How To Prioritize Core Business KPIs To Accelerate Value Realization From Data Science
Zohar Bronfman is the CEO and co-founder of Pecan.ai, a predictive analytics platform for solving business problems .
Machine learning and data science can transform businesses with valuable insights that have a direct impact on profits. However, many companies are reluctant to integrate data science because it is a complicated and expensive endeavor.
Many companies today view their needs, business goals, and data as unique to all other companies. I've worked with companies that wanted to integrate every existing data set and then spent six months validating all of their data.
Actually, your organization is not that unique. Integrating AI-powered predictive modeling into your business processes does not require you to manually review each data set or manually build each model. Organizations looking to integrate predictive modeling can achieve results by starting from the basics and moving forward strategically.
Trust people who know the data and understand the problem.
First, start with the business problem you want to solve. Focus on what matters most and prioritize the use cases that directly impact revenue. Put aside the idea of building a large data science team, ingesting all kinds of data, and modeling your client's "end". This is often a waste of time because customer behavior and market conditions change rapidly. By the time the "ideal" model becomes reality, customers have adopted new tastes and habits, especially in today's rapidly changing environment.
Moving from data insights to business-critical predictions requires putting business first, not data science. Data science initiatives can fail when we ask data scientists what we can learn from data to solve business dilemmas. In a recent Wakefield Research survey of marketing executives, 40% said those who build predictive models don't understand marketing goals, and 38% said data scientists don't ask the right questions about customers.
Data scientists struggle to identify useful "learning outcomes" in data if they don't understand which questions are important. Redesign projects to focus on how you can use the data, not what you can learn from the data. Identify, define, and prioritize specific use cases related to your core business issues.
Trust your marketing, sales, and BI teams—they're asking the right questions and looking for insight to guide business decisions and strategy. That's the goal. In addition, their analysts are immersed in relevant data, unlike data scientists or consultants who may move between departments and projects.
Today, marketing, sales, and BI teams typically lack the skills to go beyond reactive and retrospective data analysis without engaging external data science sources. They use BI dashboards and spreadsheets as a rearview mirror, just showing what's going on. But these data experts are uniquely positioned to formulate forward-thinking questions and proactive actions that effectively run businesses.
Concentrate on your most important information and concerns.
For B2C businesses, be it direct-to-consumer e-commerce businesses, mobile apps or media editors. Most consumer-centric companies use behavioral and operational data similar to standard measurement methods to analyze performance and ask similar questions.
Which customers can unsubscribe in the next X days? Who can become a high-value customer? Which customers will upgrade with personalized offers or buy additional products?
Not only do many companies have similar questions, but many use the same types of data organized in similar ways to solve similar problems. Transaction data, in-app event data, web analytics, or even social media data: these data sets have characteristics common to all companies.
This commonality means we can systematize and automate much of the data preparation and feature engineering, allowing us to generate hundreds of features and determine which ones are suitable to produce accurate predictions.
To focus your efforts on what matters to your business, start with the key metrics you use to track performance and align them with specific goals.
Marketing teams have similar forecasting challenges and needs.
If you are in marketing, you are most likely focused on acquisition, engagement, monetization or retention. Find out how your team or department measures ROI and which segments of the customer journey are you focusing on.
For example, if you are tasked with attracting customers through an advertising campaign, you will most likely be measured by the ad campaign's ability to attract customers that generate X dollars over a given time period. Use your data to predict the future customer value of each campaign. You can then decide which campaigns to duplicate, which campaigns to end and which campaigns to further optimize.
Whether you subscribe to Razor, a mobile gaming, or streaming media business, subscriber lifetime value prediction (pLTV) prediction modeling is similar. You want to understand the value each customer creates within days of purchase and at certain intervals (seven days, 14 days, 30 days, 180 days, etc.) based on your business and relevant purchase frequency. is for your business model. Also, modeling the CAC payback period for each customer is similar for most organizations.
Move forward with a forward-thinking approach to business challenges.
Top KPI values to consider as you start rolling out predictive analytics to your team. Identify the team with the best knowledge of relevant data and metrics. Then ask yourself if the questions guiding your company to achieve those goals are similar to those asked by other companies. If so, there may be no need to build predictive models manually. On the other hand, an automated approach based on the business community can effectively meet your needs. Then start with basic information that can generate predictive insights, rather than consolidating everything available into one comprehensive (and large) project.
Getting started with predictive modeling can seem daunting. But it helps to realize that many organizations are broadly similar in terms of key business KPIs. With that in mind, it suddenly seems so much more interesting to be less unique. Sharing common business challenges can quickly advance your predictive analytics plan and enable you to use your data to achieve important goals.
The Forbes Technology Council is an invite-only community for world-class CIOs, CTOs, and technology leaders. Am I eligible?
Comments
Post a Comment