It is critical to give social impact partners the flexibility to test, learn and adapt, to find what works to meet goals.

The Ministry of Justice launched the world’s first Social Impact Bond (SIB) in 2010, in partnership with Social Finance. Since then, more than 100 SIBs have been launched around the globe. They are pushing not only our traditional structures and ways of working but also our thinking: social investment demands an articulation of and focus on outcomes.Social impact investment in the public sector is still in early days but the evidence suggests that tools like SIBs could be an effective way to deliver outcomes more efficiently and more effectively, improving accountability—by joining up disconnected parts of the value chain—and increasing the quality of data and transparency – by necessarily including those things from the start.

The presence of investors and innovative commissioning structures also introduces elements of risk, which need to be managed.

To mitigate risk and ensure maximum impact, I recommend a few guidelines:

It is important to set specific, measurable, achievable, realistic and time-bound outcomes but also ones that are not overly complicated to measure. One way in which this can be achieved is by looking at all available evidence and then setting fixed (not relative) outcome targets or using interim indicators (which are correlated with desired outcomes) if such indicators offer clarity, simplicity and meaningful indication of results.

Risk can be mitigated by working with stakeholders with aligned values and motivations.

It is also critical to give social impact partners the flexibility to test, learn and adapt so as to find what works to meet goals. In a structure focused on outcomes, such flexibility is key for actually achieving those outcomes, by driving innovation.

Developing authentic relationships of trust with the investment community and stakeholders is absolutely key. By shifting from a service provider model (‘Here’s my list – can you do it?’) to a collaborative development model (‘What problem are we trying to solve? Think about it with us’), challenges can be addressed more openly, and risk goes down. In fact, I recommend being more open about challenges than one would normally be comfortable with.

Given that repayment of investment is contingent upon results, metrics for evaluation in theory would be built into any model from the start. However, it is worth thinking carefully up front about ways to track and learn from key variables. If, at the end, one simply learns, “this model worked”, one may miss crucial learnings as to why it worked and which elements should be repeated and scaled, and which ones you wouldn’t. In short: start with the end in mind, get crystal clear on what you want to achieve, work backward as to what you want to learn, and build your evaluation strategy (with experts) from the start.