With digital marketing now more easily tracked than offline marketing (even on mobile devices), and the cost of servers and technology needed to track and store all the data a fraction of what it cost a decade ago, there is no reason today to make a marketing decision without consulting the data.
That's why VCs have invested billions in advertising technology companies that are finding new ways to improve digital marketing based on different uses of data.
But if data in the offline marketing world was expensive and time-consuming to collect, today we suffer from too much data. It’s hard to separate the signal from the noise.
That’s why just having data is not enough.
Here’s what marketers need in order to derive the fullest marketing value from their data.
1. A continuous stream of historical data: Possessing data is critical, but before marketing predictions can be made, you will need a historical stream of data in order to understand and uncover data patterns (for example, let’s say that Lowe’s analyzed snowfall in Boston based on this year’s record snowfall to stock snowblowers for next fall instead of consulting historical snow fall data from the last decade. This might result in overstocking of snowblowers, which will then be discounted at the end of the season, causing lower profitability and revenue.)
At least two years of historical data is optimal, but shorter periods of data can also be used to test promotions such as user sign-ups. What’s important with historical data is the similarity of the parameters around the data. For example, comparing user sign-ups from two email campaigns where many of the variables were similar is okay. But comparing user sign-ups from one email campaign versus a user sign-up campaign via Facebook or Twitter will be more challenging because the platforms are different.
2. The algorithms to analyze the data: Once you have a continuous stream of historical data, the fun begins as you look for data patterns that can help you predict future actions. The challenge is to crunch the billions of data points in order to reveal the data that will yield actionable insights. To analyze the data you will need predictive algorithms that will be able to analyze your data across the relevant parameters/segments in order to uncover patterns. Patterns could include the fact that users who bought product X are 17 times more likely to also buy product Y (but are highly unlikely to buy product Z) or even what times of days/days of the week are best for targeting which segments of users.
3. The data scientists to uncover the actionable insights from the data: Even with algorithms, it still takes an experienced data scientist in order to analyze the data and uncover the insights to drive future marketing campaigns. Given the vast quantity of data and the range of possibilities, a trained professional is needed to guide the process and to determine the data sets to analyze. And once the initial analyses are made, the data scientist will then recommend new/different analyses in order to uncover the strongest patterns that will generate the best performance in future marketing campaigns.
4. Test, test & re-test: Once you have your data patterns and know how to improve the targeting and conversions of your next campaign, you need to remember that data changes over time. This means that you must continuously analyze and optimize your data sets. A data pattern or data source which is very effective now might be much less effective in four months.
There is no doubt that the ease of gathering data has changed predictive marketing forever. But data is only the raw ingredient. It takes a skilled team using the latest technology in order to harness the power of your data.