AI/Machine Learning Solutions Guide 2018

CGT: On which areas of the business are CGs focusing their early AI initiatives? Which areas do you think are being under-considered at this stage?

FINLEY: Consumers, packaging, goods: the industry name says it all. Three vast subjects combined into a single field. The sheer quantity of data that can be collected from one of these areas is compounded by the need to merge data from all three and try to compete with other companies swimming in the same data tsunami.

AI has already been applied to manufacturing and customer service, but many CGs are ignoring its value in augmenting human decision-making ability. Two possible reasons why: They fear AI will replace jobs (whereas we find people are freed up for more advanced, less repetitive work) and the benefits aren’t as quantifiable as say, preventing machine downtime.

Humans by themselves can’t keep up with this unprecedented level of data. Today’s brand managers barely have the tools to project a few market scenarios; tomorrow’s augmented business leaders will implement strategies that simultaneously mitigate threats and exploit opportunities.

WAGNER: One of the early uses of AI was to analyze emails, chat logs and social posts to determine the sentiment of the consumer. This all falls in the category of “unstructured data” analysis, where the information can be text, numbers or images, spread across various platforms and formats.

While this is important, businesses live or die based on how well they prepare for the future — either bracing for headwinds or capitalizing on tailwinds. AI can powerfully analyze “time-series data” (such as weekly sales volume or monthly revenue figures) and identify leading indicators not only with internal metrics, but with external factors such as weather, income or consumer sentiment.

AI can provide a holistic understanding of what drives business performance and can very accurately predict future performance based on signals happening now. 

KEANE: Most companies are making significant investments in supply chain efficiency and focusing a lot on assortment optimization. We anticipate that the industry will begin to pay more attention to process optimization for things like pricing, stocking frequency, dynamic routing, alternate offers, intelligent visit planning, fixed asset deployment, and workforce enablement and training. Sales rep performance optimization is very important, too.

You should focus on implementing algorithms that can deliver real value. AI tools need to be integrated with actual processes and workflows to ensure adoption. You have to learn from the field experts and incorporate their expertise into the machine learning algorithm — they know the consumer better than anyone.

Right now, we’re looking at how organizations like Amazon, Uber and some of the big trading banks are using recommendation engines to influence behavior. We’re also focused on shifting the primary interface from screen to voice.

CGT: How difficult will it be for companies to implement AI throughout the enterprise? Is this a 20-year, bank-breaking effort or will the consumer goods industry be “powered by AI” before we know it?

FINLEY: AI already exists in places that, as consumers, we don’t really think about. We use natural language to interact with our phones, and our software recognizes our faces. AI, and any new technology, should be implemented in manageable, measurable sequential steps.

This means starting sooner rather than later. A business that waits until it has all the data will be outflanked by more agile teams. They will likely miss many of the benefits and opportunities that they would have found by letting smart people and machines work together.

This is not to say that AI won’t cause any disruptions or that it won’t have any cost. But progress always requires some level of change management, and AI is no different. It would be a mistake to overspend or to not spend enough. As in all aspects of life, the difference between medicine and poison is the dose.

WAGNER: Cloud computing and AI go hand in hand. As a result, gone are the days of spending millions on new hardware and software only to spend months — if not years — implementing a new solution.

CG companies large and small can embrace cloud-based AI solutions in a phased approach, leveraging existing IT investments, databases and reporting platforms. Insights delivered with AI through the cloud can start off with web-based applications. Then, as best practices are developed, these insights can be integrated within existing processes.

The additional benefit of cloud-based AI is that companies don’t need to invest in new hardware every time there are advances in computing power. Cloud-computing providers allow companies to scale up or down as needed, without the need for large capital expenditures.

KEANE: I see it both ways. The shift to AI will take some time, but it’s going to happen much faster than most people realize. It has to. The benefits of integrating AI with distribution, merchandising and selling are too great to ignore. Inexpensive tools are everywhere, so trial and error is key. You have to accept that mistakes will be made, because benefits outweigh the costs.

Achieving success will not “break the bank” as long as AI engines are focused on driving real-world efficiencies and growing revenue. I believe AI will become pervasive in our industry over the next five years. The more it’s infused into software, the more that costs and risks will decline and adoption will accelerate.

We’ll soon start to see an algorithm for optimizing every process imaginable. This will create the need for algorithms that reconcile competing algorithms and governance for legal, ethical and cultural concerns.

Click below to download a comprehensive listing of solution providers for artificial intelligence and machine learning.