Business strategy first, AI strategy second
Allen Bates, Unilever North America's chief digital and information officer, describes the company's approach to AI in a way that deliberately avoids treating it as its own category of initiative. The company's investments run through a business strategy that happens to be powered by AI, rather than through a parallel AI roadmap that competes with the rest of the business for funding and attention. That framing is a quiet rebuke of the chief AI officer model other large enterprises have adopted this year, where AI gets its own executive, its own budget line, and its own set of metrics disconnected from the operating units it is meant to serve. Bates runs digital and information for North America specifically, a scope that keeps him close enough to plant floors and brand teams to insist every AI dollar answer to a number those teams already report.
The practical effect at Unilever is that every AI investment Bates approves has to point to an existing business metric a plant manager, a brand team, or a finance partner already tracks. That discipline shows up concretely in two places the company has been willing to put real numbers behind: manufacturing capacity and marketing production speed, both measured against baselines that existed before any AI tool entered the picture. Neither number depends on a reader trusting an AI specific benchmark invented for the occasion, which is precisely why Bates is comfortable citing them outside the organization.
A 10 percent capacity gain on the factory floor
At Unilever's manufacturing plant in Raeford, North Carolina, AI powered digital twins contributed to a 10 percent increase in capacity, a figure Bates ties directly to the plant's operating metrics rather than to an isolated AI project scorecard. A digital twin in this context models the physical production line closely enough that engineers can test changes virtually before touching the real equipment, catching bottlenecks and failure points that would otherwise only surface after a costly line stoppage.
That 10 percent is the kind of number a plant manager already reports up the chain regardless of what technology produced it, which is precisely the point of Bates's approach. The AI tool does not get a separate line item claiming credit for the gain, it gets folded into the capacity number the business was already measuring, which makes the return far harder to dispute in a budget review than a standalone AI pilot metric would be.
Beauty AI Studio and the marketing speed bet
On the marketing side, Unilever's Beauty AI Studio has helped the company produce creative assets up to 30 percent faster, a gain built on top of a five year partnership with Google Cloud to establish an AI first technology foundation across the business. Faster asset production matters specifically in beauty and personal care, where campaign cycles move quickly and the cost of creative production has historically been one of the larger line items in a brand's marketing budget.
Unilever CEO Fernando Fernandez referenced AI driven marketing gains on the company's Q2 2026 earnings call, giving Bates's technology investments visible support from the top of the organization. That kind of executive sponsorship matters for a CIO trying to justify continued AI spend internally, because it moves the conversation from a technology team's internal metric to a number the CEO is already citing to investors. It also raises the stakes if the gains stall, since a figure cited publicly on an earnings call is harder to quietly walk back than an internal dashboard metric nobody outside IT ever sees.
Change is everyone's job, not IT's job
Bates is explicit that the change management required to get a plant floor or a marketing team to actually use a new tool differently than they used the old one is the harder part of this work, well beyond the technology itself. His description of the principle is direct: change is everyone's job, in his view, a responsibility spread across the organization rather than one outsourced to IT or to a dedicated transformation office while the rest of the organization waits to be told what changed.
That distributed ownership model is a deliberate contrast to the common pattern of standing up a center of excellence that owns AI change management on behalf of every business unit. Bates's version pushes accountability for adoption back onto the managers and teams closest to the work, with IT functioning as the enabler of the tools rather than the enforcer of their use, a split that tends to produce slower initial rollout but more durable adoption once it takes hold.
What other CIOs should take from Unilever's numbers
The most transferable lesson from Bates's approach is not the specific technology choices, digital twins and a generative creative studio are available from multiple vendors, it is the discipline of refusing to let any AI initiative report its value on a metric the business was not already tracking. That constraint forces a harder conversation up front about what outcome a given AI investment is supposed to move, but it produces numbers that survive scrutiny in a way a standalone AI adoption metric rarely does.
For CIOs still building out a 2027 AI roadmap, Unilever's two data points, a 10 percent capacity gain and a 30 percent production speed gain, are useful less as targets to match and more as a model for how to frame any AI result going forward: against the business metric that already existed, not against the AI project's own internal yardstick. A CIO who adopts that single rule before the next budget cycle will spend far less time defending AI spend in the abstract and far more time pointing at numbers the business already trusted before the technology showed up.
