Last week I spoke at the Transforming Data With Intelligence (TDWI) virtual conference. The topic was the reuse of analytics models – where an analytics model is designed once and reused for multiple purposes.
When done right, the reuse of analytics models can save companies a lot of money. But it requires even more of the tough stuff; good alignment between the teams that build and the teams that use analytics.
In this week’s Tech Tuesday, I will talk about effective ways to facilitate such an alignment and share an example from my personal experience.
Facilitating Teams Alignment
One of the most effective ways to facilitate a good alignment between building teams and analytics teams, is having senior analytics folks “sit-in” on weekly stakeholder meetings. I recommend this for lots of functional teams, such as marketing, sales, and product teams.
When done effectively, the information sharing and team building will skyrocket. What often (quickly) follows is a sense of inclusion, a sense of “we’re all in this together”. I’ve seen it firsthand, many times, and it works wonders for team collaboration and a pleasant work atmosphere.
As I was discussing this sort of symbiotic alignment and inclusion at the conference, I was delighted to see this terrific question come up, from a Vice President running a global team:
[Do you have] any recommendations on getting engagement and changing culture among stakeholders on the process and challenges of ML [machine learning]? The value seems to be clear, but understanding of what’s needed and some business process change is still a major challenge. Any recommendations?
It reminded me of a quote by Theodore Roosevelt; “Nobody cares how much you know until they know how much you care.”
And those words apply wonderfully to this particular question. Success and growth don’t come from simply adding analysts to product meetings, but rather from open discourse.
However, it is important to note that plopping a highly-math-literate person down next to a marketing genius who “just knows” what to do can feel forced. To get around this initial growing pain, there are several things you can do to make it easier for all parties to remove perceived walls and build real bridges.
Removing Walls and Building Bridges
There are a number of ways in which you can stimulate a fruitful collaboration between your building teams and analytics teams. How about organizing a hackathon? This is an opportunity to put folks with multiple disciplines on a single team with a unified purpose.
Another idea is to extend an invitation in two directions: In this case, data scientists could for example be invited to weekly product meetings while the head of product could be invited to weekly high-level analytics meetings.
The important thing here is for each team to show how much they care about the other and realize the value of their complementary skill sets. It’s clear today that product teams often benefit from a data-driven approach. It’s also obvious that without buy-in from the product team, the best data-driven approach will go unused.
Many years ago, I worked as the director of data science for a company that needed company-wide insights in the form of daily dashboards. There were lots of stakeholders with as many needs and preferences in terms of what they needed to see day-to-day to do their jobs efficiently. I hired someone to build the dashboards and the first thing I asked her to do was go around the company with a crayon (so as to not collect too many small details) and ask each person what they needed to see daily to inform their decision making.
She did this and delivered the relevant dashboards to each person. A week later, I asked her to show me a report of who in the company used the dashboards and who didn’t. It’s easy to see this sort of information using Tableau, for example.
The results were unsettling but predictable: more than 50% of the stakeholders weren’t using the dashboards she built. Huh? She did exactly what they asked for!
So I asked her to go to all the stakeholders who weren’t logging in to their charts and say, “Sorry I built the wrong chart. Would you tell me how I can make this better for you?”
This was the key! With a few small adjustments, adoption rose to over 90%. That bit of forced humility was the necessary olive branch. From there, we had the beginning of trust and were able to begin feeding important insights to stakeholders and creating a truly data-driven company.
Alignment starts with inclusion. I have found that when teams feel like they are important to other teams, they align and achieve much more than they might otherwise. And through open communication, teams tend to find ways to work together and achieve more, and this is the root of concepts like analytic model reuse.
Be safe and have a wonderful week!
What Data Scientists Do, According to 35 Data Scientists
Modern data science emerged in tech, from optimizing Google search rankings and LinkedIn recommendations to influencing the headlines Buzzfeed editors run. But it’s poised to transform all sectors, from retail, telecommunications, and agriculture to health, trucking, and the penal system. Yet the terms “data science” and “data scientist” aren’t always easily understood, and are used to describe a wide range of data-related work. Here’s an overview of what data scientists really do, according to 35 data scientists.
Why the Future of ETL is not ELT, but EL(T)
How we store and manage data has completely changed over the last decade. We moved from an Extract, Transform, Load (ETL) world to an Extract, Load, Transform (ELT) world, with companies like Fivetran pushing the trend. However, we don’t think it is going to stop there; ELT is a transition in our mind towards EL(T) (with EL decoupled from T). And to understand this, we need to discern the underlying reasons for this trend, as they might show what’s in store for the future.
Real-Time Interactive Data Visualization Tools
Take a look at these interactive data visualization libraries. Can you do more to make sure the consumers of your data understand the story you want to tell? Maybe interactive data visualization could help. Here are a few examples.