How to Seize the Advantages of Image Recognition, Visual Intelligence

Artificial intelligence is top of mind for many brands and retailers, who are rushing to prepare for agentic AI. They are looking to build on the success of generative AI, which is being used across operations from packaging development to customer service and advertising. Generative AI has enabled advances in image recognition and visual intelligence, and agentic use cases have the potential to be just as impactful, if not more, improving supply chains and reshaping how products are marketed in stores.In 2025, 25% of companies that use gen AI will launch agentic AI pilots or proofs of concept, growing to 50% in 2027, according to predictions from Deloitte. To unlock this potential, however, companies must figure out how to make the most of these emerging technologies even as the entire consumer packaged goods industry faces major challenges.“They’re fighting a supply shock due to geopolitical impacts and demand shock, which is inflation, declining confidence and a squeeze on spending,” Lokesh Ohri, EY Americas consulting consumer products sector leader, told CGT. “Brands need a new strategy to regain confidence with their consumers and to drive organic growth. Visual intelligence is a tool that could help them build a competitive advantage.”Retailers are using visual intelligence to reduce out-of-stocks but also to gather information on shoppers that they can sell to brands. Meanwhile, third-party compliance companies are using it to revolutionize their services, moving away from sending agents to stores to take pictures. “Brands that harness visual recognition are keeping up with the times,” Ohri said. “They are starting to set the trend quickly. This will not be a massive competitive advantage alone. It will be table stakes. Today, brands are leaping forward, turning every shelf, every screen, every social feed, into a source of actionable insight. This is going to be a level playing field pretty soon, given how fast this is evolving.”Capitalizing on this new technology will require strong relationships between retailers and brands, integrated teams that can navigate local regulations on image recognition, and a clear outline of KPIs to ensure companies make the most of the data they receive.1. Rethink compliance strategiesVisual intelligence gives brands a multidimensional view of how their products are displayed on shelf, something that is far more accurate and cost-effective than paying people to conduct in-store checks. “Stores are becoming massive data trawls,” Till Dudler, Accenture strategy lead for consumer industries, told CGT. “I think there’s a real opportunity for CPG and retailer collaboration here because you get actual real-time data that you can do something about.”Dudler said many companies have been cutting back on field sales, but argues that they should reconsider to ensure they maintain strong, collaborative relationships with retailers.“Stores will optimize their stock outs, but for you as a brand, it will be even more important because sometimes there are substitute products,” he said. “There might be a case for CPGs investing in more field agents because they can get alerts and go to the source and help sort it out quickly.”Shelf placement is critical in how brands compete against each other and private labels, and visual intelligence ensures compliance. But it can also gather data that can be combined with digital models to create simulations of new shelf layouts.“Planogram design is an art and a science,” Ohri said. “Before you run an experiment at-shelf that can take 12 to 16 weeks to execute, you can land some control sets and have the eye tracking and step back and simulate what could have worked, so now you know where to place your product.”“You go from AB testing to A to Z testing,” he added. “You can find the most optimal [planogram] that’s not just appealing to the eye, but profitable for the business as well.”PepsiCo has been using image recognition technology within its planogram process, getting sales reps the exact percentage of LBR category and SKUs of every item in a cooler within two minutes and a few pictures. “From a compliance standpoint, the biggest win is that we now have an electronic way to audit a store and take a planogram picture,” a director of sales capability for the company previously told FORM. 2. Retailers will have most of the powerRetailers are using digital shelf tools to prevent out-of-stocks, but also gathering data on how customers move through the stores and what displays their eyes are drawn to. They’re using that information as a new revenue source by selling insights to brands.“Consumer goods companies should get in, but should figure out and test the KPIs and what they can do with this and the level of access they will get,” Dudler said. “If you’re a category captain, you can probably get more access.”Brands will distinguish themselves by how effectively they can use the data they gain and their ability to lobby retailers for changes. Crunching all that data can be expensive, so companies need to optimize how they’re using visual intelligence to avoid cutting into their profits with runaway technology costs.“If you’re a challenger brand, and you know your placement isn’t right, what do you do?” Dudler asked. “I guess you can call headquarters and get it changed, but this will be better for companies that can afford field sales teams. Challenger brands typically will be more data-native, so they probably will be better on the analytics side of things, but then actioning might be harder for some of them.”The new technology is likely to be integrated into retail media contracts that retailers have already been aggressively pushing, but having access to real-time data can give brands an edge in negotiations.“You can actually change the contracts saying, ‘I pay you on your performance,’ and that helps increase the conditionality of some of the investments consumer goods companies make,” Ohri said.3. Test the technology where you can, and expect legal changesLatin America has some of the most relaxed rules concerning the use of visual recognition, and Dudler said some retailers there are setting up cameras to start gathering data on customers by scanning license plates and connecting them to past purchases before they even get out of their cars.“You segment by the car and they have TVs with personalized advertisements,” he said. “In most countries, you’re not allowed to do this.”Visual intelligence can allow retailers to put in a request for more product as soon as shelves start getting empty, but as they gather more information on shoppers, it can even start triggering restocks based on who has walked into a store, similar to how Amazon uses its predictive algorithms to locate items near likely customers.“The threshold of what people feel comfortable with moves all the time,” Dudler said. “It’s scary when you think about what teenagers now feel comfortable sharing versus what we are. The [General Data Protection Regulation] moved the threshold back on what you’re allowed to do, so within the EU there will be continued strong consumer protection and regulation, but in other markets, probably less so.”Ohri expects that smart shelves in stores will eventually be able to recognize shoppers standing in front of them and use agentic AI to tailor their experience based on visual cues and data from their phones.“It could be a promotion based on what’s in your basket,” he said. “It could be the right product size fit for you. You want the consumer to take one more thing home. But for that to happen, you need the agent to work, and you need to watch out for privacy [restrictions].”4. Reduce the manual strain in packaging processesVisual intelligence has the potential to reduce the friction of requiring people to scan products throughout the supply chain by using digital watermarks that can be instantly recognized. They can also be helpful as consumers shift from using text-based search to find products online to using AI tools that allow them to simply upload a photo of what they want.“Those watermarks on your products quickly help the algorithms — which are doing visual search in a very crowded world with millions and millions of products — identify what that product is and help your consumer get it,” Ohri said. “That needs to happen for the consumer experience to become seamless.”Watermarks have additional advantages, sharing information with less friction than smart labels or QR codes.“What if the product is your bar code?” Ohri asked. “You have to present these packages to the [AI] models to be able to recognize them to build the patterns. That’s starting to happen more and more.”5. Bring new efficiencies to manufacturingVisual intelligence is only expected to grow. The global market size was estimated at $11.4 million in 2024 and is projected to expand at a CAGR of 19.1% from 2025 to 2030, according to Grand View Research.It can improve the efficiency of supply chains not just by ensuring products get to stores before they run out, but by reducing labor costs and errors in production. Ohri calls fully automated lights-out manufacturing “the holy grail of manufacturing,” but explains that visual intelligence is also important for lower levels of automation.“Robots have to recognize where other robots are, where the shelves are,” he said. “Incorporating image and visual recognition at the warehouse is like turning the lights on in a dark room for machines and robots to assist humans in completing their tasks.”The downside is that incorporating all that data into factory processes can make a cyberattack even more dangerous.“You now have a complete digital map of every product that you have,” Ohri said. “If not protected correctly, you could be losing your competitive edge of how you run your facility, what your products are and how much stock you have.”

How to Seize the Advantages of Image Recognition, Visual Intelligence

Related Topics

Business Topic