I was talking to a potential client recently who was frustrated by the lack of ‘real analysis’ happening in his organisation. He had a team of analysts working for him, most of whom were using Excel when he really wanted them to be using Python. Although we don’t generally recommend Excel for advanced analytics – see my earlier blog on this topic – it’s still a useful tool that can generate real insight and is heavily in use in many organisations.
Quick wins are important
This idea that ‘real analytics’ only happens if you’re using what are, for many people, some pretty hard core coding techniques is something that I’ve come across a few times now. Obviously we understand the urge to build a team of analysts who are all working at the cutting edge, using the latest tools and techniques, but while you push towards that it’s important not to overlook the value of the real insights that can be gained by using much simpler techniques and tools. These kinds of insights are important because they give a project momentum and build morale, as well as demonstrating the value of analytics to the wider organisation.
In the example above, I suggested to the analyst in question that he might consider running a series of decision trees in order to get some quick insights that would have real value for managers. He’d got so hung up on wanting his team of analysts to move from Excel to Python that he risked completely skipping over some very practical easy insight techniques that would add real value to the organisation and prove the benefit of analytics to business users quickly and easily.
It’s very easy to get bogged down in a complex analytics project, particularly if you dig straight into the most complex and sophisticated techniques. If you’re not careful it can be months before you generate any real insights and when that happens you can quickly find that you’re losing goodwill internally. If you’re not careful you can find yourself losing management buy in for your project, and your team can lose morale and momentum.
Simple techniques can still deliver real insight
As an analyst it can be tempting to think that if you’re not building complex models or writing sophisticated code then you’re not adding value, but that’s not the case. A lot of valuable insights can be generated using very simple techniques, particularly if your organisation has not done much in the way of analytics before. It’s also important to bear in mind that what can seem simple and obvious to you as the analyst, can also be a significant insight or a major breakthrough to a business user. Don’t overlook the value that simple techniques and seemingly obvious findings can have to your organisation, particularly in the initial stages of a project.
It’s very tempting to want to go hell for leather when you get the go ahead on a new analytics project. The urge to immediately dive in and start swimming about in the deep end of your data, trying to answer the most complicated questions, is strong. Whilst there’s obviously value and insight there too, and the ultimate aim of your project will almost certainly take you there, don’t forget the importance of coming up with something quick and simple at the start to help keep everyone on board with the project and demonstrate its value to the organisation. Apart from anything else, a small benefit now is much more politically valuable that the promise of a massive benefit at some undetermined point in the future.
Business users care about insight, not methods
I’ve worked on numerous projects where the initial stages of the analysis generate some of the most important insights. If you’re digging about in your organisation’s data for the first time, then even the most basic descriptive statistics will most probably tell people things that they don’t already know about what’s going on in the organisation. The types of techniques you’d typically associated with the data cleaning stage of the project still have value in terms of the information they can provide. Some simple visualisations, even as straightforward as basic pie charts or histograms, can help people understand their data in a way that they have never really been able to before.
So, my advice is not to overlook the value of simple techniques when it comes to generating real insight early on in a project. As an analyst you’ll always want to get out the big guns, but remember that business users don’t see things in the same way. The tools and the techniques that you use are almost irrelevant to them. What they care about is what they can learn about the organisation that will help them to do their jobs better, and whether that nugget of insight comes from a simple crosstab or decision tree or from a complex neural network or predictive algorithm really doesn’t matter to them.
