Half of Software Vendors Report Low AI Feature Adoption by Users

Software companies are shipping AI features at unprecedented speed, but a significant portion of those capabilities are sitting unused. According to a survey of more than 260 vertical market software (VMS) companies, 51 percent report that fewer than one in four of their customers are actively using the AI features they've built.

The findings, shared by Max Risen, co-founder and president of M&A at Banyan Software, suggest the bottleneck isn't demand—it's readiness, alignment, and the ability to translate AI capabilities into measurable business outcomes.

Why it matters

This adoption gap represents a strategic inflection point for both software vendors and enterprise buyers. As AI development accelerates, the challenge is shifting from "can we build it?" to "will customers actually use it, and will it create value?" Companies investing heavily in AI features risk wasting resources if those tools don't align with customer workflows, change management capacity, or clear business objectives. The data suggests that closer vendor-customer collaboration during development—not just faster shipping—may determine which AI investments pay off.

The velocity paradox

Banyan Software owns and manages over 120 VMS companies that serve specialized industries with highly tailored software solutions. Risen developed the survey to understand how these businesses were navigating AI feature development compared to larger enterprise software providers.

While AI enables faster software development cycles—allowing features to be built, tested, and deployed in weeks rather than months—speed alone doesn't guarantee adoption. Customers express strong interest in AI but remain hesitant to integrate it quickly due to concerns about workflow disruption, security and privacy risks, and the unpredictable nature of generative AI systems.

Many enterprises simply don't yet know how to extract value from AI tools, particularly in vertical markets where large-scale AI adoption is relatively recent.

Usage without outcomes

Ben Schein, Chief AI and Analytics Officer at Domo, describes the challenge in stark terms: capability is easy, deployment is hard, but outcomes are the hardest.

Schein points to a phenomenon he calls "tokenmaxxing," where employees consume large volumes of AI tokens, creating the appearance of strong adoption without corresponding evidence of business value. He recommends that leaders inventory their AI tools to understand what's actually being used, how, and at what cost—expenses may far exceed meaningful productivity gains.

For AI to deliver on its promise, Schein says enterprises need robust data governance to ensure the data feeding AI systems is accessible, high-quality, and secure.

Rethinking the development model

Risen argues that vendors and customers need to work more closely together to determine which AI capabilities to build. The traditional approach of observing user behavior across a customer base may not translate well to AI, where the technology's transformative potential often requires reimagining entire processes rather than simply accelerating existing workflows.

He encourages rapid iteration—deploying features in weeks to test what works—but with tighter customer collaboration to address change management, integration, and security concerns upfront.

Risen offers several recommendations for leaders: move forward with an AI roadmap now, but measure results quickly to redirect investment toward what's working. Resist the urge to build everything in-house; buy when possible, build only when the capability is highly specialized or unavailable. And ensure the CEO champions AI strategy—without top-level commitment, adoption may struggle.

The survey results underscore a fundamental shift: shipping AI features and creating value from them are two different challenges. As Schein puts it, companies shouldn't celebrate usage—they should celebrate results.

These findings were first reported by Dr. Jonathan Reichental in Forbes.