One of 5 topics within Change leadership & adoption.
23 documents from 18 organizations address why people who have the tool do not use it. It is the smallest topic here and the one with the sharpest single insight in the whole theme.
Most of the material treats non-adoption as a communications problem. McKinsey's change-leadership work argues the opposite, and gives the reason: the fears underneath are rational. An employee who says "if the agent gives me the wrong answer and I act on it, it's still my mistake" is not being difficult. They are correctly describing where accountability sits, and no amount of enthusiasm changes that arithmetic.
The practical consequence is that resistance rarely looks like refusal. Deloitte describes the subtler forms - withholding information, bypassing the recommendation, quietly disengaging - and McKinsey names the version that defeats most adoption dashboards: the employee who says yes and changes nothing.
The issues, by agreement
How many independent organizations name each issue as a problem. An issue is only as real as the number of separate publishers that identify it, so the count is the ranking. Bars are organizations, not documents. Where the count reads ours, no publisher here states the issue and the analysis is our own.
Each issue carries the organizations that name it, the numbers behind it, and the remedies proposed - with the concrete steps under each. Every citation points at a section of a named document, so any count here can be checked.
Issue 013 organizations name it2026 evidence
The tool is deployed, nominally accepted, and changes no work
Adoption dashboards record licences issued and sessions opened. The common failure is the employee who agrees, logs in, and does the job exactly as before.
Count what changed in the process, not who logged in. Seats and sessions will rise whether or not anything happens.
Done when Each deployment reports one process metric that could only move if the work changed, licences and logins no longer appear in reporting upward, and any high-usage flat-metric case is recorded as a finding.
Pick one process metric per deployment that would move only if the work changed.0-30 daysCOO
Stop reporting licences, logins and sessions upward.30-90 daysCIO
Where usage is high and the process metric is flat, treat that as the finding.30-90 daysCOO
The evidence — 4 documents
Organization
Document
Position
DeloitteConsultancy · March 2026
Deloitte's AgenticAdopt frameworkOur reading Describes resistance appearing as withholding critical information, bypassing AI recommendations, or disengaging from the process rather than as refusal.Subtler forms of resistance
names it
GenpactConsultancy · June 2026
$18 trillion in trapped AI valueOur reading Reports leaders saying accumulated organizational debts actively limit AI value today, with a large share holding no funded plan to address them.Trapped value behind enterprise debts
names it
McKinsey & CompanyConsultancy · August 2026
Agentic AI change leadershipOur reading Names the pattern precisely: the employee who says yes but does not alter any work, and the tool that is technically deployed but barely used - and warns that reading this as resistance to be managed misreads the signal.The employee who says yes
names it
The Conference BoardInstitution · December 2025
Critical success factors for AIOur reading Argues sustainable transformation needs a top-down vision joined to organizational redesign, rather than deployment alone.Redesigning the organization
proposes a fix
Issue 022 organizations name it2026 evidence
The resistance is rational, and treating it as reluctance guarantees it persists
People decline to rely on a system whose errors they will personally answer for. That is a correct reading of where accountability sits, not a failure of enthusiasm.
This is the finding that reframes the whole theme. McKinsey puts four fears at the heart of agentic resistance and argues that leaders who treat them as obstacles to push through will fail, while leaders who name and design for them will not. The quoted version is the clearest statement of the problem in the research: if the agent is wrong and the employee acted on it, it is still the employee's mistake. Until the accountability question is answered, reliance is irrational and no training programme will produce it.
State who carries the error when someone follows a recommendation and it is wrong. Until that is written down, withholding reliance is the correct individual decision.
Done when Every deployed system has a written statement of who carries the error when a user acts on a wrong output, and that statement has been published to the users of the system rather than only to a risk committee.
For each deployed system, write down who is accountable when a user acts on a wrong output.0-30 daysRisk
Publish it to the users of that system, not just to the risk committee.0-30 daysCOO
Where the answer is "the employee", say so plainly and expect low reliance - or change it.30-90 daysCEO
Design for the four fears rather than messaging at them
Competence, judgment, identity and future relevance are addressed by changes to the work, not by communication about the work.
Done when Each function names which of the four fears applies to it, and at least one structural change per fear is in effect - a review right, an escalation route or a rewritten role definition - not a communication about it.
Name which of the four fears applies in each function before designing the rollout.0-30 daysCHRO
Change something structural per fear - review rights, escalation, role definition.30-90 daysCOO
The evidence — 4 documents
Organization
Document
Position
DeloitteConsultancy · March 2026
Deloitte's AgenticAdopt frameworkOur reading Reports that where foundational understanding, trust in AI decisions and confidence in the implications are missing, resistance is inevitable - and that it seldom appears as outright rejection.The impact of workforce resistance on agentic AI ROI
names it
McKinsey & CompanyConsultancy · August 2026
Agentic AI change leadershipOur reading Locates the resistance below the tool, in competence, judgment, identity and future relevance, and separates leaders who push through such fears from those who design around them.Four fears at the heart of agentic resistance
names it
McKinsey & CompanyConsultancy · August 2026
Agentic AI change leadershipOur reading Records the reasoning directly: if the agent gives the wrong answer and the employee acts on it, it is still the employee's mistake, so people are reluctant to rely on tools they do not fully understand where accuracy, professional judgment and reputation are on the line.Why reliance is withheld
names it
Boston Consulting GroupConsultancy · November 2025
To invest in and upskill peopleOur reading Frames capability investment as the route through, which addresses competence but not the accountability question underneath it.Investing in people
proposes a fix
Who is represented
This dossier is drawn from 20 organizations working on the subject, 5 of which are cited directly in the issues above.
Consultancy — 11
McKinsey & Company 3Deloitte 2Boston Consulting Group 1Genpact 1KPMG 2Accenture 1Capgemini 1Cognizant 1EY 1PwC Malaysia 1UST 1
Institution — 4
The Conference Board 1Cloud Security Alliance 2IAB 1NIST 1