AI in CPG and Retail: How Winners Are Pulling Ahead

AI has introduced new sources of disruption along with new opportunities for advantage in the consumer industry. Both consumer packaged goods companies (CPGs) and retailers are feeling the effects of this ongoing transformation as shoppers use new tools to search, compare, plan, and decide on purchases. In response, CPGs and retailers are launching pilots, testing agents, and applying AI across most functions, from innovation and marketing to merchandising, replenishment, and on-shelf availability. Few, however, are capturing substantial value at scale.

Companies that are pulling ahead are doing so not by adding solutions or building more sophisticated models, but by applying AI more deeply to the core commercial levers that build sustainable advantage: faster innovation, higher-fidelity demand sensing, more localized assortment, and better execution across physical and digital channels. Boston Consulting Group (BCG) and The Consumer Goods Forum (CGF) surveyed 39 senior CPG and retail executives, combining those insights with focused interviews and BCG’s expertise to explore how frontrunners are achieving advantage.

State of Play: Real Value, Unevenly Captured

Although CPGs and retailers have long used analytics and machine learning, AI’s potential impact has expanded materially across the demand value chain recently. CPGs are applying it from concept through revenue growth management, customer planning, digital shelf, and brand engagement. In retail, AI is informing pricing and assortment, demand forecasting, inventory management, store operations, and marketing. Agentic commerce is emerging as a new frontier for both segments, reshaping discovery, digital shelf visibility and consumer acquisition.

In our survey sample, we found that maturity varies across sectors. Most CPGs (75%) remain in pilots and exploration, with only 18% scaling significant impact. Among retailers, a bifurcation is visible: 45% are scaling impact, while a similar number (40%) have barely begun. Frontrunners are focusing their efforts in areas where value is clearest, including demand forecasting, pricing, and transport optimization. BCG analysis suggests that scaling the full set of relevant AI initiatives across the demand value chain can deliver 220 to 350 basis points (bps) of cumulative earnings before interest and taxes for CPGs and 180 to 360 bps for retailers. These gains do not represent a consistent translation to margin uplift, as appropriate reinvestment is likely to follow in the form of better pricing, improved offerings, and the like. As agentic capabilities mature and companies build stronger transformation muscles, the opportunity could expand to as much as 1.7x for each segment.

Measurement is a weak link, as more than half of survey respondents say that they do not measure the ROI of consumer AI investments. Part of the challenge is structural, and the pilot trap is real: pilots succeed in controlled environments, but complications compound at scale. Meanwhile, many pilots track data but do so without a clear baseline or threshold for scale.

Some of the next value pools involve the broader ecosystem. The near-term opportunity can be to advance voluntary precompetitive foundational elements that help AI-enabled interactions work more reliably (for example, product-data taxonomies, and supply and stock signal definitions).

Recommended CEO Considerations: Six Questions that Matter Now

Addressing six questions can help CEOs and their teams pressure-test their current AI agenda and turn it into growth, productivity, and advantage in the demand value chain:

1. Are we aligning investments with strategic priorities? Most companies do not align their AI investments with the processes that they say are most strategic. The leadership challenge is to focus on a set of core commercial priorities before accelerating.
2. Are we ambitious enough, and how do we measure impact? As capabilities mature, the value pool has vast potential for expansion. Leaders should take this into account by thinking ahead of today’s capabilities and targeting improvement based on rapid growth. They should also challenge their teams to unlock value beyond copilot by pursuing autopilot initiatives where applicable—that is, by shifting from using AI to augment human work to allowing it to operate more independently within guardrails, with humans reviewing exceptions. Leaders must be diligent about measurement, tracking meaningful KPIs tied to real value.
3. How do we improve the odds of successful and sustainable transformation? Successful AI transformations start with clarity on how far the organization intends to go. Three approaches contribute to positive outcomes, with different impact at stake. Deploy puts AI into existing work; reshape changes how the work is done; and invent creates new propositions. The larger, more measurable prize comes from redesigning a small number of priority functions that embed AI in the process. That means redefining workflows, governance, and ways of working.
4. What are the broader impacts on the workforce and the operating model? Building an effective AI operating model entails reshaping the functions that power the demand value chain. Work becomes more cross-functional, with human roles shifting toward orchestration, oversight, relationship management, and tradeoff decisions.
5. How should we think about our data assets and tech partnerships? Deep advantage comes from codifying proprietary knowledge and enterprise context. As companies move toward more agentic workflows, they should make more deliberate platform choices while preserving interoperability, portability, and governance.
6. How do we move quickly with losing control of risks and costs? As AI becomes more agentic, companies need to establish strict guardrails for its actions. Three are vital: maintain operational stability with clear rules governing decision-related execution and judgment; monitor return on tokens at the level of workflow unit economics, varying oversight from the deploy approach to the invent approach; and engage the workforce honestly with clear communication about the transformation.

Considerations to Turn AI into Advantage

Companies that have begun to separate from the pack are not just more active in AI—they are also more disciplined. Corresponding to the six questions posed above are six considerations to move AI into advantage:

1. Point your AI at the two or three battles that you’ve chosen to win.
2. Set ambition to anticipate exponential change, and measure it as you would any P&L item. Think multidimensionally, considering what the technology can delivery tomorrow, where you can push for more depth, and how you should transform your ways of working and operating model.
3. Focus on a few reshape and reinvent big bets; co-create with the functions involved in delivery to build the necessary change muscle (for example, R&D, marketing, IT, finance).
4. Redesign roles and workforce strategy to meet transversal needs. Going forward, a large share of human work will shift toward oversight, judgment, orchestration, and relationship management.
5. Build data assets in parallel with deployment, and be consistent on agentic ecosystem choices for autopilot.
6. Include three guardrails in every AI deployment: real-time monitoring of output quality, cost tracked to value, and upskilling.

For CEOs, the next step is to pressure-test whether the current AI agenda is strong enough to create lasting advantage in the demand value chain, narrow the focus to achieving strategic priorities, and execute transformational initiatives to capture value.