Framework Recommendation Report

Context Analysis:

• Stage: Early-stage startup (50 pilot customers)
• Team: 8 engineers
• Data availability: Minimal (pilot phase, limited metrics)
• Features: 20 MVP ideas
• Timeline: Quarterly planning (3 months)
• Bias detected: 16 out of 20 features marked high-priority (80% inflation)

---

Recommended Framework: ICE (Impact, Confidence, Ease)

Why ICE is optimal for your situation:

✓ Works with limited data

• You don’t have usage metrics yet (only 50 pilot customers)
• Confidence score explicitly acknowledges uncertainty common at early stage
• Impact can be estimated based on pilot customer feedback and interviews
• No “Reach” calculation needed (RICE requires established user base)

✓ Fast to execute

• Score 20 features in 1-2 hours
• Simple 3-factor model avoids analysis paralysis
• No complex calculations needed
• Perfect for MVP velocity

✓ MVP-friendly

• “Ease” factor helps identify quick wins for rapid validation
• Prioritizes learning velocity over precision
• Ideal for product-market fit phase

---

Why NOT other frameworks:

❌ RICE (Reach, Impact, Confidence, Effort)

• Requires “Reach” data (user counts, usage metrics, adoption rates)
• You don’t have this yet with only 50 pilot customers
• Premature quantification leads to false precision

❌ Value/Effort Matrix

• Only 2 factors makes it hard to differentiate 20 features
• Too simple for your volume of ideas

❌ Weighted Scoring

• Adds complexity without benefit at early stage
• Requires defining criteria weights (too subjective without data)

---

CRITICAL: Bias Detection Report

🚩 SEVERE SCORE INFLATION DETECTED

Issue: 80% High-Priority Clustering

Your situation:

• 16 out of 20 features marked high-priority (80%)
• Only 4 features rated medium or low priority (20%)

Healthy distribution should be:

• 20-30% high priority (4-6 features)
• 40-50% medium priority (8-10 features)
• 20-30% low priority (4-6 features)

Why this happened:

Common causes in early-stage startups:

1. Everything feels urgent - Racing to product-market fit creates artificial urgency
2. Fear of missing out - Don’t want to deprioritize the “winning” feature
3. No baseline - Without historical data, everything sounds important
4. Political dynamics - Multiple stakeholders championing their ideas
5. Optimism bias - Early-stage enthusiasm inflates perceived impact

Consequences of score inflation:

❌ Loss of prioritization power - If everything is high priority, nothing is
❌ Team overwhelm - Engineers don’t know what to focus on
❌ Resource diffusion - 8 engineers spread across 16 “priorities” = 0.5 engineer per feature
❌ Missed opportunities - True high-impact features get diluted attention

---

Calibration Framework

Step 1: Define Scale Anchors

Impact Scale (1-10):

• 10 = Game-changing - Doubles activation rate, prevents 50%+ of churn
• 7-9 = High impact - Moves key metric 30-50%, affects majority of users
• 4-6 = Medium impact - Moves key metric 10-20%, quality-of-life improvement
• 1-3 = Low impact - Incremental improvement, affects small segment

Confidence Scale (1-10):

• 8-10 = High confidence - Built similar features before, strong pilot evidence
• 5-7 = Medium confidence - Some pilot feedback, reasonable assumptions
• 1-4 = Low confidence - Pure hypothesis, limited validation

Ease Scale (1-10):

• 8-10 = Very easy - 1-2 weeks, low complexity
• 5-7 = Medium ease - 3-4 weeks, some complexity
• 1-4 = Hard - 6+ weeks, high complexity

Step 2: Force Ranking Exercise

1. Ask: “If you could only build ONE feature this quarter, which would it be?”

   • That’s your only “10” on Impact
   • Everything else must be scored relative to that anchor

2. Ask: “If you had to cut 10 features completely, which would they be?”

   • Those should be scored 1-4 on Impact, not 8-10

3. Ask: “Which 4 features would make the biggest difference to pilot conversions?”

   • Those are your Tier 1 priorities (top 20%)

Step 3: Evidence-Based Questioning

For each of the 16 “high-priority” features, ask:

Impact evidence:

• “How many pilot customers explicitly requested this?” (Actual number, not “several”)
• “What happens if we DON’T build this? Do pilots churn? Or just disappointed?”
• “Is this a blocker for conversions? Or a nice-to-have?”

