The other day, as one often does, , I was killing time by scrolling through social media content when I was stopped in my tracks by a post that really resonated with me. “We tend to focus on the business of analytics, when in fact we should be focusing on the analytics of a business.” As someone who has seen firsthand how companies can get caught up in the latest analytics tools, dashboards and complex models – often at the expense of real business impact – I dwelled on this statement for a few days.

It is estimated that humans generate 2.5 quintillion bytes of data every day: that is 18 zeroes or a billion multiplied by another billion. In this landscape, businesses are inundated with analytics-based solutions promising transformative insights. As a result, the ‘business of analytics’ puts the focus on the mechanics – choosing the right software, fine-tuning machine learning models and debating methodologies – rather than on using analytics as a tool to drive tangible business outcomes.

The hype around the ‘analytics of business’ is real and amped up by studies like a PWC survey of 1,000 executives claiming that “data-driven organisations are three times more likely to report significant improvements in decision-making compared to other firms that leverage data less.” However, the sheer volume and complexity of data often leads to analysis paralysis, which hinders effective decision-making. In my experience, when it comes to marketing and behavioural analytics, executives often prefer straightforward, actionable insights that directly inform strategic decisions.

I practice K.I.S.S. – or ‘Keep it Simple Stupid.’ At the end of the day, the best way to convey a message is with a simple percentage, ideally in a manner where everything adds up to 100%. Or if you want to generate extra oomph, throw in that ‘X times’ variable to really send across that message. The reality is that simplification makes any number accessible to audiences, helping them to grasp the scale of an issue much better. Percentages are also easier to use in explaining a behaviour, as opposed to an index score that morphs multiple variables into one number or a data science model that spits out a probability.

Whenever I have worked on studies like the ‘Brand Health Tracker’ with marketing and strategy teams, my observation is that it is far easier to use basic percentages, such as ‘awareness’ or ‘consideration’ as opposed to a ‘brand equity score’ (an index score), which is trickier for the majority of the stakeholders to comprehend. Index scores may be useful for some advanced statistical analysis, but in the end, those outputs are also used to support a narrative that is built on simple KPIs. Ultimately, it is important to recognise that effective data analysis does not necessarily equate to complexity.

Human behaviours are complex and understanding what drives them is a messy proposition. The most complexity that I add to my analysis is to combine two simple KPIs to model customer behaviours. One of the complexities that I add to my analysis is by using answers derived from market research sessions during which we pose questions about the brand (for instance, about its negative attributes) to people who were either using a particular brand or had used it in the past (churned customers) to arrive at the reasons why people stopped using the brand. Similarly, you can create different levels of digital ‘engagement’ in your customer base by looking into how many channels a customer uses (email, website, app etc.) and by adding their recently used channels (last month, last year, etc.). This can potentially allow one to form hypotheses about why customers may be churning, which can then be tested using available data sources.

This said, it is also extremely important to understand the limitations of simplicity. The Net Promoter Score (NPS) is a widely used instrument in boardrooms across the world to measure loyalty. It is also straightforward and easy to measure by using a 0-10 scale. Many companies often use NPS scores to evaluate employee compensation or bonuses. However, the founder of NPS has categorically recommended that NPS scores be delinked from employee bonuses because “only bad things will happen”; in other words, employees start caring less about pleasing customers and more about getting a high rating.

Don’t get me wrong; simplicity does not mean that data science models are not effective. The main issue with data science is usually a lack of understanding about the data that is input into models. Understanding exactly how the data is being sourced leads one to an understanding of exactly how a business functions – and it is this business context that then acts as the foundation that summarises the outputs of the models, leading to simplification for stakeholders. This is critical from a marketing or behavioural perspective, as these data variables are much more subjective and vague as opposed to variables in operations (like inventory) for instance.

Analysts should take time out to talk to the customer or, at the very least, to the customer-facing operatives. Similarly, they should interact with peers and be up-to-date with the latest news about their industry. Peers often help put numbers into perspective and help set the context, and more importantly, prioritise insights and recommendations. I usually prepare a one-liner answer to the most common question at the end of a report: “What should we focus on?”

This leads to the final contradiction: simplicity is an illusion you cast for your audience. Drilling down to ‘simplicity’ often requires an analyst to tackle a lot of data complications in order to distil it into bite-sized portions for stakeholders. It’s a painstaking process to go through data and reprocess it from multiple angles, only to discover that there is yet another lens to look at. Processing the data or writing SQL is the easy part – personally, I spend much more time staring at a blank slide/page debating the order in which I present my data points so that it seems coherent and clear.

The aim of focusing on simplicity is to help drown out the noise and help stakeholders hone in on the next steps. Simplicity also allows one to scale work more easily and make it quicker to reproduce if necessary. At the same time, simplicity also demands an understanding of what the data means, how it was sourced and calculated and where any blind spots may be. So yes, simplicity does require going down the rabbit hole without entering Wonderland.

Ans Khurram is an analytics and insights professional. anskhurram@gmail.com