Incentives & what gets measured

The state of it

One of 5 topics within Change leadership & adoption.

30 documents from 20 organizations address what the organization rewards. The material here is thinner than the subject deserves, and the most interesting entry is not a change-management document at all.

Equilar tracks AI appearing as an executive performance metric - the point at which adoption stops being an initiative and becomes compensation. BCG's survey supplies the misalignment underneath it: CEOs see AI return weighing on their own performance evaluation more heavily than their boards do. When the person accountable and the body assessing them disagree about how much it counts, the incentive is unclear at the top before it reaches anyone else.

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.

The issues in full

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 011 organization name it2026 evidence

Nobody owns the decision about whether a human or an agent does the work

The choice between assigning work to a person and assigning it to an agent is now a live, recurring allocation decision. It sits with no function, is made informally inside teams, and is measured by nobody.

Microsoft, reporting from a forum of 250 customers running this at scale, frames the shift precisely: AI used to be assessed against the IT budget, a number leaders already understood in a category they already managed. Once it can do real work the relevant comparison stops being a software line and becomes the cost of a person doing the same thing - which is a different decision, taken by different people, against quality and time as well as cost. It also does not stay put, because the cost of the machine side falls quarter by quarter. So the organization has acquired a standing allocation question it has no owner for, no method for, and no record of having answered.

How to fix it — 1 approach, 3 steps

Give the human-or-agent decision an owner and a method

Name who decides, on what evidence, and how often it is revisited. An unowned recurring decision defaults to whoever moves first.

Done when One role is named as signing off the move of a task from a person to an agent, the sign-off record shows quality, time and fully-loaded cost on both sides, and a revisit date is set.

Name the role that signs off moving a task from a person to an agent.0-30 daysCOO
Require quality, time and fully-loaded cost on both sides before the switch.30-90 daysCFO
Revisit each decision on a fixed cadence, because the cost side moves.ongoingCFO

The evidence — 2 documents

Organization	Document	Position
MicrosoftHyperscaler · June 2026	Tokenomics is the new headcountOur reading Sets out the reframing: the comparison moves from an IT line item to the cost of a person doing the same work, and the answer keeps changing because the machine side keeps getting cheaper.Tokenomics is the new headcount	names it
EquilarInstitution · February 2026	AI as a performance metricOur reading Proposes putting the outcome into how executives are measured, which is the only mechanism that would make an unowned allocation decision somebody's job.AI as a performance metric	proposes a fix

Issue 021 organization name it2026 evidence

The CEO and the board disagree about how much AI return counts

Executives believe AI results weigh on their assessment more than the body doing the assessing does. An incentive that unclear at the top does not clarify further down.

How to fix it — 1 approach, 3 steps

Put it in the objective or accept it is optional

If AI outcomes are not in someone's written objectives, they are competing with things that are. Decide, and say which.

Done when Either AI outcomes appear in executive objectives with a measure agreed with the remuneration committee, or the organization has stopped describing AI as a priority. Both are answers; silence is not.

Establish whether AI outcomes appear in executive objectives today, in writing.0-30 daysCHRO
If they should, agree the measure with the remuneration committee before the cycle.30-90 daysBoard Chair
If they should not, stop describing it as a priority.0-30 daysCEO

The evidence — 1 document

Organization	Document	Position
Boston Consulting GroupConsultancy · May 2026	CEOs & Board DividedOur reading Reports CEOs seeing AI return on investment as a bigger factor in their performance evaluation than boards do, across a survey of 625 CEOs and board members.Split decisions survey: five areas of divergence	names it

Who is represented

This dossier is drawn from 20 organizations working on the subject, 3 of which are cited directly in the issues above.

Consultancy — 7

Boston Consulting Group 2 Deloitte 3 McKinsey & Company 3 Capgemini 2 EY 1 Genpact 1 Infosys 1

Institution — 6

Equilar 1 NIST 2 Citi 1 Cloud Security Alliance 1 The Conference Board 1 World Economic Forum 1

Academic — 2

Carnegie Mellon SEI 1 National Bureau of Economic Research 1

Hyperscaler — 2

Microsoft 2 IBM 3

Frontier lab — 1

OpenAI 1

Enterprise — 1

Palantir 1

Other — 1

Dario Amodei (Anthropic) 2