Join top executives in San Francisco on July 11-12, to hear how leaders are integrating and optimizing AI investments for success. Learn More
For several years now, pundits have declared that data is more valuable than oil. But are companies really succeeding in extracting the most value from their data? What are some of the hidden costs of gathering and storing data, and how can companies get more from their data?
Storms of data
Today, companies are faced with an enormous amount of data. Collecting, storing and securing that data in a warehouse or data lake comes at a big cost. The pandemic exacerbated the problem by spurring digital transformation and moving the entire buyer’s journey process online. That movement prompted many companies to put increased efforts behind data collection to make sense of a shifting world.
But data in and of itself is not valuable. It’s only valuable when you can use it to understand a shifting world, and capitalize on those shifts to improve your company’s performance, such as by increasing revenue growth, gaining a competitive edge or raising the bar on operational excellence.
An organization may have a pile of gold bricks, but if it doesn’t have a way to turn the gold into cash flow, that gold is essentially worthless. This is the challenge many organizations are facing right now when it comes to data. Many companies are sitting on a gold mine of data. But they have no way of turning it into valuable, prediction-driven insights that could inform the many “million-dollar” decisions and actions that revenue teams make every day.
Join us in San Francisco on July 11-12, where top executives will share how they have integrated and optimized AI investments for success and avoided common pitfalls.
By prediction-driven insights, I mean the type of algorithmically-derived, probabilistic information that can help guide daily actions and predict what is highly likely to happen in the future — and most importantly, make an outsized impact on the bottom line. Today, most companies analyze their marketing data by focusing on the past: What did this segment of people do in the last quarter or the same period last year? But to shift from historical analysis to prediction-driven intelligence, the underlying question needs to be reframed as: Which specific individuals are most likely to do something in the future?
Predictive insights: Using data to look ahead
This shift to a predictive mindset gives a marketing person a lot to work with and numerous possible insights. They could create a personalized offer to influence the customer’s behavior to shift course or take an action sooner. They can also create much more accurate lookalike audiences, making their targeting more precise, or expand audiences in a highly strategic way by focusing on lookalikes of future high-value customers. Another option is to predict which customers are likely to churn and take action to try to retain them before they leave. Even a small increase in customer retention can provide a dramatic boost to profits.
Say you’re a large D2C lifestyle subscription brand spending millions of dollars a month on acquisition campaigns. You’re also likely offering your potential new customers significant discounts on their first order, and perhaps even on their second and third orders, to really hook them in for the long run. Those acquisition costs can be sizable and eat into margins. These kinds of promotions are often directed by an established heuristic or business intelligence (BI) rule.
For example, the rule might mandate offering a promotion to every VIP customer. But, in doing so, it extends promotions to those who’d buy again without the promotion — and also misses out on offering promotions to those who are likely to become VIPs. This rule-based approach is expensive and ineffective. It gives discounts to customers who don’t need them, and it fails to build loyalty with other customers who are likely to engage for the long term.
Continuing with the subscription box example: There’s a good chance that less than 20% of your subscribers are profitable, and not until they’ve ordered at least six subscription boxes. Wouldn’t you want to know who those 20% are in the first week or two and quickly identify your “future best customers”? How about those who could turn into future VIPs with a bit of a nudge? Finding these premium customers early on will help identify similar audiences earlier in the engagement funnel.
This type of predictive intelligence and insights can be generated from the customer event and transaction data that companies already gather as part of their day-to-day operations. AI-based predictive analytics can surface that information.
5 steps to using predictive insights
When companies want to use predictive insights to drive more significant business results, they should focus on the following steps.
Evaluate whether business intelligence rules are actually driven by the data
Is your company using predefined rules or, worse, outdated rules to make decisions? Are you tracking actual results linked to those rules, and then adjusting them as needed to show real results? Ask yourself how your company defines a good customer, and how often that customer actively interacts with your brand.
Churn might also take on different definitions in specific businesses. Churn may mean a customer vanished entirely, or it may mean their interactions have become much less frequent. The most common definitions may not really be indicative of your business performance, yet we base so much of planning, forecasting and budgeting on those definitions.
Make sure your definitions of active user, good customer, and churn are regularly refined. These definitions need to work for your business — even as your business, the market conditions and the competitive environment evolves.
Eliminate data silos
With the proliferation of SaaS tools, we seem to be collecting so much more data, yet most companies still struggle to integrate it properly to extract insights that would be indicative of future performance. There are a variety of reasons for that: internal data privacy, legacy mindset around who owns what data, lags in data warehousing strategy or operational know-how about the mechanics of integrating it.
Even within well-defined disciplines like marketing, siloing is still a challenge that hinders performance. The CMO Survey found that after a decade of integrating customer data across channels, marketers are still struggling, with most giving their organization a 3.5 out of 7 score on the effectiveness of their customer information integration across purchasing, communication and social media channels. Ironically, this score has actually gone down since 2014, with marketers saying their programs are getting worse over time. Creating a complete, integrated view of the customer by abolishing data silos will drive the best decisions.
Watch out for the separation of the BI and AI disciplines
When the BI team is reporting to the chief revenue officer, and an AI team is reporting to the CIO, it’s easy to create information silos that make it difficult to see the broader picture. It becomes even more challenging to find useful insights. Some companies are solving this by merging the two groups under the office of a chief data officer, but progress is slow here, thus hindering results.
Don’t be over-enamored with actionable insights
Most analytics endeavors will yield some useful information that can be acted upon. But does every insight that’s actionable offer equal value? Absolutely not. You need to focus on building data strategies and spending resources on surfacing the precise insights you need to achieve your most important business objectives. This focused approach is far more efficient than sifting through a haystack of actionable insights in the hopes of stumbling on the one that will give you just the right boost to your revenue or a major efficiency gain at this very moment.
Go beyond observing dashboards and reading reports
Too often organizations are overly focused on dashboards and analyzing past trends to determine future actions. Dashboards and reports are often thought of as the final deliverables of data, but this thinking is limiting data’s value. Think about how your acquisition, monetization and retention journeys are orchestrated today, then feed predictive scoring data right into those business systems and tools. This integration directly impacts your top line and bottom line, instead of just looking at the past.
Predictive Insights: Getting the most from your data
Calling data the world’s most valuable resource makes sense, especially given the importance and credibility that more and more organizations place on capturing and analyzing data. But if you don’t use your data correctly, you’re not going to get the best results from your marketing campaigns.
Companies need to look at how they’re using their data and identify the most valuable insights they can glean from it — and then they can see what data is truly useful for their goals. After all, if 87% of data science projects never make it into production, is data being used in the most valuable way possible?
Zohar Bronfman is cofounder and CEO at Pecan AI.
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!