Skip to content Skip to navigation

Analytics Seminar: 10 (or more) Things We Wish Someone Had Told Us About Customer Analytics

Analytics Seminar 18th October 2018 by Mr Mike Sherman

10 (or more) Things We Wish Someone Had Told Us About Customer Analytics


Mr Mike Sherman
Independent marketing, big data and insights consultant

Business today is vastly different than what it was a decade before, driven by changes in the larger social context, especially with regards to the amount of data that is now available. It is no longer the case that customer preferences, habits and decision-making processes were concealed behind black box models; in fact, businesses now have tremendous information about their customers. With such data and tools available to make sense of that data, it is surprising then that managers are not leveraging this phenomenon by making analytic-driven decisions. On the contrary, studies have shown that less than half of all decision makers base their decisions on insights gleaned from customer analytics, and that percentage is only going down year-on-year.

To correctly utilise data to derive insights, however, it is important for managers to take note of some underlying principles behind how analysis should be done on data to benefit the business, and to maintain a higher level perspective on how the results of the analysis benefits the business. First and foremost, it’s important to remain humble, and remember that predictions from data analytics are just that: predictions. The results are not always right, and depending on them like they do, to the exclusion of other tools or information might lead the company down a disastrous path. Just look to Thomas Watson, CEO of IBM who led the growth of the company from 1914 to 1956, who famously predicted that “…There is a world market for maybe five computers”. Projections are often false.

Data analytics must also necessarily serve the higher business objectives. No matter how advanced or technical the statistical analysis, what ultimately matters is the insights generated by the analysis, and that is what clients ultimately require to improve their business. Managers often spend too much time deliberating over the choice of tool when in fact it’s the ultimate result that matters- that’s lesson number two. Following on from that, the next lesson is that the insights must lead to measurable outcomes. More often than not, this represents an increase in revenues. Keeping this in mind and taking a step back, then, managers must consider if the analysis will be meaningful in the decisionmaking process. If the potential revenue increase (or decrease) is too small, then effort could be spent more meaningfully doing something else. If there exists a high degree of certainty in the scenario for managers, similarly, it might be more valuable to model something filled with more unknowns.

Having established the purpose that data analytics serves, the fifth lesson is to know your data, to understand it. This does not imply any measure of how well you know the facts, but rather whether the wider context and underlying reasons behind the data are well-understood. An interesting business case shared by Mr Sherman was a South Korean telecommunications company which was trying to make sense of its customer demographics- the company’s data indicated that an overwhelming majority of its customers were in the agricultural business, when that clearly did not make sense in the context of the larger Korean population-level demographics. It turned out that sales reps were simply ticking off the first option on the list because asking customers about their occupations was sensitive and uncomfortable in the Asian context. This ultimately highlights the need to fully understand your data before any conclusions can be drawn from them.

The next lesson, while known by most but not fully understood and internalised, is that correlation does not imply causality. It is often tempting to draw the most intuitive conclusion from two seemingly connected trends, but there exists a wealth of well-documented examples of how this has backfired. Related to this is the seventh lesson, which is to challenge the metrics. Since it’s important for data analytics to serve the business objective, and different metrics used in the analysis could lead to very different results, it is therefore of paramount importance that the right metric to analyse be picked to produce meaningful impacts. Hence, focus on what will be useful. Again, this is related to the business significance. An example given by Mr Sherman was churn modelling in the context of the telecommunications industry. The problem was that no matter the metric chosen, it couldn’t lead to actionable responses.

Finally, in the context of data analytics it is important for marketing campaigns to be targeted and tailored to the demographic. With everything online, the marginal costs of an additional unit of advertising sent is negligible, so creating sensible slices of the population to market to, and correctly tailoring the messages to these slices is what ultimately drives successful campaigns now. In the ninth lesson, Mr Sherman highlights the importance of testing and continuing to learn with each iteration. Business is in constant flux and learning from previous failures, as well as successes, is what maintains a firm’s competitive advantage.

Even after having done all that, without communicating the value of everything you’ve done to clients, they’d still be in the dark as to exactly how the business is being improved. Doing so in a concise but clear manner should be something you’re able to do for every project out there. Synthesise but don’t summarise the information! Doing so, together with the above ten lessons, will yield a cohesive idea of how consumer analytics should be conducted and the way in which it can ultimately deliver value to your company.

Last updated on 30 Oct 2018 .