Why do AI change management plans fail to produce adoption?
AI Navi's Navigate-Execute-Land methodology treats the Land phase as a live sequence of decisions, not a document. Strategy sits in Navigate. Delivery sits in Execute. The moment the team goes from receiving a system to owning its output sits in Land, and Land cannot be scheduled in advance.
The standard approach to AI change management produces two things: a training programme and a communication plan. Both are written during the Execute phase, based on stakeholder mapping done before the system is complete, before the team has seen its output, and before the commercial implications of using it have become concrete enough to raise. The plan is then handed to the business at go-live alongside the system itself.
The problem is sequencing. At a £40M UK food brand midway through an AI-enabled deduction recovery programme, the commercial team raised a concern three days before launch that had not appeared in any stakeholder mapping exercise: several of the flagged deductions involved retailers the business had traded with for over a decade. Who was going to make those calls, and what would they say?
That question is not a change management problem in the abstract. It is a commercial conversation that requires someone present and senior enough to answer it in the language the team uses, before the moment of adoption, not in a plan that was finished before the question existed.
What happened inside the Land phase at the £40M food brand?
The deduction recovery programme had been built to scope. The AI matched retailer claims against order and delivery records and surfaced recoverable deductions in 15 minutes per case rather than three hours. By the end of the eighth week, 60% of previously unchallenged deductions had been overturned. None of that was in doubt. What was in doubt was whether the commercial team would use the system at all.
Three things happened in the Land phase that turned a stall risk into a 60% recovery:
1. The AI output was translated into a commercial argument, not a technical explanation.
The session held with the commercial and finance teams before go-live did not explain how the matching model worked. It explained what to say to a named retailer buyer when challenging a specific deduction: here is the paperwork, here is the dispute window, here is what the brand recovers if the challenge succeeds, here is why this is now the right moment to raise it. The team left with something they could repeat in a conversation with a supplier, not just something they had seen demonstrated on a screen.
2. The first challenges were co-owned, not handed over.
A senior person worked alongside the finance analyst to file the first ten disputes. Not to teach. To co-own the outcome. The difference is accountability. If the first challenge goes badly and the analyst has been left to file it alone, the system gets blamed. If it goes badly and a senior person is in the room, the conversation is about what to try differently next time. By week four, the analyst was filing independently. By week six, the team had built enough history with each retailer's dispute portal that they no longer needed the process explained.
3. Adoption was measured as a commercial KPI from week one.
The number of disputes raised per week, investigation time per case, and overturn rate appeared in the weekly commercial review from the first week of live operation. Not in a project update. In the same call where revenue, margin, and case fill rate were reviewed. When adoption metrics sit in a project report that reaches the implementation team, they disappear from commercial attention within two weeks of launch. When they sit in the commercial review, the finance director notices if the number drops and asks why.
What is the parachute consultant problem in AI change management?
The fractional leadership model exists partly because of a specific failure mode in consulting engagements: the senior person who understood the business well enough to address the commercial team's real concerns is no longer present when those concerns surface.
A consulting engagement delivers a system and a set of accompanying documents, including a change management plan. The plan is produced during Execute by people who have spoken to the stakeholders but have not yet seen the system in the hands of the people who will use it commercially. At go-live, the consulting team exits. If a concern surfaces three days later, the business contacts the project manager, not the senior lead who understood both the commercial context and the technical delivery.
Fractional leadership keeps the senior accountable person embedded through the Land phase, which is exactly the period when the concerns that were never in the plan arrive. McKinsey’s most recent global survey of 750 leaders, conducted between February and April 2026, found that organisational readiness explained 48% of the gap between companies capturing value from AI and those that did not, compared with 25% for personal readiness. Companies that redesigned workflows were 5.3 times more likely to report enterprise value capture than those that left workflows unchanged. Workflow redesign is a live activity, not a document. It happens in the weeks after go-live, between people, and it requires a senior person who is still in the building.
AI Navi has covered the general cost of change management being treated as optional, and what to budget for adoption, in separate pieces linked at the end of this article. The Navigate-Execute-Land framework provides the structural context for why the Land phase is specifically where the fractional model produces different outcomes: the senior person’s accountability does not end at go-live.
Frequently Asked Questions
What is the Navigate-Execute-Land methodology?
Navigate-Execute-Land is AI Navi's three-phase delivery framework. Navigate covers strategy and scoping. Execute delivers the first working AI system in production. Land builds internal ownership and measures adoption until the team runs the process independently. The Land phase is the most frequently underresourced of the three.
Why do AI change management plans fail to produce adoption in UK FMCG?
They are usually written during Execute, before the team has seen the system’s output and before the commercial implications of using it have surfaced. The concerns that determine adoption appear in the days around go-live, not in a stakeholder mapping session held during the build.
What three tactics worked at the £40M UK food brand?
Translating the AI output into the commercial argument the team could repeat in a buyer conversation. Co-owning the first ten disputes alongside the finance analyst rather than handing the system over and stepping back. Tracking disputes raised, investigation time, and overturn rate in the weekly commercial review from week one.
What is the difference between a fractional CAIO and a consultant in the Land phase?
A consultant produces a change management plan and exits at or shortly after go-live. A fractional CAIO stays embedded through the Land phase and addresses concerns as they arise, in commercial language, with the authority to make decisions about how the programme adapts. The distinction matters specifically in the two to four weeks after launch, when the real adoption issues surface.
How long does the Land phase last in an AI FlightPath™ Sprint?
The Land phase runs from go-live to roughly week ten, with the accountable senior person stepping back progressively as the team builds independent capability. For deduction recovery specifically, most teams reach independent operation by week six to eight.
What does the SCALE AI™ Leadership dimension actually measure?
Whether the person accountable for an AI programme has communicated why the change is happening to the team responsible for using it, before deployment. It is the lowest-scoring dimension across UK mid-market CPG and FMCG clients in AI Navi’s FlightCheck™ diagnostics, averaging 18% of benchmark.
Next step
Book an AI FlightCheck™ diagnostic to find your Leadership dimension score and identify where the Land phase is most at risk in your current AI programmes. Or take the AI Readiness Scorecard for a first read in under ten minutes.
Related reading: The cost of treating change management as optional and what to budget for adoption are covered in AI Navi’s two existing change management pieces linked above.