A mere eleven months ago, OpenAI launched the public version of its AI-based, large language model, ChatGPT. In that short time, the noise surrounding AI has dominated the media, boardroom conversations, and almost every social gathering on the planet. Retail advertising and marketing directors everywhere are inundated with questions. “What are we doing about artificial intelligence – and how will we survive without it?”

Thankfully, Comosoft and its AI development partner, DecaSIM, have a practical answer. They now offer a clear path for retailers searching for real-world AI solutions, relieving the pain of turning vast amounts of complex data into meaningful, more cost-effective marketing campaigns. In a recent case study, a leading regional grocer implemented a live test of Comosoft’s marketing automation system, LAGO, in combination with DecaSIM’s AI-based data analysis tools, generating positive results from week one. By optimising which promotions to feature in its advertising and creating more customer engagement, the company demonstrated increased weekly sales and an eight percent increase in EBITDA profitability.

Artificial intelligence for the rest of us

Much public anxiety over AI stems from common misperceptions about the technology. ChatGPT and other “generative” AI systems often look for patterns in vast amounts of unstructured data – meaning there are no preexisting identification tags or metadata to help us sort or classify them. The Comosoft/DecaSIM AI model is based on a smaller set of structured data, such as the known purchase history of specific products, prices, and seasonal buying patterns. The AI tool uses the data to solve equations to create new (and precious) metadata describing shoppers’ behavior and preferences about specific products.

In other words, the AI is finding meaningful inferences from the data based partly on the domain knowledge of advertising and marketing professionals. It does so with volumes of data that, while not as vast as the trillions of data parameters required by generative AI systems like ChatGPT are still too large for humans to analyse without help.

All this puts the Comosoft/DecaSIM approach in the category sometimes known as “Narrow AI,” or artificial intelligence designed to perform specific tasks. This differs from purportedly self-aware, artificial general intelligence or AGI, which is still hypothetical at best. However, since simplified tasks are critical to retail success, advertising and marketing directors have little to fear and much to gain. The belief that AI is taking over from humans is offset by a growing awareness that AI can be a critical part of any retail marketing team.

During the testing process with retailers, DecaSIM co-founder Chris Antipa noted that advertising and marketing departments are finding their work enhanced, not threatened.

“There’s always this natural pushback, with people saying, ‘Hey, computers are coming for my job.’ But as time passes, they begin to understand what the AI is doing for them and talk about the platform as a teammate. They even joke that they should give it a nickname.”

Putting AI to work

The combination of DecaSIM’s AI tools and Comosoft LAGO can only be described as a high-impact data solution. LAGO already automates various vital functions for retailers, including integration of campaign planning, numerous data sources, and streamlined production of print and digital campaigns – including multiple regional versions of each catalog, flyer, or digital equivalent. Using AI elevates that process even further. Historically, structured data from LAGO and other sources are fed to the DecaSIM system, deriving meaningful shopper habits and preferences for specific products and product categories. This enhanced metadata is fed back into LAGO, providing an objective basis for promoting particular products at specific locations and times of the year.

The results have been remarkable. Co-founder Doug Edmonds recounted the response of one retail executive, who said the DecaSIM/LAGO combination “made every ad like a holiday ad,” thanks to the system’s ability to anticipate shoppers’ preferences and product attributes. “The data we utilize contain context and meaning hidden to most of the retail world,” Edmonds continued. “AI uses those data to define that context and meaning, rate its importance, and add all that to the metadata for each product and category.”

The LAGO approach, which integrates marketing-relevant data from product information management (PIM), digital asset management (DAM), and other sources, is ideally suited to optimise the new DecaSIM data. “Traditionally, more than half of our implementations start with some rich data feed,” said Comosoft Product Manager Steve May. “Typically, this is a collection of product SKUs, digital assets, and offers. With DecaSIM, the data are enhanced before we even get it.”

This is especially important to advertising and marketing directors at the planning stage of a campaign. LAGO’s whiteboarding capability already gives them an overview of which products to feature prominently based on criteria such as regional availability, profit margin, and sales history. With the addition of DecaSIM data, they also can have reliable indicators of what shoppers will more likely purchase, including their preferences for related or alternative products.

“LAGO and DecaSIM are both ‘painkillers’ for retail advertising and marketing professionals,” said Comosoft US President Randy Evans. “One stops the pain of integrating PIM, DAM, and other data to execute campaigns at scale. The other removes the guesswork of which products are more likely to sell.”

The future of AI in retail marketing

The success of this approach is not limited to the grocery market. Retailers depend heavily on regular foot traffic and a known population of frequent shoppers – a dimension that fits everyone from home improvement and hardware retail chains to fashion and lifestyle brands. Knowing objectively what shoppers want and advertising accordingly is a holy grail for all retail marketers – one that DecaSIM’s AI can find and LAGO can implement, even down to the local level and on multiple print and digital platforms.

The two companies are actively engaged in combining and expanding these capabilities shortly. Some of these enhancements will be the subject of next month’s article.