Most of the chief growth officers I know have been hired to unlock future organic growth potential by evolving the culture and structure of their respective organizations. Organic growth (as opposed to growth by acquisition) is especially hard for mature brands. Once your product is widely distributed and you’ve fought for all the retail facings you can fight for, velocity becomes the key growth driver.
Let’s assume, for this writing, that the CMO is awesome and that the marketing department is totally on top of line extensions, product attributes and programs to optimize velocity. Let’s also assume that there are fully optimized advertising campaigns driving awareness of the marketing programs and that everything that traditional marketing and advertising can do is being done. This is a huge assumption, but let’s accept that we have the best people in the world working for us and they are doing their jobs. Now, how do you grow?
Hurricanes and Pop-Tarts
Back in 2004, Linda M. Dillman, Walmart’s CIO, instructed her staff to mine Walmart’s customer data to forecast sales in advance of Hurricane Frances. According to the New York Times,
The experts mined the data and found that the stores would indeed need certain products – and not just the usual flashlights. “We didn’t know in the past that strawberry Pop-Tarts increase in sales, like seven times their normal sales rate, ahead of a hurricane,” Ms. Dillman said in a recent interview. “And the pre-hurricane top-selling item was beer.” Thanks to those insights, trucks filled with toaster pastries and six-packs were soon speeding down Interstate 95 toward Wal-Marts in the path of Frances. Most of the products that were stocked for the storm sold quickly, the company said.
Data to the Rescue
With a fraction of the computer power we have at our disposal today, Walmart was able to use a simple algorithm to predict demand, adjust inventory levels and increase sales.
Association Rule Mining
The algorithmic tool the analysts used is known as “association rule mining.” It focuses on finding co-occurring associations in data sets. This technique is often used for “share of basket” or “affinity” analysis. While the specifics are best left to someone with excellent math skills, let’s think about how Walmart might have approached the problem.
Beer and Pop-Tarts
By analyzing past Walmart transaction data ahead of Hurricane Charley (a hurricane that hit a few weeks in advance of Hurricane Frances), Ms. Dillman said her team determined the top-selling pre-hurricane item was beer. What surprised her was the sevenfold increase in strawberry Pop-Tart sales. The “association” of Pop-Tarts and beer required the application of association rule mining. Let’s illustrate this with some fake data.
Let’s say Ms. Dillman’s team analyzed 1,000,000 transactions in advance of Hurricane Charley.
100,000 transactions included beer (10 percent)
15,000 transactions included Pop-Tarts (1.5 percent)
10,500 transactions included both Pop-Tarts and beer (1.05 percent)
Beer and Pop-Tarts are statistically independent, so because 10 percent of all pre-hurricane customers bought beer, we would expect only 10 percent of Pop-Tart buyers to also buy beer (15,000*0.1=1,500). But that’s not what happened. In the pre-hurricane timeframe, 70 percent (10,500/15,000=0.7) of Pop-Tart buyers also bought beer – an increase (or lift) by a factor of seven over expected sales (7*1,500=10,500).
The ability to predict the need for a sevenfold increase in pre-hurricane Pop-Tart inventory levels or, better yet, having time to set up an end-cap or checkout area pallet or outpost is a guaranteed velocity driver. Where might I find such a data set?
I asked my friends at AccuWeather how far in advance they could predict hurricanes. It was the wrong question. Chief Commercial Officer Casey McGeever told me, “People don’t go to the store when there is a hurricane . . . they go to the store when a hurricane is predicted.” For that, you need insights into one of the biggest big data sets: weather. As it turns out, AccuWeather makes this information available. So in practice, you could apply predictions from these data against association rule mining of your sales data to predict required inventory levels, suggest sales efforts or even inspire seasonal programs.
Data Science Tools
It is hard to find frequent item-sets in a large database. It requires good mathematics and a computer science environment that is purpose-built for the task. But the world is changing fast. Recently, Amazon announced its new Amazon Machine Learning (AML) service. Amazon says, “It’s a service that makes it easy for developers of all skill levels to use machine learning technology.” The service includes visualization tools and wizards that guide you through the process of creating machine learning models without having to learn complex algorithms and technology.
This new Amazon service is directly competitive with Microsoft’s big data offerings, and there is clearly a trend toward Data Science as a Service tools – which is a huge win for chief growth officers everywhere – unless, of course, they are replaced by an AML dashboard – which might just become your new CGO.
We have a team ready to help you get ready to work with your data, understand the opportunities afforded by machine learning and pattern matching and even do a data science readiness assessment. Just shoot me an email, and I’ll be happy to work with you to help you achieve your business goals.