AI Use-Case Prioritization Two-by-Two
Rank AI ideas by market impact and internal breadth before assigning resources
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 98%
Collect proposed AI applications through a lightweight intake process, then evaluate each one against two dimensions. The vertical axis estimates potential impact on acquisition, demand generation, or brand awareness; the horizontal axis estimates breadth of internal use and the resulting productivity gains. Apply rough high, medium, or low ratings rather than delaying for precise forecasts. Plot the highest-potential ideas, stack-rank approximately 10 to 15 candidates, and allocate dedicated technical resources in that order. Keep the ranking dynamic through recurring reviews so that new technologies, market changes, and newly discovered opportunities can displace earlier priorities without requiring a complete planning reset.
Origin
Developed by HubSpot's marketing technology team to prioritize more than one hundred proposed AI marketing experiments, as described on Marketing Against The Grain.
Core principles
- 01Finite resources require explicit prioritization.
- 02Potential demand impact and internal productivity measure different forms of value.
- 03Rough estimates are sufficient for an initial ranking.
- 04Priorities should change when technology or market conditions change.
How to run it
- 1
Open a lightweight intake
Invite the entire relevant team to submit AI use cases through a simple channel such as Slack and a form. Optimize the intake for speed and participation rather than procedural perfection.
Pro tip Ask submitters to identify the expected user, outcome, and affected metric.
Watch out A complicated submission process can suppress useful frontline ideas.
- 2
Estimate external impact
Give each idea a rough high, medium, or low rating for its likely effect on demand generation, acquisition, or brand awareness.
Pro tip Use directional estimates when reliable historical data does not exist.
Watch out Do not disguise unsupported precision as certainty.
- 3
Estimate breadth of use
Rate how much of the internal team could use the capability and how much productivity it might unlock.
Pro tip Distinguish a narrow specialist tool from a capability reusable across many roles.
Watch out Broad adoption does not automatically imply meaningful value.
- 4
Plot and shortlist
Place the strongest ideas on the two-by-two and select roughly 10 to 15 for leadership review. Focus first on ideas that score highly on the dimensions that matter most.
Pro tip Retain the complete scored backlog so lower-ranked ideas are not lost.
Watch out Do not force every idea into the shortlist merely because it has an enthusiastic sponsor.
- 5
Set the order of operations
Stack-rank the shortlist and decide which experiments receive resources first. Make the trade-offs explicit.
Pro tip Assign a named owner and technical capacity as soon as an idea is selected.
Watch out Prioritization without resource assignment creates a list rather than execution.
- 6
Review priorities repeatedly
Reconvene on a regular cadence to account for new technologies, macro changes, and evidence from active experiments.
Pro tip Use a two-week cadence when the technology landscape is changing rapidly.
Watch out Do not treat the initial ranking as permanent.
In the wild
A marketing team receives more than one hundred AI ideas. It rapidly scores each for likely top-of-funnel impact and breadth of internal use, plots the strongest candidates, and selects AI-powered email and website chat from the high-impact quadrant. A centralized marketing technology team then receives dedicated resources to execute them.
→ The team converts a large unstructured backlog into a focused portfolio of high-potential experiments.
Common mistakes
Waiting for perfect estimates
The framework is intended to create direction quickly. Excessive forecasting delays the learning that only real experiments can provide.
Ignoring execution capacity
A highly ranked idea produces no result unless the team secures the technical skills and ownership needed to implement it.
Freezing the first ranking
AI capabilities and market conditions change quickly, so a static priority list will become obsolete.
Is it for you?
Best for
It is best for teams with many proposed AI use cases but limited technical capacity.
Not ideal for
It is not ideal when regulatory risk, implementation cost, or data readiness must dominate the initial decision.
From the transcript
“And so we put together a two by two that really plotted what we felt were two very important criteria to measure these use cases…”
“On the why, we had the impact that a use case might have on top of funnel metrics.”
“And then on the X, we had breadth of use, meaning what size of the marketing team could really leverage this use case and how…”
From the episode
The Ai Strategy That Increased Our Email Conversion Rate By 82%