How to apply data science to marketing and sales
We live in an age where driverless cars will soon fill our streets, Siri is on every iPhone, traders rely on algorithms, Alexa runs our smart homes and the mass automation of labour will impact everyone. The McKinsey Global Institute estimates the total potential impact of artificial intelligence at around $13 trillion in additional economic output. Data analytics capabilities have become essential.
Given such numbers, plus the omnipresent media buzz surrounding data, the call for today’s marketers and sales professionals is clear. Borrowing the words of Andy Grove of Intel: “Only the paranoid will survive!”
The message to “use data” might have reached many sales and marketing departments - but the ‘how’ is less obvious. As a start, I’ve laid out a step-by-step approach that can help unlock the value of data for your organisation, without it becoming a project of Ben Hur proportions.
Applied statistics and machine learning have a lot to offer
Given the repetitive nature of sales and marketing, there are many opportunities for data science to add value across the function, but some are easier to unlock than others. As the saying goes, “Success breeds success” so it makes sense to start with an area of highest potential and ease to gain buy-in for more ambitious projects further down-the-line.
Sales and Operational Planning (i.e. forecasting accuracy) is often the best place to start, and the one area where it is hard for man to beat machine consistently over time (assuming the data quality is there). This doesn’t necessarily mean that no human intervention is required, but if the two are combined in the right way and a learning loop is established, it will be the easiest place to quickly unlock significant improvements.
One added benefit is that the accuracy delivered here feeds into a lot of analyses for other areas. This is because its main deliverable provides an accurate baseline which then serves as the foundation of promotional effectiveness and pack/price architecture. It’s also a cross-functional area where sales, marketing, supply and finance come together with a joint incentive to get it right. Which makes it a great place to start as well.
Take it one step at a time (but with a medium-term perspective)
An analytics capability that’s gradually built in the right way can liberate a business. Ineffective noise (and often internally-directed emotional energy) is replaced by a common understanding and acceptance of the facts. To unlock the value of data in a business in a pragmatic manner, three main elements are required:
1. Clarity on the question(s) to ask
It is surprising how often businesses start a journey into data science without being clear on the questions they would like to see answered. The very first step must be to understand what business questions need answers.
To start this process, begin with a simple list separated into three categories: (a) things we know, (b) things we think we know and © things that we really need to know. When this is done well, often you’ll be humbled by how many things are in the “we think we know” and “need to know” bucket compared to the first category.
2. Make it a central commercial effort (not just left to IT)
The greatest pitfall of any data science initiative is that it is “promoted” to a major IT project. There is a mistaken focus on pulling together data into well-structured databases instead of addressing the initial list of business questions. Unfortunately, this often results in another failed IT project, with the resultant hit to motivation and an expensive invoice to pay.
A smaller, multi-functional team will prove more valuable. Working together with the commercial leadership through iterative cycles of key questions, hypotheses, aligned analytical approaches and insights generated. Admittedly, this is a much more confronting approach as it involves senior people getting engaged and joining the problem-solving cycle. However, it also ensures that any real breakthrough insight generated can be immediately applied.
3. Focus on building the capability and understanding
Finally, building an analytics capability is not a single project or a finite task. Like exercise, it is a journey of continuous improvement, building valuable muscle memory along the way.
In general, there are three groups of customers/stakeholders to keep in mind. First, it is important to build the actual capability in-house over time. Ideally, these are people from the business with an analytical focus who can evolve into capable data analysts. Ideally, with an ability to program basic scripts in R or Python whilst also being able to translate the data into valuable business insights.
The second group considers the leadership that directs the work, sense-checks the outcomes and translates this together with the team into tangible actions. Last but not least, there are the sales and marketing teams themselves. Lethal is a combination where great insights are generated and senior leadership endorses these, but the why and how are not explained in an accessible manner to the team having to do the work.
These steps aren’t overly complex; the challenge comes with consistently keeping them going. For those who stick with it, the value and potential business transformation makes it well worth the effort.
The secret to data science success
Data will become the lifeblood of every good, fact-based business decision, as long as it is well-structured, understood and combined with pertinent questions to drive relevant insights. Companies have become increasingly aware of the need to build a data science capability. Too often, however, this fails - because it’s handed over to the IT department and re-written into a major database and data management project or set up as an external team instead of being woven into the culture of an organisation.
The secret for companies to do this well is evolution. Where, often with some outside help, data science is driven by the key questions to be answered rather than the data that’s available or the urge to be viewed as data innovators.
Data science needs to be insight-driven, capability building and at the heart of the business. In that way, it will be liberating for everyone involved. It clears the “fog of trading” and separates facts from myths. Knowledge is power, and well-executed data science is a powerful force for every company.
Note: The author is grateful for the contribution of Bas Bosma, part-time Professor of Data Science at the University of Tilburg and Managing Director of Simplxr based in the Netherlands