Product function
Opportunity scoring and prioritisation
The scoring model Merjora uses to rank AI opportunities: value, confidence, effort, risk and dependency — applied consistently across every candidate.
Scoring makes opportunities comparable. Each candidate receives the same five scores — value, confidence, effort, risk and dependency — so the ranking reflects a consistent standard instead of whoever argued hardest in the meeting.
What this function does
- Applies one scoring rubric across every candidate opportunity
- Separates value from confidence so a big uncertain number cannot outrank a solid one
- Surfaces dependencies that must be resolved before an opportunity is startable
- Produces a sequenced shortlist, not an undifferentiated backlog
What it needs from you
- The candidate opportunities from discovery
- Any constraints you already know: budget window, regulatory limits, system freezes
- Your appetite for change in each affected team
What you get back
- A scored table with the rationale visible for every score
- A recommended sequence: start now, prepare next, deliberately not yet
- The specific blocker attached to anything held back
How it works
- 01
Score
Value, confidence, effort, risk and dependency are scored on the same scale for every candidate.
- 02
Explain
Each score carries the reason behind it, so it can be challenged rather than trusted blindly.
- 03
Sequence
Dependencies determine order; the highest-value item is not automatically first.
- 04
Park
Opportunities that are not yet viable are documented with the condition that would change that.
What it deliberately does not do
- It does not hide low scores to make a roadmap look fuller
- It does not treat effort estimates as commitments from a delivery team
- It does not rank by hype or by what is easiest to demo
Questions
- What if we disagree with a score?
- Change it. Every score shows its reasoning, and the ranking recalculates. The point is a shared standard, not an unarguable verdict.
- How many opportunities should we run at once?
- For most mid-market companies, one — occasionally two if they touch different teams and systems. Parallel AI projects usually compete for the same scarce internal attention.