Less Talk More Action in Advanced Analytics

Does it seem like leadership is begging for data-driven decisions, AI, Machine learning and Advanced Analytics, but it’s all talk?  A culture shift like this takes education and change management, starting from the top.  Here are some of the top excuses when it comes to Data Science Initiatives and how to help change their mind.

 

1. “That’s not in our budget.”

Identifying an outcome that has ROI or cost savings is helpful to stand up against the inevitable budget questions. 

For example, identifying product cross-sell opportunities may increase current customer value by 3%.  Over even 100 customers this may add up pretty quickly. 

 

2. “Our clients/shareholders/board of directors will never go for it.”

Getting a data science initiative approved is not like purchasing software.  A single piece of software isn’t going to get you answers to all your business questions. Shift your language:

You aren’t “buying a tool”, you’re “buying a competency.” 

This is an investment in creating the foundation to answer important questions.  Those questions and outcomes should directly align with leadership’s goals/roadmap.

For example, an analytics tool won’t give you any predictions. However,  an analytics tool combined with analysts that know how to use it could build an analysis that predicts potential churn in your most valuable customers?

 

“But we already tried that once…and failed!”

Data Science is changing quickly.  New tools are making it more accessible every day to gather & analyze data, share insights and implement outcomes faster.  While the previous attempts may have “failed” they provide great insight into the organizational gaps that need to be handled for a successful implementation.

I can’t tell you how many times we help companies correct this example: Your data scientist created a great predictive model that nobody trusts. We correct this by investing in collaboration tools and including team members with deep business knowledge that will not only accelerate the process but ultimately make a more accurate model.

Do any of these sound familiar? If so, we can help. Let's chat.