analytics teamFollowing on from tips for your research and database marketing teams, this week we turn our attention to the 3rd quadrant of customer insight, your analytics team.

Given the emphasis of advertising & events for this community, it’s tempting to focus on technical skills.

So much social media airtime is given over to new software, coding skills or data solutions, you would think that was all that was needed for effective analytics teams. But there is so much more to developing analysts who will make an impact in your business.

So, without further ado, here are 3 top tips for maximizing the impact of your analytics team (mostly lessons learnt from getting this wrong first time)…

Tip 1: Train your analysts in story-telling skills:

This tip is a reiteration of a learning point I’ve shared before. Over the last decade, I have seen more value driven from analytics due to it being communicated with a clear & compelling narrative than from sophisticated statistics, innovative SQL coding or use of new data. It can take time to master and to learn what truly insightful analytical output should look like, but analysts will only learn by doing (in my experience). The first barrier to overcome for analysts is the temptation to think of their work as completed once the numbers have been crunched, appropriate statistics applied and conclusion reached (at least statistically). However, this can leave a business customer with a pile of Excel graphs & tables pasted into a Powerpoint template to wade through, which rarely if ever helps to communicate a clear point.

Rather, an analyst should be trained to see her true final output as a compelling story communicated in as few slides as necessary – learning to cope with the twin pains of simplifying (when self-esteem can come from a feeling of mastering complex problems) and summarising (when their can be a desire to reflect the weeks of effort in the number of slides produced). However painful this may be at first, the effort is so worthwhile. Simplifying and summarising are key to clarifying the key points to be made in a clear enough way to be shaped into a story. It should be emphasised that the real goal is action, to improve customer experiences & increase profits, so the only information that really matters is that which decision makers need to know in order to instigate action.

In his classic work, “The Leaders Guide to Storytelling”, Stephen Denning lists 8 story templates that will be used most often by leaders. Most of the time, I’ve found the number of templates needed for analysts is even shorter. Using the phrases that Stephen lists, I’d say the three most often needed are:

  • “What if…” (identifying an opportunity for more value creation);
  • “Gosh! We better watch out for that…” (identifying a ‘burning platform’/problem which must be fixed);
  • “When do we start?” (envisioning business on areas where insight could add value if applied).

Tip 2: Allow time for thorough data exploration:

As Big Data and Predictive Analytics are the dominant themes at conferences and in the marketing press, it’s easy to lose sight of some of the methodological basics. I’ve found that many analysts don’t work to a clear methodology, being guided more by the software they use or how they were trained to code. This can cause data preparation to either be neglected or simply tolerated as a necessary evil in order to get the data you need into an analysis tool, so that the fun work can begin and more valued stages like model building or visualisation can be started. But this mindset misses an important opportunity. Data preparation is not simply a stage that is needed to help an analyst cope with dirty data, rather it is an opportunity for discovery and ensuring data quality as a firm foundation for all future work.

Whilst checking for outliers and appropriately categorising or grouping variables, an observant analyst has a chance to explore this new data topography and keep their eyes open for unexpected relationships or data anomalies which may provide new perspectives on a problem (or highlight other problems). Here the job title of ‘data scientist’ really fits, an analyst needs to develop the restless curiosity of a scientist and with the rigour of applying the scientific method analyse these new data ‘elements’ to ensure these are fully understood before embarking on sophisticated analysis in powerful tools.

My earliest analytical background, over twenty years ago now, was data mining and artificial intelligence rather than statistics. Over the subsequent years I have learned much from those trained as statisticians, but I still feel some of the methodological progress made during the development of ‘data mining’ as a discipline can serve analysts well. CRISP-DM or SEMMA are useful methodologies and both emphasise the importance of Exploratory Data Analysis as a stage. For more on both methodologies, Jess Hampton has published a useful comparison which highlights the key stages.

Tip 3: Mentor your analysts in data visualisation skills:

Visual communication is so powerful and as many studies have shown more rapidly and accurately understood by most audiences. However, learning the skills to communicate effectively visually is not a quick fix, in fact it is more akin to learning a craft. The first priority is to think this way, for an analyst to take time to appreciate how much more effective a visually clear & compelling graphic can be than loads of numbers on a page or even just a cut & paste of Excel graphs. A number of approaches can help with this learning stage, if you already have experienced analysts or have good visualisation examples from other companies (or suppliers) then these can be worth sharing in a workshop setting.

As Graeme McDermott has commented previously, he has also seen benefit from Guardian’s Masterclass on Infographics. However, I would want to stress that fancy infographics using specialist skills are not the only way to achieve this visual up-skilling. Modern versions of PowerPoint or Keynote have powerful graphical capabilities, especially when combined with the graphing capabilities within Excel or Numbers. So, often it is a matter of just overcoming the temptation to automate production of slides into a template and spend time to sit down and think about the most effective way to visually communicate your information. When viewed as an artistic endeavour or craft, which is locally recognised & praised, a data scientist can come to relish and enjoy this (even if it is a more manual stage). For some it will give expression to an artistic side of their skills which they had not previously acknowledged.

Two other aids which I have found useful when mentoring analysts to strength this aspect of their technical skills are the work of Edward Tufte and agencies who freely share creative infographics which inspire new ideas. Prof Edward Tufte is not only a very capable statistician and computer scientist, he has dedicated much of his life to promoting better information design or visual literacy. He is well worth reading especially on the topic of the ethics of data visualisation (avoiding misrepresenting facts through misleading visualizations – which is more often done than we may realise). Two agencies which I have found frequently share interesting infographics or other visuals which can spark creative ideas for in-house analysts are Visual.ly and Home (although I’m sure there are many others, so please do share your tips).

I hope that was helpful. The above list is by no means exhaustive and I’d love to hear your thoughts on what has helped. But I hope it gives you food for thought & perhaps help to compliment the focus on software tools & qualifications which can somewhat dominate advise for analytics teams.

Do please also share what has worked well for your analytics team.