One of the great challenges for Data Scientists is the need to deliver real value to their organization. This was clearly apparent at the end of November, when over 100 data science and AI leaders came together in San Francisco to share experiences, tackle common challenges, and discuss the role of leading data science & AI within their organizations.
At Chief Data Scientist & Artificial Intelligence Officer, USA this one common theme struck me more than any other; the need for data science & AI to offer real value to the wider organization, and the struggle that this can represent. Many believe that the path to achieving this is through using more complicated algorithms, or achieving greater accuracy in their models, or perhaps by using Artificial Intelligence. And maybe they are right.
However, several speakers highlighted ‘the missing steps’ that focusing on such areas leaves out. Without these crucial elements, you risk allowing potential value to slip through the cracks! I like to see these steps as the ‘bookends’ that should surround each data science project, one before and one after, that ensure the knowledge gained in-between doesn’t simply slip off the bookshelf.
The First Step: Identifying Your Intent, and your ‘Data Status’.
Before beginning any project, you should seek to identify the intent of your analysis, by asking yourself “Do I know what I’m looking for?” Then you should work out whether you have the data you need in order to find it. Anand Iyer, Chief Strategy Officer at WellDoc referred to this as defining your ‘Analysis Intent’ and ‘Data Status’.
Based on these, you can better define the nature of the project you are undertaking, what the value you are trying to create will look like, and what approach you should follow in order to achieve your goal.
Anand shared the above slide, which categories the four areas your project might fall into very neatly. By using this table, you can define your goal (Discover, Adapt, Inform, or Extrapolate), as well as the exact nature of the value that your are trying to create for your business.
Doing so should the first step for any data science project, so resist the temptation to drive right in and ensuring you have a clear goal in mind first.
The Last Step: Communicate! Use the data to tell the story
Once you have first identified the value of your project, you of course go on to undertake the work, and achieve the results. Job done.
Wrong! Your project is only as valuable as the knowledge it imparts to others, so that they can act on it, resulting in meaningful impact within your business, so it is vital to perform the last step and communicate your results to others.
You may think that you already do this, but it only counts if others can truly understand the output of your project. Therefore, you must become a ‘story teller’ – this was the lesson from Bob Filbin, Chief Data Scientist for Crisis Text Line.
Bob introduced the concept of the ‘Data Faucet’. Rather than having a Data Lake which requires self service, which can often mean insights never reach their intended audience, instead insights are delivered straight to the audience in a format they can immediately digest. It’s also very hard to miss as the information goes straight into their internal messaging app, Slack. This means that 90% of staff access this information daily, with no time wasted fishing in a data lake.
So, the last step of your data science project, if you really want to deliver value to your organization, should always be ensuring that insights are received and fully understood.
If you want to join the conversation next year, and learn new approaches to leading the data science and AI efforts for your organizations from those who are doing it day in and day out, be sure to check out the 2018 edition of the event.
The world’s largest community of C-Suite executives
Get access to ‘members-only’ content, VIP conference invites and other updates from the CXO Hangouts community.