Friday, 20 February 2009

Dashboard danger

The world has been seduced by dashboards. At a glance, in colour, and massively visual - dashboards are a successful concept that has gained huge popularity. Executives want them, and IT departments have done their best to oblige.

But dashboards are a great deal more complex than their glamorous surface would have us believe. The raw material that is fed into a dashboard is a bunch of numbers. Numbers about sales, numbers about customer visits, numbers about a range of things that are supposedly important to the business. I say supposedly, because there are two ways of approaching a dashboard – with one being vastly more popular than the other.

The first is to ask what figures are available, and how can they best be represented on a dashboard. Debates then start about whether to use a pie chart or a line graph, whether it should be just the UK or all EMEA, how much detail should be included, etc, etc. These are mostly closed-loop discussions that often lead to a project, and a dashboard.

The second is to ask what the business is trying to achieve, and how it can best be supported in its ambitions. Debates then start about strategy, difficulties in getting information which is genuinely useful, and what the figures really mean. These are mostly open-loop discussions that often lead to more arguments and a general uncomfortable-ness about problems that people feel will take too long to sort out.

I recall a case study about a bank that is now far too controversial to name. They went on record as saying that opening up the strategy debate would be too time consuming, and so they would just take the measures they had and report on those. That particular bank has caused more problems to the UK economy than most of the population is happy with. At the time its management was hailed as a great success: we can now see that they were expert in pushing problems under some very expensive carpets.

Visuals do help – but the numbers beneath the visuals have to be relevant, accurate, and meaningful. A sale is just a sale, right? Well, maybe. It’s not a sale if revenue is recognised too early, or you don’t get paid.

I suggest it is better to open some messy discussions about what the business needs to create real customer value, and which numbers will help to manage that process. Only then is it time to represent your solid and well thought out data in a dashboard. You won’t do it in a day, but you will have a better dashboard, and a better business.

Projects I’ve worked on where the upfront work has been done have produced excellent results. Work needs to be done to keep everyone on side, but in my experience it is well worthwhile. Management must trust the underlying data, or they won’t trust the pie chart that sits on top.

In these difficult economic times, companies need to be asking where value is being created for customers and what resources should be expended to provide the best returns. More than ever it is a time for good thinking, and asking the right questions.

Time for a dashboard redesign?

1 comment:

  1. Caroline

    100% agree, measures need to support strategic aims and initiatives. Sometimes I believe it is appropriate to start an initaitive without the measure being in place to avoid measures for measures sake.

    There is always the risk that the new measure created for the initative is so complex that no one buys in, keep it simple and make it the one the business buys into, EVA & economic profit was popular theoretically but did businesses use it to manage use of capital - not usually.

    The other key point I would offer is that the dashboard should feature some thoughts, comments and analysis - the so what. The little bit in top LH corner of your example I believe. I want to see as with project reports that rather than saying the numbers show that sales are below target, I want to know why and what is being done to improve or is target wrong even.

    Also it should be possible to see at a glance the status Red,green or amber maybe.

    Final thought does a dashboard imply rather a set of instruments? maybe a little reactive? maybe we need more forward forecasting sensing window?

    Robert

    ReplyDelete