AI/Machine Learning Solutions Guide 2019

CGT presents a comparison chart of solution providers on the forefront of artificial intelligence and machine learning for the consumer goods industry. Plus, industry experts provide thought leadership on challenges, opportunities, and implementation issues for companies navigating this new playing field. Are there any obstacles remaining to the widespread implementation of artificial intelligence at consumer goods organizations? If so, how do companies get past them? MORAN: There is no barrier to widespread implementation per se. Rather, the opportunity is to create the user applications to solve specific business challenges. Otherwise, pushing AI is a bit like a hammer running around looking for nails. I’d rather start with what you want to solve. Machine learning — the branch of AI that has generated the most buzz — is really good at looking at lots of data to pick out patterns. But without a context and an application, it’s a bit like saying “big data.” All data has been big for a while, that’s why [the field of] statistics was invented. I’m far more interested in starting with an opportunity or a problem, and then working backwards to see how a range of modern analytics solutions might be applied to address it. WAGNER: While there is certainly enough quality data available to build models that provide meaningful learning and results, consumer packaged goods manufacturers are struggling to find relevant use cases to model after. When we establish a small proof of concept that delivers evidence of value and demonstrates the overall worth of artificial intelligence across the organization, it goes a long way toward greater implementation of AI within the industry. In addition, support from the c-suite early in the initiation process will gain their buy-in on implementation throughout the entire organization. Beginning simply with a pilot program to build upon is the first step in the direction of more pervasive deployment. When we do these pilots, we focus on high-level numbers and inquiries that impact the entire company as well as the sales and performance of a large region or category. In which business function have companies enjoyed the most measurable success so far? In which areas has implementation been lagging? WAGNER: We’ve seen the most calculable success within the finance areas of CPG and retail companies, undoubtedly due to the fact that our solution utilizes AI to focus specifically on predicting future headwinds and tailwinds to demand. In our experience, even small use cases yield significant value to the bottom line in these areas. On the other hand, we have observed that the biggest challenge comes when clearly defined business objectives for using AI are not established. Identifying objectives with executive visibility is paramount to quantifiable success and, ultimately, to more widespread implementation. Executives who are willing to maintain an unbiased attitude and embrace the latest progressive solutions will enjoy the most success as AI becomes more prevalent within the industry. MORAN: Certainly, programmatic advertising has been transformed by AI. Machines do a great job of looking at lots and lots of inventory for ads and determining the best balance of cost, quality, and reach to maximize marketing objectives. Programmatic selling isn’t there yet, but I think it will get there in the next five years. In a lot of ways, selling should be easier to have machines do than advertising. Prices just go up or down, and perhaps promotions are much more complex, but at the same time they’re still quite measurable: either the sale happened or it didn’t. So, compared to the nuances of brand building and activities at the top of the brand funnel, the selling function could dramatically retool. There’s a lot of friction to this being feasible today, from siloed data and access to manual processes to POS limitations, but I’m hopeful. What are the next steps in technology development?How do we improve the tools that we already have? MORAN: I was once told that software ages like fish. There’s a reason that technology as a whole has broadly turned over to software-as-a-service applications. Improving old, on-premise software is often a waste of time and a magnitude more expensive than just swapping to a modern stack. A lot of retailers especially have been slow to do this, and as a result the big innovations in POS software, for example, have (ironically) happened in the small to midsize business market instead. WAGNER: Unlike 10 to 20 years ago, companies can now leverage cloud-based AI solutions to augment or enhance current systems without the tremendous expense of having to replace them. Tools across the analytics workflow have required a significant amount of manual analysis and human interpretation, an approach that isn’t scalable in today’s environment. Undoubtedly, the next steps in AI technology development will involve embracing and implementing cloud technology as well as augmented techniques to modernize solutions. New development will complement — not replace — existing data and analytics initiatives. As the technology progresses, solution providers will help CPG and retail executives develop a strategy to evolve roles, skills and responsibilities as well as support increased investment in data literacy. Do you have any examples of how AI has dramatically changed a client’s go-to-market strategy? WAGNER: We worked closely with a large beverage producer to implement the use of AI to help the company understand how external data — specifically, economic changes — impact consumer demand for its products. We discovered that factors such as the number of hours that architects build is a strong influencer in driving product demand. As a result of this improved insight, the client was able to develop predictive models that tightened its demand variable and, thus, dramatically reduce safety stock and improved distribution with- out impact to service levels. MORAN: Our retailer clients are using shelf-edge experimentation tools to better compete with e-commerce in dynamic pricing while optimizing the total store experience. Merchants spend less of their time on pricing administration and more on the highest-value judgment tasks, and as a result their categories are performing better. In a recent steering committee meeting, a retailer saw it was driving 1.5% higher sales, higher transaction volumes, and approximately 50% higher EBITDA margins vs. control stores by taking this AI-led “software-as-a-coach” approach for merchants. To download the full report, including a comparison chart of 23 solution providers on the forefront of AI/machine learning, click on the link below.

You May Also Like

Exploring the Future of Retail & Consumer Goods Analytics

Getting to the ‘So What’: Unilever’s Dan Cook on Democratizing Data for Actionable Insights