7 Ways Coca-Cola Uses AI and Big Data
While we happen to be in a unique position to dispense investing advice, we never tell people what stocks they should invest in. Instead, we talk a bit about what investments we make and why we make them. Put your money where your mouth is and all that. While we predominantly cover disruptive technologies, pure-play stocks on such themes are far and few between. That’s one of the reasons why the lion’s share of our investment dollars have been sunk into Dividend Growth Investing stocks or DGI stocks.
One good example of an ideal DGI stock is Coca-Cola (NYSE:KO). This is a company that has not only paid a dividend but raised it for 55 years straight. Go crunch those numbers using the most basic growth rate and you’ll see why Warren Buffet’s yield on cost (look it up) is over 50%. Essentially that means if he invested $1000, he now gets more than $500 a year in dividends. Incredible, innit?
What’s not so incredible though is the extent to which the world’s 27th greatest brand (down from 16th in 2016) seems to not be paying much attention to what gets placed on their web pages these days. Just go to http://www.coca-colacompany.com/ and scroll to the bottom of the page:
25% of the link real estate being used on this footer is dedicated to something called “The Opener”, which reeks of some brainstorming session that sounded good on paper but completely flopped when it came to execution. What sort of insightful, engaging information can we expect from blogs like “Girl Loves Life” or “Social Nerdia“? Who knows, because one blog no longer exists and the other is one of the saddest excuses for a web presence we’ve seen in a while. How about “My Stilletto Life” where Tillie blogs about her move from NYC to Los Angeles with her new boyfriend and shares with the world a new biscuit recipe along the way? Maybe it’s because she’s an “Instagram celebrity”, but she certainly isn’t a coder:
What exactly is Coca-Cola, a company with one of the biggest marketing budgets in the world, trying to accomplish here by linking to this drivel on one of their main pages?
Perhaps what’s most depressing here is that you just know they had a team of about 10 “social media marketing experts”, probably some on six-figures, working on “The Opener initiative”, not one of whom could cobble together a solid list of “thought leaders” to represent the Coca-Cola brand, or even to make sure the links don’t lead to defunct blogs. As investors in Coca-Cola, this is troubling, and made us concerned enough about our investment to take a closer look at how they’re using technology. The only way that Coca-Cola will continue increasing profits so that they can increase those dividend payments for another 55 years is by using technology like artificial intelligence, not by engaging with the Tillies of the world. We were pleasantly surprised to find some interesting applications of AI and big data.
Big Data Flavor Shots
We all know that artificial intelligence is only as good as the delicious big data you feed it. That’s why Coca-Cola started monitoring their self-service soda machines to see what “flavor shots” people preferred. As it turns out, this was the inspiration for their launch of “Cherry Sprite”.
We might be inclined to think that you wouldn’t need AI to make those types of decisions, since all you need to do is just aggregate the data using traditional querying methods, and that’s probably true. However, they also use some natural language processing tools to scour the Internet and see what people are saying about their flavor creations. We learned about this from an article by Forbes on the topic of Coca-Cola and AI which talks about how “its products were mentioned somewhere in the world an average of just over once every two seconds“. This brings us to our next topic.
Scanning Social Media
These days you need to be aware of what people are saying about you on social media because of the outrage machine. While we’re still amazed why companies even bother trying to apologize, we understand how valuable it is to know what people are saying about your products and brand. With more than 23 million retail outlets, Coca-Cola needs to make sure they’re emitting consistent brand messaging without offending anyone – because outrage machine. According to an article by Nastel Technologies, in 2015 Coca-Cola needed to analyze over 20 billion impressions coming from 120,000 pieces of content. This sort of data monitoring is also how they figured out it was a good idea to bring back Surge:
Because 90% of consumers now base their purchasing decisions on social media content, Coca-Cola uses computer vision to detect photos of their products and then more algorithms to gauge the sentiments of what people say about said photos, or their brand in general. Speaking of which, let’s look at some other ways they’re using computer vision.
