At times my excitement about data leads to some embarrassment. Yet, it’s not my excitement itself that embarrasses me (I don’t think anyone should be self-conscious about showing enthusiasm), but the blank, equivocal, or sometimes confused looks that can greet my excitement. These looks say ‘Are you crazy? Why are you excited about data?’

I accept that to many data is, well… boring. It can be. To others it’s just way too complicated. It’s often made to be. But, as a whole, data excites me for the simple reason that it stimulates creativity.

It’s important to make a couple distinctions before I rev up. The first is that data management is different from data analysis. Data management involves databases, servers, hygiene, etc. and, although hugely important, it can get a bit dry. Data analysis is what excites and inspires me.

The second distinction to make is that I am a marketing data analyst, not a data analyst of another kind. Data can be analysed for a variety of purposes: risk analysis, corporate asset planning, financial modelling, whatever. I am a marketer who analyses data for marketing purposes.

Now, on to the creative muse that is data. I often sit in front of a spreadsheet all day, creating through questioning: what’s the gender split and the age breakdown of customers, and how do these compare to national averages? How many customers have purchased product A or product B, and do those who’ve purchased either have a higher propensity to purchase product C? What’s the average time lapse between purchases – is that impacted by seasonality? Can I uncover relationships between customers by looking at surnames or addresses or phone numbers?

The definition of creativity is ‘the use of the imagination or original ideas’. To most, a spreadsheet of names and numbers is not very interesting, so it takes creativity to find value within it. Yet, generating insights is only part of the creative challenge involved in data-driven marketing. With regards to data, I’ve found that marketers respond best to simplicity. Put another way, they tend to appreciate the whole story more than they do the detailed account of an individual character.

Data analysis involves mathematic formulas and algorithms, and if you forget to carry the one or if you put the X where the Y should be, insights will lack validity. So the detail is important when working with data.

Yet, data analysis isn’t just about the numbers – if it was, then nobody would understand it. I mean, a standard deviation tells me plenty, but if I were to go into a meeting armed only with z-scores I’d be shown the door within two minutes. The detail is important, but so too are the pretty graphs, and that’s what most people just don’t appreciate about data analysis: once the insights are identified statistically, they need to be communicated in English, and that takes considerable creativity.

Here’s an example for you. If I were to segment customers based on their product preferences, I’d need to consider a lot: an appropriate sample size, the types of variables to consider, correlations between variables, degrees of separation between clusters, etc. But when I finally develop the first-class segmentation I know the client is after, should I explain it to him by showing evidence of the above? Only if I want an extremely confused client.

No, what I would show him are pretty yet powerful graphs that demonstrate – from a marketing point of view – that the segments are valid and above all useful. The statistical concepts I considered are not easy to understand let alone explain, so I need to get creative in translating the statistics.

I’d probably start by showing him the size of each segment, explaining that each is manageable (i.e. that none represents 95 percent of the base). I’d then put up graphs comparing percentage value versus percentage volume, average transaction values, and length of tenures with the program. I’d demonstrate how an understanding of these characteristics could prove useful in determining marketing objectives and priorities for each segment.

Finally, I’d prove to him the validity of the segmentation by showing that although only purchasing behaviours were considered.