Don’t let your data science become a ghost of Christmas past

Increasingly, we are approached by organisations who started out on an exciting journey in analytics and data science, but whose progress has faltered, regressed or died completely.

They’ve invested significant sums in their data journey, with some even investing in dedicated innovation labs. But now their programmes are a thing of Christmas past, not because they’re lacking in skills or that the demand isn’t there, it’s because they fail to clearly articulate or quantify the value that data is bringing to the organisation.

We’ve seen a universal struggle in monetising the results from data analysis and data science. Organisations get wrapped up in the excitement of (Christmas) present often with little thought towards the future.

Typically we come across three common threads, where organisations have:

• early support and push for analytics because they believe it’s the right thing to do
• made a huge investment in data and analytics without truly understanding the overall impact and potential
• not determined what their exact return on investment is and as a result are seeing wavering stakeholder support

Our universal solution

When we get approached by new companies wanting to explore what we do, we typically ask three initial questions:

1. What are the top 10 questions the organisation has answered using data and analytics?
2. What is the return on investment for each of these pieces?
3. What is the size of the potential further opportunity that could be realised?

Interestingly very few have been able to confidently share their experience and results with ease - unfortunately, it leaves some scratching their heads. Even where numbers can be shared, they are rarely codified into use and value cases for onward sharing and re-use.

Organisations that ask for help in delivering value from their data avoid it becoming a relic of the past. Those that focus on value too late in their strategy will find themselves trapped in a cycle of challenge, frustration and loss of support, or loss of freedom in experimenting and driving real innovation for their organisations resulting in unfulfilled ambitions and Dickens-level regret.

The first steps in delivering value for the future

In starting to work with such organisations we introduce some simple steps that help unblock this challenge to truly drive and measure value from analytics and data science.

So what are the steps we take?

1. Blending Science with Art – the science (statistics, mathematics, economics) is essential, but so is art (experience, knowledge, storytelling, user design and judgement). A good data team has a blend of these skills and focus on the right conversations to develop business needs, with creative solutions and science. Thinking like a data scientist but acting like a business leader.

2. Develop a consulting mindset – consulting skills can hone an individual’s ability to quickly surface problems and opportunities, turn these into meaningful business questions and find solutions. It is as much about relationships and taking people on the journey with you as it is problem identification. It is also about telling a story that inspires change and drives meaningful results.

3. Start small and grow – mapping out an analytics journey is tough and knowing all the answers up front is - not only impossible - but hard to get support for. Start with small use cases, a clear pipeline of value-driven activity and deliver rapid results to win confidence. We suggest starting your analytics project by focusing initially on the minimum, what is your analytics Minimum Viable Product that will show value? Think: Skunk > Mock-up/Prototype > MVP > Production.

4. Embrace the art of conversations – analytics and data science cannot be transactional otherwise it will be sub-optimal. Commissioning a piece of analysis and then getting the results is dangerous unless it is a very fixed, specific and isolated problem. In 95% of cases, great data science is a conversation between a group of different stakeholders with users of the output at the heart. Like a small group of 6-8 involving: designers, developers, analysts, data scientists and business users / decision-makers engaged in conversation using the ICS 7 steps of The ICS Conversational Analytics TM.

Get key stakeholders involved and ensure users are central to this. These include decision makers, business stakeholders, data scientists, data owners and so forth. Help them all to appreciate each other’s roles and needs such that a user understands the challenges faced by a data scientist, that a data owner understands the value that can be created and the data needs of the analysts, that a data scientist knows the importance and value of the question being asked, that everyone understands how the user will use the solution, analysis, product and design it accordingly.

5. Establish an ongoing value-driven pipeline – building a case for investment based on one question, or even a small number of questions, is risky. Knowing your demand and your audience’s needs is critical and so building a clear pipeline of business questions, with needs, costs and associated value is essential to building a successful, repeatable and scalable model for analytics. Develop a pipeline of initiatives ready and waiting which can be ranked and assessed for value. Commercial, operational and ethical value helps build confidence, but also provides a simple operating model for success.

6. Put the user at the centre – embrace design thinking principles and true agile analytics, where the user is at the heart of the solution. To do this, it is essential to consider the user, the user personas and their unmet user needs. By adopting design thinking principles, the user will always be at the heart of the analytics question, problem and solution and it will be tailored for them and built with them. Making the user part of the analytics discovery and build process will not only create buy-in but ensure that the right solution is designed, and value is created. Adopting fast iterative sprints and involving users, even little and often, will enable the right feedback loops to test the analytics story and use cases.

7. Do more based on value and success – once you have started small and developed some early value and wins – scale – look for re-use and scaling potential on the cases you have already proven. Once again think: Skunk > Mock-up/Prototype > MVP > Production. The key is to lead on change and value returned to the business. Without a change and return, there is little point in any of this, and without articulating it, there is little recognition or chance of growth. Assess value early, quickly and indicatively and then refine and test from there.

8. Accept that it’s hard – data science and analytics in business is a world apart from academic perfection. For instance, data is hard to get to and imperfect. It takes time to access and exploit fully but following the above steps can make it easier to drive value quickly. It also requires discipline, but at the same time needs to embrace the three f’s of data science: fun, freedom (to experiment) and failing (accepting that occasionally you’ll fail fast to learn and then jump ahead). 

Unfortunately, we are educating a future workforce to expect a perfect environment for data science and potentially creating a problem. We have a whole new workforce that want to make an impact, have a purpose and drive change… this is great, but it also needs to accept that elements of the mundane exist too: accessing data, wrangling it into shape, making stuff valuable and re-usable.

Three quick gifts…

 1. Keep it simple, start small and grow a pipeline of opportunities. Enable projects to be started as simple skunk projects, or rapid prototypes and scale quickly from there.

2. Realise value quickly and scale from there – only once value is realised should you worry about scaling to production. Having a simple yet clear route to move prototypes into production, whilst at the same time enabling the user community to get early access to test and use seamlessly throughout is key.

3. Blend Science and Art and a Consulting mindset – embrace problem-solving, judgement, experience, storytelling and design thinking, as much as core analytics, data science and experimentation.

For more information contact us for a chat over a coffee so we can explore with you the power of our approach and share our experiences.