Reading Product Codes
Patrick Brandt from Coca-Cola spent some time on the Google Developers Blog talking about how “they’re using AI and TensorFlow to achieve frictionless proof-of-purchase“. The effort was around a consumer loyalty program, one that required customers to enter a 14-digit code into their mobile phones. They first tried to solve the problem using old-school OCR (Optical Character Recognition), but ran into problems trying to quickly and accurately scan codes like these:
They then turned to deep learning with Google’s TensorFlow platform that we talked about in a previous article. The end result allowed the user to take a picture of the bottle cap that was then translated into a code at a 99.97% accuracy. It takes about one second to process the image, and the success of this effort led to the technology being a “core component for all of Coca-Cola North America’s web-based promotions“.
Machine Learning for Selfies
Reading product codes isn’t the only clever use that Coca-Cola came up with for computer vision. Way back in 2015, Microsoft had come up with a cool tool that guessed people’s age. If you go upload a picture of yourself holding a Coke bottle, it will unlock “the special Coca-Cola experience” which guesses your age and tells you the age of the Coke bottle you’re holding.
Gimmick? Sure, but it’s a clever way to get people to interact with your brand and that it also might come in handy for making those AI algorithms that are constantly scanning social media a bit more intelligent.
AI-Powered Vending Machines
Even in the land of vending machines, Japan, the actual function of the vending machine hasn’t changed that much. As it turns out, there are a lot of cool features you can add to vending machines that make them much more interesting. An article published by Coca-Cola in August of last year talks about “AI-powered vending machines” that are both easier to manage and operate. These new machines can “offer specials, track sales, preempt maintenance and refill needs, and even accept mobile payments” so that the user can order in advance of even arriving at the machine.
Chambers imagined sitting on a hot train and having a phone notification appear announcing a Coca-Cola happy hour special – buy a Coca-Cola and get a discounted DASANI, for example.
That’s all becoming a reality now with cloud-connected machines that will examine “the consumer’s Facebook activity, current location and tone of conversation, so the AI bot can adopt a local dialect and attitude tailored to each user“. Sounds like the “digital marketing echo chamber” we talked about before.
CRM and Einstein
Coca-Cola needs to focus on what they do best while leveraging the AI talents of other companies like Salesforce, which built them an app that helps conduct inventories more efficiently. Instead of having thousands of people checking coolers for stock, you can now just take a picture of a cooler and Salesforce’s AI technology called Einstein will then count the bottles, determine “Coca-Cola purity” (this shows if any other brands are located in the cooler), and will also make adjustments based on loads of external data like past purchases or even events that are happening in the area.
Coca-Cola can now deploy this app to all their associates in the field who can now more easily manage orders for the 16 million coolers installed in retail outlets worldwide. Since it’s all cloud-based, you can only imagine how useful this data is for their entire supply chain.
It All Comes Down to Big Data
Going back to what we said earlier, the company with the biggest, cleanest, most accurate datasets will dominate with the most effective AI algorithms. That’s where Coca-Cola can compete like nobody else given the petabytes of data that this nearly $200 billion company produces. According to an article by DataFlow, they’re now analyzing point-of-sales data from companies like Walmart, which accounts for $ 4 billion in Coca-Cola sales annually”. They’re using up to “1 quintillion decision variables to consistently deliver the optimal blend of orange juice” so that they can make sure their orange juice products taste the same every year. They’re also using big data in other areas of their business, “for example to cut overtime costs by 46%” in their employee service center.
As the United States moves closer to losing their technological dominance to countries like China, we’re becoming increasingly concerned about our DGI stock portfolio. While we are holding mostly multinational stocks to avoid overexposure to the weakening U.S. work ethic and constant consumer outrage, we still want to make sure that every stock we hold is using artificial intelligence and big data to ensure that profits will increase so that our dividend checks will continue to increase well into retirement. From that perspective, Coca-Cola (NYSE:KO) gives us some hope. If they can manage to focus as much on technology as they do on feel-good messaging, they’ll be increasing that dividend by meaningful amounts for years to come.
Here at Nanalyze, we hold the lion's share of our investing dollars in a portfolio of 30 dividend growth stocks. Find out which ones in the Quantigence Dividend Growth Investing report freely available to Nanalyze Premium subscribers.