Why are customer analytics important? Neil Hoyne, Head of Customer Analytics at Google, referenced a study done several years ago to see how often marketing decisions are based on customer data. Of the 350 marketers interviewed, only 6% of decisions were made on the basis of customer analytics, with the remainder making decisions based on the opinions of their managers, colleagues, and a whopping 19% on simply “what felt right.”
While everyone knows that customer analytics are important, few companies are actually using the data and even fewer can connect an understanding of their customers to ROI. If your company estimates that 9% of your customers are driving 80% of revenue, do you know who that 9% of people are? Do you know how to change your investments to not only acquire more customers, but keep the customers that you already have?
Many companies claim that they are already looking at customer data to inform their advertising decisions. But Neil provided an interesting example of how looking at only part of the funnel can actually be a cost to your company instead of a benefit. A woman buying a pair of shoes clicked on 72 paid ads before converting to a paid customer. The business knew the last ad she clicked before converting, so all the credit went to the last click. They ended up losing about $35 on the customer.
Customer analytics help you understand how your best and worst customers perform and tie it into benefits and loss. Neil explained that understanding relationships is at the core of customer analytics—figuring out how to measure and influence those relationships. He emphasized the importance of bridging all of a customer’s interactions together to understand them as individuals. No one understands a business better than the customer themselves. Make sure you are not taking a loss trying to retain customers. Once you begin to target and personalize customers, to know who they are, you will know how to put new products in front of them.
Customers with identical last-click behaviors are not equal. You might assume that two people making the same purchase from a luxury manufacturer through the same click are equally valuable to your business. However, after you start looking at their behavior and see that one customer has come in only once or twice and the other customer comes in every two months, you will start to understand the difference and determine where and how to focus your advertising investments.
It’s often difficult to move beyond a last-click attribution model. But an attribution model that just tells marketers what each interaction is worth is usually too simple of an approach and businesses miss key opportunities to connect with consumers. Instead of dwelling on clicks in isolation, Neil recommended taking a customer-centric approach to attribution, examining all of the data points in the customer’s path.
What if you were able to figure out who the high-value customers were and spend your money on how to convert and keep them? You might be able to ignore 80% of the audience, even putting them on a no-invest list. You can then reinvest those resources in other customers to whom your ads will always appear. Understanding individuals involves three parts: first, understanding how you acquire a customer, but then going deeper into how you develop that relationship to get the customer to spend more, and finally how to keep them as a customer. It means cross-referencing paid search campaigns, searches, and other data points to forecast the lifetime value of each customer.
So how do you get started? Success and failure is less about systems and data sets and more about individuals. If you’re not already collecting CRM data, start collecting it and use it as the source of truth. You can then supplement CRM with individual analytics platforms to determine how customers get in and how they leave. Google uses BTYD open source, which forecasts out the life or death of a customer and estimates how much they are going to spend. Currently, analytics platforms and CRM systems are being built and run separately; however, we will see more integration between the two over the next couple of years. Neil recommended not focusing on the quality of the data. He said that you can’t incentivize salespeople to enter the right data—just collect what you can.
So how do you influence the process? To handle the underlying KPIs, Neil recommended running very simple, very fast tests to make sure you’re going in the right direction. Aim toward something you can do in a short period of time, such as 6-8 weeks. Start with changing your metrics and slowly expand. Run the basic models and see how your customers are different. Understand the behaviors of good and bad customers figure out why you’re addressing everyone the same. Then, target and personalize your message.
If not last-click, what kind of attribution model should you use? Neil said there is no definitive judgement on models. Some advertisers will still use with last-click. Others will try time decay and linear with success. When companies are designing attribution models, they’ll often use all the historical data. But all of the data was built on last-click method. How are you going to explore and enrich customers earlier on the path? Many tests only cover what a user does when they come to the destination website. What if you could run a test over time—to find out where they went before and determine how to adjust messaging to keep customers coming back.
Finally, incentivise your employees to design and run the right tests. Take a step back to advance the business forward by making cultural changes to reward people for running the right tests.