Sales AI Investments Fail When They Automate Bad Processes

The Activity Trap

Companies are pouring resources into AI-powered sales tools, yet most aren't seeing meaningful revenue gains. The problem isn't the technology—it's that organizations are using AI to accelerate broken processes rather than fix them.

Art Harding, GVP of GTM Strategy, Performance and AI Operations at ClickUp, frames the issue bluntly: true sales productivity comes down to four levers—contract value, deal count, win rate, or cycle time. "If you tell a CFO you've improved productivity, she's going to expect you to point to more sales for the same heads, or dramatically lower cost," Harding explained, as first reported by Keith Ferrazzi in Forbes.

The disconnect is stark. Sales teams celebrate AI adoption rates and increased activity metrics while bookings per seller remain flat. More emails sent, more call summaries generated, more coaching sessions logged—none of which necessarily translates to closed deals.

Why it matters

As enterprises move past the initial AI hype cycle into what Harding calls "the trough of disillusionment," the gap between AI investment and sales performance is becoming impossible to ignore. Leaders who can't draw a straight line from their AI tools to revenue outcomes will face tough questions from CFOs and boards.

Rethinking AI Coaching

AI coaching tools represent one of the most popular—and potentially misguided—applications in sales. The fundamental question leaders should ask: if reps weren't listening to their frontline managers before, why would an AI agent change that?

The real opportunity isn't delivering more coaching through AI. It's using AI to identify which sellers are coachable, diagnose where they're stuck, and determine which interventions will actually move performance. This frees managers from administrative work to focus on talent development and acceleration.

Removing Friction, Not Adding Tasks

Elite enterprise sellers are spending a day and a half preparing for major meetings—researching accounts, building slides, gathering competitive intelligence. This represents a fundamental misallocation of high-value talent.

AI should eliminate preparation burdens by synthesizing account history, surfacing buyer priorities, drafting materials, and recommending next moves before sellers enter the room. But the bigger opportunity lies in real-time assistance during live conversations: in-moment pricing guidance, mid-call deal coaching, and competitive positioning surfaced on the fly.

The goal is simple: maximize the time sellers spend in front of prospects actually selling.

The Ramp Time Multiplier

Beyond helping experienced sellers, AI can dramatically compress new hire ramp times. One organization Ferrazzi spoke with collapsed a nine-month seller ramp to a fraction of that duration by teaching only the minimum required to start, then using AI-powered role-play to build skills. As one rep put it: "I'd much rather fail with an agent than with a customer."

Governance and Discipline

Scaling AI in sales requires clear boundaries. Use cases touching customer data, core systems, or high-stakes decisions need tight governance and engineering oversight. Smaller workflow improvements can sit closer to end users. The challenge is the middle ground—deciding what to build versus buy, and who owns ensuring tools actually improve performance.

Harding has reorganized his team around this principle, renaming his enablement function "Go-to-Market Performance" and treating it as product management for sales workloads. "Engineers need guidance on what to build," he noted. "Without product management, who decides if we build or buy?"

The Reset

The path forward requires forcing every AI investment through a tougher filter: Which parts of the sales process genuinely require humans? Which are legacy artifacts no one has questioned? Which investments trace directly to higher contract values, more deals, better win rates, or shorter cycles?

As Harding observed, we've mass-produced food with less nutrition, and mass-produced messages while communication has worsened. "Why are we all so excited about more software?" he asked. "It is time for 1st principle thinking not implementing SaaS memes."

AI is an asymmetric amplifier. Deployed strategically with top performers on high-impact actions, it can generate extraordinary gains. Spread indiscriminately across broken processes, it simply makes mediocrity more expensive.

These insights were first reported by Keith Ferrazzi in Forbes.