Strategic Bets

Strategic Bets help you test business hypotheses systematically. Define what you’re testing, link metrics to track progress, set success and kill criteria, and make data-driven decisions about whether to continue, pivot, or stop.

What is a Bet?

A bet is a structured hypothesis test. Instead of launching features or initiatives without clear success criteria, you define upfront:

• What you’re testing (the hypothesis)
• How you’ll measure success (linked metrics with targets)
• When you’ll decide (timeframe)
• What would make you stop early (kill criteria)

This prevents zombie projects that drag on without clear outcomes.

Creating a Bet

Navigate to Bets and click New Bet.

Give your bet a clear name that describes the hypothesis. For example: “Adding onboarding emails will improve 7-day retention” or “Premium tier pricing at $99/month is viable”.

Add a description explaining the context, what you expect to happen, and why this matters.

Setting the Timeframe

Choose how long you’ll run the experiment before making a decision:

• 1 Week — Quick tests, minor changes
• 2 Weeks — Feature experiments, A/B tests
• 1 Month — Larger initiatives, behavior changes
• 1 Quarter — Strategic bets, market tests
• Custom — Set a specific decision date

The decision date is when you’ll evaluate the bet’s outcome, not necessarily when you expect to hit targets.

Assigning Ownership

Assign a Team and Owner for the bet. The owner is responsible for tracking progress and making the final call on the outcome.

Initial Confidence

Rate your initial confidence (1-10) in the hypothesis. This helps you track how your thinking evolves. A bet that starts at 8 and ends at 3 tells a different story than one that starts at 3 and ends at 8.

Linking Metrics

After creating a bet, add metrics to track its progress. Click Add Metric to link an existing metric.

For each linked metric, set:

Success Target

The threshold that indicates your hypothesis is validated.

• Minimum — Value must reach or exceed this number (e.g., retention > 40%)
• Maximum — Value must stay at or below this number (e.g., CAC < $50)
• Range — Value must fall within a band

Kill Target

The threshold that indicates you should stop early. This prevents wasting resources on clearly failing experiments.

• Minimum — If value drops to this level, abort (e.g., if retention < 20%, stop)
• Maximum — If value exceeds this level, abort (e.g., if CAC > $150, stop)

When you activate a bet, the starting value of each linked metric is captured automatically. This lets you track progress from the baseline.

Bet Lifecycle

Bets follow a structured lifecycle:

Draft

New bets start as drafts. Use this state to refine the hypothesis, add metrics, and set targets before committing.

Active

Click Activate when you’re ready to run the experiment. The starting values of all linked metrics are captured at this point.

While active, the bet tracks progress toward success targets and monitors kill criteria. You can:

• Pause — Temporarily halt tracking (e.g., during holidays or incidents)
• Resume — Continue a paused bet

Resolved

When you’ve gathered enough data or reached the decision date, resolve the bet:

• Validate — The hypothesis was correct. Document what you learned and next steps.
• Invalidate — The hypothesis was wrong. Document what you learned and what to do instead.
• Pivot — The hypothesis needs adjustment. Create a new bet based on learnings.

When resolving, you can update your final confidence level and add an outcome summary.

Evaluating Results

Progress Tracking

Each linked metric shows:

• Starting value — Captured when the bet was activated
• Current value — Latest metric entry
• Progress percentage — How far toward the success target
• Value change — Delta from starting value

Kill Trigger Warning

If any kill criterion is hit, the bet is flagged. This doesn’t automatically stop the bet — you still make the final call — but it highlights that early termination should be considered.

Success Achievement

When all success criteria are met, the bet is marked as achieving its goals. This doesn’t mean you must validate immediately — you might want to see sustained performance — but it indicates the hypothesis is looking correct.

Pivoting

When a bet reveals unexpected learnings, pivot rather than simply invalidating. This creates a child bet linked to the original, preserving the learning chain.

The new bet inherits the context of its parent, and you can carry forward your final confidence as the starting confidence for the pivot.

Best Practices

Start small. Begin with short timeframes and clear, measurable hypotheses. Build confidence in the process before running quarter-long experiments.

Set honest kill criteria. It’s tempting to set kill thresholds too low (“we’d only stop if retention went to zero”). Set them where you’d genuinely reconsider the approach.

Link specific metrics. Vague bets with no linked metrics become opinion debates. Tie every bet to numbers that will definitively answer the question.

Document learnings. The outcome summary and next action fields are valuable. Future-you will want to know why past-you made certain decisions.

Review regularly. Check active bets at least weekly. Stale bets that nobody’s watching defeat the purpose.

Asking Elimo About Bets

You can ask Elimo (the AI assistant) about your bets:

• “What bets are currently active?”
• “Which bets are overdue for a decision?”
• “Show me bets that have hit their kill criteria”
• “What’s the status of the retention experiment?”

Elimo can retrieve bet information, linked metrics, and progress to help you stay on top of your experiments.