When we think about measuring the ‘value’ of something we tend to think like an accountant. That is, we imagine value to be:

1. Easy to measure, like money: monetary value is consistent, fungible, universally accepted and simple. There’s no confusion about what a pound means.
2. A closed loop, like a balance sheet: a business spends money in its own interests. If its activities benefit the wider world that doesn’t get counted, so the only question that matters is whether you made money or not.
3. Primarily concerned with the bottom line: it is certainly beneficial to understand how you increased profits this year but you don’t need to in order to enjoy them.

This works great most of the time for most things.

When it comes to measuring social value, the assumptions underpinning an accountancy mindset don’t hold. 

For one, social return on investment can be far from easy to measure as there is no ‘one currency’. For example, figuring out the ‘exchange rate’ between preventing someone from going to prison and increasing healthy life expectancy is really hard.

It is also not a closed loop. For example, if a non-profit invests in a programme to improve educational outcomes for children in care, it doesn’t reap the benefits when its graduates go on to have better jobs in later life. 

In addition, because of the first two complications, figuring out the ‘bottom line’ – that is, the exact magnitude of impact you had and who it accrued to – can be nightmarishly difficult, even in seemingly simple cases. 

Even more inconveniently, the amount of impact a piece of work might have and your ability to evaluate that impact accurately are not correlated. For example, imagine you’re trying to get more people to install a low-carbon heating technology in their homes. It is relatively easy to evaluate which variant of an email to individual consumers is best at getting them to buy a specific product. This is because you can control who gets exposed to what and directly measure the results through click-through rates. 

Now let’s say you get a well-loved celebrity to push that same product on their TV show. It’s likely you have a much larger audience who are much more persuaded than they would be by the email. But you cannot possibly know how effective the celebrity was because you can’t know who saw the message when or what they did with it. This leaves you in a bind: evaluability shouldn’t determine which intervention you go with, potential impact should. But you won’t be able to say anything about whether that impact potential was reached if evaluability is low.

But measuring and communicating social value ie, impact is important if we are to make the best use of scarce resources. We can’t just give up because the accountancy mindset fails us! 

So what can we do?

Like anything, the key to success is understanding and accepting the binding constraints at play and the trade-offs that they impose.

When it comes to measuring and communicating the results of their work, I think there are three key problems that people working towards social impact wrestle with:

1. Making credible estimates: there are loads of ways you can make a rough -  or even precise -  guess at how much impact you had. But lots of them are nonsense. For example, you can look at unemployment numbers before and after your jobs programme and claim the decrease shows your intervention worked. But the change in trend could have absolutely nothing to do with you. Apart from being disingenuous, this kind of measurement is bad because if you care about social impact you should care a lot about whether you are paying for deadweight - that is, riding the wave of changes that would have happened whether you got out of bed or not. If it turns out that this is the case, then resources can be moved elsewhere (by contrast, a business owner whose product is flying off the shelves because of some wider trend in the world can simply go to bed happy). This means that it’s important to invest in deploying robust causal inference methods to have a shot at giving a credible answer on whether and to what extent your work had an impact.
2. Being comprehensive with your estimates: if you’ve committed to being credible in your estimations of impact,you’re then left with a second issue: how much of your work should you do this for? The ideal answer, of course, would be ‘all of it’; but it’s more expensive to robustly evaluate something than to not bother. And every pound you spend on evaluation is a pound you can’t spend on programming. 
3. Communicating your impact in a clear way: let’s say you run a portfolio of projects and somehow manage to credibly evaluate impact on all of them. It’s unlikely you will have the exact same outcome measures for everything you did so when you try and add them together it’ll turn out to be a real Frankenstein-job on the communications. But much of the time you need to be able to say something people can intuitively grasp about the impact of the work. Whether it’s donors, voters, or the media: if your stakeholders don’t get it you probably won’t be doing it much longer.

These three problems present us with an ‘iron triangle’, or a triple constraint model.

In other words, as shown in the graphic below, you can choose any two but not all three:

To give an example from our work, consider Nesta’s blueprint for halving obesity. Our goal with this work is to clearly convey the effect of a comprehensive package of policy interventions based on credible evidence. The iron triangle of impact measurement dictates that something has to give:

• If we choose to optimise for something credible and comprehensive, it will be extremely hard to convey it clearly enough that someone can digest it at a glance.
• If we choose something credible and clear, it will be hard to go into comprehensive detail.
• If we choose something comprehensive and clear, it’ll likely fall short on credibility (which is why so many seemingly great examples of communicating social impact come from corporate social responsibility portfolios or other public relations-driven elements of a business’ work!).

Feedback from its users suggests we did a great job on the toolkit. This is in part because the work was guided by a clear choice to prioritise credibility of evidence and comprehensiveness. That’s not to say the end product is unclear, just that the reader inevitably has to deploy some cognitive effort to engage with it! 

While iron triangles can feel frustrating, they're actually incredibly helpful as a way to figure out what matters and set expectations with stakeholders, as long as you can agree which two constraints you’re going to choose up front.