What Kotter's Eight-Step Model Looks Like Applied to an AI Transformation

Step one, creating a sense of urgency, requires honest external benchmarking for AI. The urgency case for AI transformation in most UK organisations is genuine: competitors are deploying AI, productivity gaps are widening, and the regulatory environment is creating new requirements. But urgency created through exaggeration or fear produces defensive rather than constructive responses. For AI, the most effective urgency narrative is a specific competitor comparison: 'Our main competitors are achieving [specific outcomes] through AI that we are not currently capturing.'

Step two, building a guiding coalition, requires broader representation for AI than Kotter's original model anticipates. Traditional transformation guiding coalitions are senior leader-heavy. AI transformation coalitions work better when they include practitioner-level AI champions alongside senior leaders. The practitioners provide the ground-level credibility with the broader workforce that senior leader sponsorship alone cannot generate. For AI specifically, include two or three respected professionals from the functions most significantly affected by AI; their involvement in the guiding coalition is more persuasive to their peers than any senior leader communication.

Step three, developing a vision, is where AI transformation most frequently diverges from standard change management approaches. AI transformation visions are often either too vague ('we will be an AI-enabled organisation') or too technologically specific ('we will deploy Microsoft 365 Copilot to 3,000 users by Q3'). An effective AI vision is expressed in terms of what the organisation will be able to do differently for its customers, its workforce, and its stakeholders: 'By [date], our staff will be able to [specific capability], our customers will experience [specific improvement], and our cost base will reflect [specific efficiency].'

Step four, communicating the vision, requires the segmentation approach that AI transformation demands and that Kotter's original model underemphasises. A single organisation-wide AI vision communication is insufficient. Different employee segments have different primary questions and different reactions to the same message. The guiding coalition should communicate differentiated messages to: senior managers (strategic implications, governance responsibilities), frontline managers (team leadership through change, personal development), and frontline employees (role impact, support available, timeline).

Step five, removing obstacles, has a specific AI meaning: the governance, IT, and risk barriers that prevent employees from using AI tools effectively. The most common obstacles in UK AI transformations are: over-restrictive data classification that prevents employees from using AI with work-relevant data; IT security policies that block AI tools; and risk and compliance requirements that are applied to AI without calibration to actual risk levels. The guiding coalition's job in step five is to systematically identify and remove these barriers, not to advocate for regulatory non-compliance but to distinguish genuine risk management from risk aversion that prevents all AI use.

Step six, generating short-term wins, is critical for AI transformation and often underinvested. AI transformation timelines are long. The organisation needs early, visible evidence of AI value before the full programme delivers its expected outcomes. Design short-term wins deliberately: choose early AI deployments specifically for their speed of value delivery and their visibility to senior stakeholders, not just their strategic importance. An AI deployment that produces measurable results in 90 days and can be discussed at the board by month three is more valuable for sustaining transformation momentum than a strategically important but slow-cycle deployment.

Step seven, consolidating gains and producing more change, is where AI transformation diverges most sharply from traditional technology change. The lesson from step six AI wins should directly inform the next phase of deployment. Kotter's model implies a relatively linear progression; AI transformation is better described as a continuous learning cycle, where each deployment phase generates evidence that reshapes the next phase. The guiding coalition should explicitly review and publish what has been learned from each deployment phase before the next begins.

Step eight, anchoring new approaches in the culture, requires connecting AI adoption to the organisation's performance management, talent development, and recognition systems. AI adoption that is celebrated in launch communications but absent from performance reviews will not be sustained. Concrete anchoring actions for AI transformation: include AI capability development in performance objectives for all roles in AI-affected functions; recognise AI-enabled outcomes in the annual talent review process; incorporate AI literacy into the professional development standards for the organisation.