The news: Marketers are reluctant to hand high-stakes decisions over to AI despite their near universal adoption of AI tools, per StackAdapt’s “The AI Delegation Gap” report conducted by NewtonX.

• 88% of global marketing and advertising professionals report AI-driven performance improvements, including faster manual optimization time, campaign setup or launch, and optimization cycles, and better audience targeting.
• However, marketers are still ignoring AI recommendations. A relative majority (42%) do so because they feel too generic or unrelated to campaigns, and 22% because they don’t align with strategy.
• 33% say a clear explanation or rationale will make them act on an AI recommendation, and 31% want a direct connection to a KPI they care about.

What it means: Advertisers are treating AI as an analyst rather than a strategist, and high usage coupled with frequent rejection of recommendations shows that increasing AI use isn’t improving teams’ trust or the usefulness of AI adoption.

• Benefits increase when AI systems have strong signals on brand voice, customers, objectives, historical performance, and desired outcomes.
• That adds pressure for brands to connect first-party data and measurement infrastructure to AI workflows and tackle data siloes to avoid giving AI fragmented information.

There are also limitations around handing campaign decision-making over to AI. Automation can reduce workload, but if recommendations don’t understand brand tenets, importance of specific KPIs, or campaign strategy, it can add a new task of deciding when to trust the machine.

Recommendations for marketers: The next hurdle of AI adoption includes giving the technology enough business context to make decisions that marketers trust.

• Train AI tools on brand identity, products, prior campaigns, and customer relationships to help shape recommendations, but use only company-approved models when sharing proprietary information.
• Use human judgment to vet AI outputs to avoid generic or strategically misaligned recommendations that fail to aid campaign development.
• Explore creating customized tools, like ChatGPT’s GPTs or Gemini’s Gems, for specific workflows, and train them on specific content tailored to the task at hand.
• Set up “undo” functionalities to ensure decisions can be reversed if they don’t improve workflows or cause brand concerns.