Tejas Manohar is the cofounder/co-CEO of Hightouch.getty​Every CMO I talk to is investing in AI. How is it moving the business? Almost no one can say.​That's the disconnect I've heard in conversations with hundreds of marketing leaders over the past year. Nearly every organization is testing AI in some form, whether for content creation, audience targeting or campaign optimization. On the surface, adoption looks strong, and momentum feels real.​But when you ask a more direct question about results, the tone shifts and a clear pattern emerges. Few marketing leaders tell me they're seeing consistent, meaningful impact on performance.​The reason this is happening comes down to how AI is being applied.​Most early efforts with AI have focused on speeding up individual tasks. Teams are producing more content and generating more ideas in less time. Although that can be helpful, it often creates a sense of progress without meaningfully changing outcomes. Content still goes through the same approval cycles. Campaigns still rely on manual coordination. Insights are generated, but turning them into action remains slow and fragmented.​​​Three​ Reasons Why AI FailsThe result is more activity, not necessarily better results. I’ve noticed three issues tend to come up again and again when teams try to scale these efforts.​1. A Lack Of Brand ContextLarge language models (LLMs) are trained on broad, publicly available data, which makes them effective at producing generic outputs but unreliable at capturing a company’s specific voice, positioning and standards. Marketing depends on nuance, and without that understanding, outputs require significant editing or miss the mark entirely. Fixing this requires structuring a company’s brand knowledge, like voice, positioning, what's worked and what hasn’t, so AI systems can actually use it.​2. Limited Access To Proprietary DataThe most valuable signals for marketing live within a company’s own systems, including customer behavior, transaction history and life cycle data. When AI operates without that information, it works from an incomplete view of the customer. That limits its ability to personalize experiences beyond a surface level or drive measurable outcomes.​3. Workflow ComplexityMarketing execution spans multiple tools, teams and approval layers. When AI operates outside of those workflows, it produces outputs that are difficult to act on in real time. Teams end up translating recommendations into execution manually, which slows everything down and reduces the potential impact.​What You Can Do To Succeed With AIThe first step is giving AI access to the right context. The biggest breakthroughs happen when AI can see more than campaign performance in isolation. It needs access to customer behavior, purchase history, engagement signals and business outcomes. Without that context, you're asking AI to make recommendations with only part of the picture.I've spent the last several years working with companies trying to make better use of their customer data, and one pattern shows up consistently. The quality of AI outputs is directly tied to the quality of the context behind them. Even the most advanced models struggle when they're disconnected from the customer and business data that drive real marketing decisions.The second step is embedding AI into the systems where work actually happens. One mistake I see repeatedly is treating AI like a separate destination. Teams generate insights in one platform and then manually translate them into action somewhere else. That creates delays and friction. Instead, AI should be able to influence decisions as campaigns are running.This is something we've thought about extensively while building products for marketers. The challenge is rarely generating another insight. Most organizations already have more insights than they know what to do with. The challenge is operationalizing those insights quickly enough to influence customer behavior. That's where AI can have the greatest impact.The third step is changing how you think about execution. Instead of just using AI to generate ideas or recommendations, consider using it to automate decisions within clearly defined guardrails. Marketers still determine the strategy, define success metrics and approve the rules. But they spend less time managing individual tasks and more time overseeing the systems that execute them.I've found that marketers are often far more willing to trust AI when they can clearly define the objectives and constraints. These teams create systems where AI can make thousands of small optimization decisions while humans remain responsible for strategy, brand stewardship and business outcomes.Finally, measure AI the same way you would measure any other business investment. It's easy to focus on outputs such as content volume, engagement rates or campaign activity because those metrics are readily available. But you'll want to make sure that AI is improving outcomes. Is it increasing revenue? Improving retention? Lowering acquisition costs? Growing customer lifetime value? Those are the metrics that ultimately determine whether an AI initiative is creating value or simply creating more work.​ConclusionFor CMOs, the takeaway is straightforward. AI adoption on its own doesn't guarantee results. The impact depends on whether it's integrated into the core systems that drive marketing and supported by the right contextual data and workflows.​Because in the end, AI won't create a competitive advantage simply because you adopted it. The advantage comes from using it to make smarter decisions than your competitors can, and doing so consistently at scale.Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?