Alpha and beta

Alpha is internal, or a handful of trusted design partners who will tolerate sharp edges. Beta is a bounded set of external users who expect roughness but also expect you to listen. Both exist to cut launch risk: find usability cliffs, validate messaging, collect quotes, and prove the metrics move before you burn attention.

The big idea

Beta is a risk-reduction program, not a feature voting booth. If three users hit the same onboarding cliff, that is a launch blocker even if twenty others asked for a different chart type.

A workable beta for a B2B feature:

1. Recruit 15–40 accounts across at least two ICP segments — not just your friends. Bias toward people who will give weekly feedback and (for T1) might write a case study.
2. Timebox 2–8 weeks. Open-ended betas rot into a forever soft launch.
3. Instrument the same events you will use at GA (product analytics).
4. Talk weekly to a rotating subset. Five 30-minute calls beat fifty unread survey responses.
5. Triage ruthlessly. P0 blockers before GA; document known limitations; park nice-to-haves for after launch.

Collect feedback in one place with one owner. Tag each item: bug, usability, request, positioning.

Walking through scheduled reports

You ship scheduled reports as T2. Beta of 25 teams: 18 complete setup, 7 abandon at SMTP credentials. Time-to-first-scheduled-report is 40 minutes versus a 10-minute goal.

You build OAuth for Gmail and Outlook before GA, cut the success story from "any SMTP" to "Google Workspace in two clicks," and only then run the launch email. The beta did not invent the feature. It prevented a public activation failure.

Common mistakes

• Friends-only betas. You will not hear the hard objections.
• Open-ended "soft launch." Without a date, nothing becomes GA.
• Treating votes as a roadmap. Count repeated cliffs, not wishlist volume.

Check yourself

Eighteen of 25 beta teams finish setup; seven die at SMTP. What is the launch blocker? The onboarding cliff, not a missing chart type. Fix the path to first value before you announce.

Why timebox a beta to weeks instead of leaving it open? Open-ended betas rot into a forever soft launch and never force a GA decision.

Key takeaways

• Alpha is dogfood. Beta is a bounded external set that you listen to on a clock.
• Instrument the same events you will use at GA, and talk to a rotating subset every week.
• Repeated onboarding cliffs are launch blockers; nice-to-haves wait.

Further learning

The next step after a closed beta is a soft launch. See the full map in launching.