Simple Prioritization Frameworks

Making decisions with incomplete information is a common challenge in software development. While we often want perfect data before acting, waiting too long can lead to missed opportunities and delayed progress. Understanding how to work with partial information helps teams move forward effectively.

Working with Incomplete Data

Software teams frequently face situations where they need to make choices without having all the facts. For example, deciding on technology stack upgrades, estimating project timelines, or choosing which features to build first. The key is finding the right balance between gathering more information and taking action.

Three common scenarios where teams need to act with partial data:

• Market opportunities that require quick responses

• Resource planning with unclear future demands

• Technical decisions before all requirements are known

Simple Decision-Making Frameworks

These basic frameworks help teams make choices when information is limited:

The 40-70 Rule: Make decisions when you have between 40% and 70% of the information. Less than 40% is too risky, while waiting for more than 70% often wastes time.

The Two-List Method: Create two lists - "Known Facts" and "Key Unknowns." If the known facts support a clear direction and the unknowns seem manageable, move forward.

AI prompt example:

Using the Two-List decision model, help address the problem of
a 20% churn increase last month in our mobile productivity app. 

1. Create two short lists: ‘Known Facts’ and ‘Key Unknowns.’
2. Based on this model, suggest the most reasonable next step:
   should the team prioritize onboarding improvements or feature
   enhancements in the next sprint?
3. Keep the response concise (max 3 sentences). 
   Optionally, add a confidence level (High/Medium/Low).

Using Quick Tests to Reduce Uncertainty

When facing unclear choices, small experiments can provide valuable data:

• Build simple prototypes to test technical assumptions

• Release basic features to small user groups

• Try new processes with one team before wider adoption

• Use AI tools to simulate potential user behavior before launch (e.g. LLMs playing the role of different user types)

Common Decision Biases

Watch out for these mental traps when working with incomplete information:

1. Analysis Paralysis: Getting stuck gathering more data instead of taking action.
   Fix: Set clear decision deadlines.

2. Confirmation Bias: Looking only for information that supports your preferred choice.
   Fix: Actively seek contrary evidence.

3. Recency Bias: Giving too much weight to recent events or data.
   Fix: Look at longer-term patterns when available.

   

Bias checker with AI:

Evaluate the following feature proposal for potential decision-making biases,
specifically confirmation bias and recency bias:  
[Insert short description of feature and rationale]  

1. Identify any signs of these biases in the rationale.  
2. Suggest one practical strategy per bias to reduce its impact.  
3. Briefly recommend: move forward, adjust, or gather more data.
   Keep the response concise and actionable.

Real-World Example: Feature Prioritization

A team needed to choose between three features but had limited usage data:

• Feature A: 2 months of data showing moderate usage

• Feature B: Only 2 weeks of data but strong initial interest

• Feature C: No direct data but similar to successful competitor features

Using the Two-List Method:

• Known Facts: Feature A had steady growth, Feature B showed early promise

• Key Unknowns: Long-term usage patterns, full market demand

Decision: The team chose Feature B, setting a 3-month review point to check results. This balanced acting quickly with managing risk.

Key Takeaways

• Perfect information rarely exists - learn to work with what you have

• Use simple frameworks to structure decisions with partial data

• Run small tests to gather quick feedback when possible

• Watch for common biases that can affect judgment

• Set clear timeframes for decisions to avoid endless analysis