Experiment
Intervention Radar
Where could intervention create the greatest benefit?
01
Problem to address
Policy teams often know the outcome they want and who may be eligible, but have weaker evidence about where intervention would have disproportionate impact. Eligibility and impact are not the same question.
02
What the evidence tells us
Strategy asks where to target interventions for the greatest environmental return. Farmers ask whether the same intervention will work on their land. These are the same question from opposite ends of the system.
03
The proposition to explore
The idea
Help policy teams identify where an intervention, experiment or investment could create the greatest benefit or learning based on need, context and outcome evidence.
04
Hypothesis
If policy teams can combine need, local context, previous intervention performance and outcome evidence, they can target experiments and investment where learning or impact is likely to be greatest.
05
How it could work
Create a prioritisation view over a geography or regulated population. Surface candidate areas/cases based on explicit outcome need and evidence, then allow policy teams to inspect why each is prioritised and what evidence is missing.
06
Example: bring it to life
Rather than asking only which farms are eligible for peat restoration, the radar identifies catchments where restoration need is high, site characteristics are favourable and previous evidence suggests stronger potential return.
07
AI opportunity
AI can combine spatial/contextual evidence, surface candidate areas and explain which signals drive prioritisation. It should support policy judgement rather than automatically allocate funding.
08
Potential value
- More strategic targeting
- Better use of limited public funding
- Faster identification of high-value pilots
- Clearer link between policy intent and local context
09
Reduce uncertainty without overcommitting
Smallest meaningful experiment
Take one planned intervention and manually construct an intervention-radar view for one geography. Test whether it changes pilot-site selection or reveals evidence gaps.
10
Open questions and risks
What we still need to learn
- Can reinforce existing data blind spots
- Impact potential is not always measurable
- Targeting needs fairness and distributional scrutiny