Track Opinion
Download the insights industry's #1 strategic planning tool
SurveyHealthcare: Quant & Qual Healthcare Data Collection
Kick in the pants + turbocharged networking + intensive education = IIeX NA 2019 | Austin, TX | REGISTER NOW

Beyond Chasing Unicorn

Becoming a true data leader begins with building a great data insights team.

It is an exciting time for the Insights Industry. It used to be that traditional Market Research was an island unto itself, carefully designing qualitative and quantitative experiments and providing highly valuable insight that could not be gathered in any other way. Now, there are no boundaries. While primary research still is an important source of information, it now has to exist within a much larger context of widespread data availability. And that data comes in many forms, many of which will not have come from the neat, carefully controlled experimental design that we may have become accustomed.

Coupled with the change in data availability, there has been an explosion in the methods and techniques available to us. Bayesian predictive models, Machine Learning (ML), Artificial Intelligence (AI), Natural Language Processing (NLP), Data Mining… The list is endless. All of these techniques readily and (in many cases) freely available to anyone prepared to invest in the skills needed to use them.

But where are those skills to be found? In short, you need a data insights team.

This brings us to a dilemma facing all forward-thinking companies. What is the best way to bring these capabilities into our business, so we may be competitive and flourish in the new data landscape?

The answer to this question is not a simple one, and there is no one-size-fits-all strategy that will get everyone there. And the costs of getting it wrong can be huge – including threatening the viability of your company.

Many companies start by adopting a naïve “Job Req. First” approach. It goes something like this: The client-facing staff start to notice that clients are demanding more data science and analytics capabilities. They communicate this to the executives, who look around and see that they don’t really have the skills to meet the needs internally, so they task someone with writing a job specification for their first real data scientist. The person tasked does a quick search and produces a specification which includes every buzzword currently floating around the industry. The specification requires that the individual is an expert in statistics, AI, ML, Python, R, JavaScript, SQL, NoSQL, Spark, TensorFlow, Keras, Blockchain and NLP. Oh, and also must have 10 years’ experience using software packages that are only five years old (yes, this is common). Starting salary $50k. PhD Preferred.

Job ads like the above are incredibly common and are the target of much derision amongst data scientists. The common phrase used is that they are looking for ‘Unicorns’ – mythical creatures that don’t exist.

So, what can you do to avoid these mistakes? As is always the case, the first thing you need to do is to take a step back and ask yourself ‘Why?’

Different companies need different levels of data science expertise. Before you can even start to think about your data science talent you need to think about the role data – and its analysis – plays within your company. How do data assets and analysis assets support your company mission and vision? In short, you need to start with a Data Strategy.

A well-formed Data Strategy is the foundation that a data insights team needs to build upon. Without a Data Strategy any attempts to build a team will be wasted. Indeed, before you have a Data Strategy there is no way to know what it even means for your company to have a data insights team.

Surprisingly few companies have a well-articulated Data Strategy. A typical strategy might include:

  • The role(s) data plays in the current and future success of the company
    • Including specific ties to the company’s 3/5-year plans
  • An inventory of current data and analysis assets
  • An assessment of partnership, acquisition and development alternatives
  • A roadmap of how the data and analysis assets evolve
    • Including key decision points and metrics
  • Outline financial models for the value of data

Having built a solid Data Strategy, we have a framework for building the data insights team. Often the choices fall into one of these scenarios:

  • Partner with an external team
  • Build a team from scratch
  • Evolve/refurbish an existing insights team
  • Replace an existing team
  • Some hybrid of the above

All of these strategies can be used, but each comes with costs, risks and trade-offs. Navigating the path forward can be tricky and it is extremely important to have your eyes fully open and have a strong understanding of the risks (and ways to minimize them). The best way to do this is to build a team development/acquisition plan that matches the roadmap for asset development in the Data Strategy.

In our session at Converge, December 4-5 in LA, we explored these approaches and other best practices (and pitfalls) to building a great Data Insights Team. This is a journey that all forward-thinking companies have either embarked upon or need to start. With the right approach and some careful planning, your company can be a true data leader.

Please share...

One response to “Beyond Chasing Unicorn

Join the conversation

Avatar

Christopher Robson

Principal, Deckchair Data Science