Enterprise AI Adoption Fails Using 1998 ERP Playbooks

The tempo problem

Most enterprise AI programs follow a familiar script: phased rollouts, training curricula, communications cascades, and a go-live date. The structure mirrors ERP implementations from the late 1990s—and that's the problem. The change management playbook organizations rely on was designed for systems that took three years to install and ran for fifteen years. AI capabilities evolve on six-week release cycles.

According to research from Gallup published in Forbes, this mismatch creates a two-order-of-magnitude gap in tempo. Training materials approved in February describe tools that behave differently by May. Use-case libraries document what the technology couldn't do when they were written. Phase gates assume the target holds still between checkpoints—an assumption AI violates continuously.

The artifacts reveal the disconnect. A regional operations manager receives AI tools in week one, learns adoption is a priority in week six, and finally gets formal training in month nine—on capabilities she's been using informally since spring. By the time official programs reach frontline teams, employees have already moved ahead.

Why it matters

The failure isn't cosmetic. In Gallup's survey of 102 CHROs from Fortune 500 companies, 99% call AI strategically important, yet only 26% have embedded AI skills into job expectations or performance goals. Organizations invest heavily in training and centers of excellence but leave the underlying machinery untouched. Managers return from capability-building programs to job descriptions written before the tools existed, approval chains that haven't changed, and annual review cycles assessing them against outdated goals.

The old ERP critique was rigidity—systems froze companies in place. Applied to AI, the same playbook produces the opposite failure: nothing freezes except the plan. The capability underneath keeps moving while organizations defend blueprints that stopped matching reality months earlier.

Three structural barriers

The method persists for clear reasons. First, organizational reflex: under pressure, companies reach for the motion they've rehearsed at scale. Second, funding processes select for it—capital committees approve proposals with fixed milestones and benefits curves, not requests for standing funding with quarterly reassessment. Third, corporate culture penalizes plan revisions. Performance systems don't distinguish between leaders who update designs because conditions changed and those who got it wrong initially. Both look like reversals.

Some functions have solved this. Trading desks reprice daily. Clinical protocols update as evidence accumulates. Revision is the job, so changing position costs nothing. That practice hasn't crossed into general management, where reopening an approved design carries visible credibility costs.

What needs to break

Gallup's analysis, first reported by Forbes contributor Vibhas Ratanjee, identifies three structural changes. First, eliminate go-live dates and observe what survives—most initiatives exist only to reach that artifact. Second, require capital committees to fund cadences rather than curves, with every proposal stating what would invalidate it and what happens next quarter when it does. Third, calendar the reversal before approving the design, protecting whoever leads the reassessment.

The technology moves regardless. Managers are already running transitions without cover, making daily calls about what to automate and where human judgment still matters. The formal program will eventually arrive to train them on work they've been doing alone for months, using materials written when the capability was different.

The details were first reported by Vibhas Ratanjee, who studies leadership and culture at Gallup, writing in Forbes.