kept the token model as is. We agreed on what to measure and when to roll back if needed.
The PRD that followed was shorter and clearer. The build was cleaner. No one asked later, why did we do this. The answer was in the memo we all signed.
Mistakes to avoid
Checklist
- Do not treat the memo as marketing fluff. It is a decision, not a slogan.
- Do not let engineering treat the memo as the requirements. The PRD comes after.
- Do not skip rollback criteria. If it hurts, you need a way back.
- Do not keep the memo hidden. Share it so it disciplines stakeholders.
The point
Separate the decision artifact from the implementation spec. Use a press-release intro and a tight FAQ to force trade-offs, metrics, and rollback. Get that signed. Then write the PRD.


