Wednesday, 4 September 2013

An Obsession with Numbers

"84.5% of statistics are made up on the spot!" I may have the percentage wrong, but I believe the line comes from Vic Reeves. It's witty in its own right, but it also reflects our obsession with reporting numbers rather than words.

Numbers just sound more credible. Anyone can say "it's getting better" and give examples in support, but it takes an Expert (whatever that is) to say performance has improved by 12.4%. Only last month for example, the Office for National Statistics informed us that happiness in the UK had increased by 1.1% - I hope your smile is 1.1% wider as a result! A more sombre example: the Mid-Staffordshire saga was largely due to generating the right numbers in entirely the wrong way.

So too with Social Return on Investment (SROI). “£5.25 of social return for each £1 invested” has the air of authority, even if qualified or expressed as a range such as £4.50 - £6.00. Working in this field, I sometimes have to explain to clients that how this value is created should be more important than the figure itself. (There has even been talk of SROI reports abandoning the ratio figure entirely, but we'd never get away with this. People like the numbers too much.)

I'm not saying that numbers don't matter; the danger is that people fail to look beyond these. They say "if you can't measure it you can't manage it" (an old cliché, but true), but numbers alone won't do that managing for you. Knowing it has improved by 12.4% is useless unless you know why, and how you can sustain that improvement. Those who fail to understand this often become experts in making excuses rather than making changes.

It isn't a question of quantitative versus qualitative measures, but how to combine the two. That's why all good performance/evaluation models (e.g. EFQM, SROI, Balanced Scorecard) all have elements of results and ‘enablers’, or theories of change. In various ways, all these methods examine cause and effect – what changes and why.

So next time someone quotes performance data to you, try the "so what?" test. Do they know the story behind those numbers? And more important still, what will they do as a result?