Confidence evidence:

• “Have we validated this with pilots? How many?”
• “What could make our impact estimate wrong?”

Ease reality check:

• “Does the estimate include design, backend, frontend, testing, deployment, docs?”
• “How long did our last ‘quick’ feature actually take?”

---

Recalibration Process (60-Minute Workshop)

Agenda:

Opening (5 min):
“We have 16 features marked high-priority (80%). Statistically impossible for everything to be urgent. Let’s recalibrate using evidence and force ranking.”

Anchor Setting (15 min):

1. Define what “10” means on each scale (write on whiteboard)
2. Get team agreement on anchors

Force Ranking (15 min):

1. Vote: “If we could only build ONE feature this quarter, which?” (That’s your Impact 10)
2. Vote: “Which 4 features make the biggest difference?” (Those are Tier 1)
3. Vote: “Which features would you cut first?” (Those should be scored 1-4)

Evidence Challenge (15 min):
For each of the 16 “high-priority” features:

• “How many pilots requested this?”
• “Is this a conversion blocker or nice-to-have?”

Re-scoring (10 min):
Based on anchors, force ranking, and evidence:

• Adjust inflated scores downward
• Ensure distribution approaches 20/50/30 (high/medium/low)

---

Expected Outcome After Calibration

Before Calibration:

• 16 features: High priority (80%)
• 4 features: Medium/Low priority (20%)
• Problem: No clear focus, team overwhelm

After Calibration (Realistic):

• 4-6 features: High priority (20-30%) ← Build these in Q1
• 8-10 features: Medium priority (40-50%) ← Defer to Q2
• 4-6 features: Low priority (20-30%) ← Cut or revisit later
• Outcome: Clear focus, achievable roadmap

---

Quarterly Capacity Reality Check

Your team capacity:

• 8 engineers × 3 months = 24 person-months total capacity
• Minus overhead (meetings, support, bugs) = ~18 person-months effective capacity

Realistic feature count:

• If average feature = 3 person-months effort → 6 features max
• If average feature = 2 person-months effort → 9 features max

Implication:
You CANNOT build 16 high-priority features in one quarter. Physics doesn’t allow it.

---

Next Steps

1. Schedule 60-Minute Calibration Workshop

• Invite: PM, engineering lead, 2-3 engineers
• Goal: Re-score all 20 features using ICE with evidence-based calibration
• Output: Realistic priority distribution (20/50/30)

2. Gather Evidence Before Workshop

For each of the 20 features, document:

• Number of pilot customers who requested it (actual count)
• Whether it blocks pilot conversions (yes/no, with evidence)
• Technical complexity and dependencies

3. Apply Force Ranking

During workshop:

• Identify THE ONE most important feature (Impact 10 anchor)
• Identify top 4-6 features for Q1 (Tier 1)
• Be honest about what to defer (Tier 2-4)

---

Key Takeaway

The problem isn’t that you have 20 ideas. The problem is that 16 are marked “high priority.”

Reality: In early-stage MVP with limited data, you likely have:

• 2-3 TRUE high-priority features (game-changing, conversion blockers)
• 5-7 GOOD features (valuable but not urgent)
• 8-10 NICE-TO-HAVE features (defer or cut)

The calibration process forces honest acknowledgment of this reality.

Your Q1 should focus on the 2-3 TRUE high-priority features. Build them deeply, validate thoroughly, learn fast.