Introduction
A promising AI pilot can generate product descriptions, audience segments, or sales forecasts, yet still collapse once it leaves the team that created it. That is the core warning from a new readiness study by Adobe and Incisiv, with direct relevance to smaller consumer-goods brands operating with limited staff, tighter budgets, and less influence over retail partners.
What Happened
The research drew on responses from 5,633 executives across 12 industries and 11 markets, including 521 professionals from the consumer goods sector. The researchers found that 61% of consumer-goods participants sit in a category they call the “Frozen Middle”: artificial intelligence is active inside individual departments, but the organization has not built the cross-functional coordination required to move those experiments into broader use.
The report notes that the bottleneck is seldom the AI model itself. Instead, a strong use case often stalls the moment it requires resources outside the originating team—such as customer data, legal sign-off, retailer approval, software integration, or a method for measuring sales impact.
Additional figures from the study include:
- Just 16% of consumer-goods respondents reported having a designated AI owner who holds both a title and budget authority.
- Only 15 of every 100 proposed AI use cases advance to production.
- About three-quarters of respondents continue to manage business and IT relationships through siloed, transactional, or consultative arrangements.
- Just 12% evaluate AI’s financial return in a comprehensive manner.
The data also shows where momentum fades. The biggest absolute drop takes place during IT evaluation, while the steepest proportional losses happen later, during integration, budgeting, and production-readiness reviews.
Why This Matters
For a smaller consumer-goods brand, a stalled AI initiative can consume an outsized portion of limited time and budget. A marketing team might use generative AI to draft ad copy, while a sales team builds a retail-media audience, but without clear ownership of data access, partner approvals, and success metrics, the project can stall before becoming part of daily operations.
External dependencies add another layer of risk. A retail-media effort can involve brand leaders, ecommerce staff, technology providers, publishers, distributors, and marketplaces. A small business may not control the data-sharing terms, approval timelines, or technical requirements set by a retail partner, meaning an internal AI workflow can stop the moment it reaches an outside organization.
Key Takeaways
- AI readiness depends more on ownership, early measurement, and cross-functional teamwork than on simply purchasing new technology.
- Small brands should start with narrowly defined use cases, such as automating repetitive product content or summarizing customer feedback, which typically involve fewer external dependencies.
- Measurement should begin before launch. Establishing baselines for conversion rates, average order value, repeat purchases, or production costs helps create a credible basis for judging results.
- Companies should still assess privacy, accuracy, and brand-safety risks, particularly when AI produces customer-facing content or relies on personal data.
- Identifying an internal champion—such as a marketing manager, operations lead, or salesperson—and giving that person clear authority, time, and a defined budget can be more effective than forming a large committee.
Conclusion
The study’s most advanced group, 342 organizations representing the top 6% of active respondents, offers a useful benchmark. These “Leaders” were nearly six times more likely than consumer-goods companies to have a formal AI owner, and 44% measured AI returns comprehensively compared with just 12% in the consumer goods sector. They also ran business and IT at a collaborative or integrated level far more often than the broader consumer packaged goods group.
The broader lesson is that AI readiness goes beyond choosing a model or software subscription. Consumer-goods companies that define ownership, bring in technology and legal advisers early, agree on meaningful measurements, and account for partner dependencies will be better positioned to turn isolated AI experiments into repeatable capabilities.

Discussion
Join the conversation
Thoughtful reactions, questions, and follow-up ideas help shape the next story.