Why the first 90 days set a chief artificial intelligence officer's reporting baseline
A newly appointed chief artificial intelligence officer usually inherits ambiguity: no agreed KPI set, no established baseline, and pressure to show progress fast. What gets measured in the first 90 days tends to define the questions the board keeps asking for years afterward.
A 2026 global CEO survey found that a large majority of CEOs said AI success depends more on employee adoption than on the technology itself (F1). That reframes an early scorecard: the priority is not model performance, but how people are actually using, understanding and trusting the tools available to them.
How the role is scoped also varies by company. One general description positions a chief artificial intelligence officer as responsible for implementing AI technology across operational strategy and product work, alongside other executives (F4). Because the mandate is not standardized, the first report often becomes the moment the role's real scope gets defined in practice.
That first report rests on an accurate diagnosis of where AI adoption actually stands today, since early findings tend to shape the strategic choices a chief artificial intelligence officer makes well beyond the first quarter.
“The accuracy of the diagnosis determines the quality of every strategic decision the CAIO makes for the next three years.”
— Arjun Jaggi, Enterprise AI Strategy (as stated in the article byline) · Source: The Chief AI Officer Playbook: What the First 90 Days Must Accomplish · Arjun Jaggi
Sources: IBM Study: CEOs are Reshaping C-suite Roles for the AI Era, Chief AI officer - Wikipedia
Usage and adoption checks
Before reporting anything about impact, a new chief artificial intelligence officer needs a simple picture of who is using approved AI tools, how often, and for what kind of work. This is descriptive data, not a judgment on productivity.
The table below outlines four starting metrics and where the underlying signal typically comes from in a first-quarter review.
| Metric | What it captures | Example data source |
|---|---|---|
| Active usage rate | Share of employees who used an approved AI tool in a given period | Baseline employee survey |
| Tool and feature breadth | Which functions or tools people use versus what is available | Baseline employee survey |
| Usage by department or role | Where adoption is concentrated or absent | Baseline employee survey, segmented |
| Awareness of approved tools | Whether employees know which tools are sanctioned for use | Baseline employee survey |
Skills and readiness checks
Usage numbers only tell part of the story. In a 2025 survey, senior leaders reported that insufficient worker AI skills was the biggest barrier to integrating AI into existing workflows (F2). That makes a skills and readiness read a natural companion to usage data.
A first-90-day report can ask employees how confident they feel using AI in their role, whether they have had any formal introduction to the tools, and whether they know where to go for help. These are self-reported perceptions of confidence and readiness, not measured competency tests, and should be labeled that way.
Sources: The State of AI in the Enterprise - 2026 AI report | Deloitte US
Governance and risk checks
Even an early report benefits from a governance section, however informal. NIST's AI Risk Management Framework is a voluntary framework intended to help organizations build trustworthiness considerations into how AI is designed, developed, used and evaluated (F3); referencing it can anchor language used with legal and risk stakeholders.
Regulatory context is also moving. A vendor guide notes that the EU AI Act's prohibitions took effect in February 2025, and that U.S. federal agencies were required to name a chief AI officer under an OMB memorandum (F5); this claim comes from a vendor with a commercial interest in CAIO advisory work and was not independently cross-checked against primary legal text here.
A useful first-quarter governance check simply asks employees whether they are aware of any policy on approved AI tools, data handling or disclosure. That awareness metric is a leading indicator, not a compliance audit.
Sources: AI Risk Management Framework | NIST, Best Practices for Chief AI Officers: A Practical Guide for CAIOs | SFAI Labs
Early business-impact indicators: self-reported vs. measured
It is tempting to include a productivity or ROI figure in the first board update, but that number needs care before it lands on a board slide.
The CEO and leader figures cited earlier in this article are themselves self-reported perceptions from large, multi-country surveys, not audited organizational outcomes (F1, F2). The same standard should apply to any employee-reported impact metric a chief artificial intelligence officer includes in an early report.
The safest approach is to report what employees reported as impact in a distinct, labeled category, separate from any financial or operational metric the finance or operations team can independently verify.
Sources: IBM Study: CEOs are Reshaping C-suite Roles for the AI Era, The State of AI in the Enterprise - 2026 AI report | Deloitte US
Reporting cadence and board communication
A single snapshot at day 90 is a starting point, not a trend. Re-running the same baseline questions at fixed intervals turns a one-time percentage into a trend line, which tends to be more credible to CFOs and boards evaluating a new executive.
The stages below outline one practical cadence for building that trend without changing the underlying questions each time.
Suggested measurement cadence for a new CAIO
- Day 30 baseline: Run the initial employee survey on usage, skills, readiness and governance awareness before making organizational changes.
- Day 90 report: Present the baseline results to the board as a starting point, not a verdict on performance.
- Month 6 re-survey: Repeat the same questions to see early movement against the baseline.
- Month 12 re-survey: Repeat again to show a full-year trend rather than an isolated figure.
Pitfalls that undermine early credibility
Several habits quietly erode trust in a new chief artificial intelligence officer's early numbers, often before the board even questions the data itself.
The checklist below is offered as an editorial starting point for avoiding the most common ones, not a validated standard.
Frequently asked questions
Who should own the first-90-day metrics baseline, the CAIO or another executive?
Editorial guidance: since the chief artificial intelligence officer's mandate varies by company, ownership is usually clearest when the CAIO leads the baseline survey design but coordinates with HR or L&D on distribution and with legal or risk on governance questions.
How long should the first employee survey take to complete?
Not documented in the supplied sources. As editorial guidance, keeping the baseline survey short and focused on usage, skills, readiness and governance awareness makes it easier to repeat consistently at later checkpoints.
Should the first 90-day report include comparisons to other companies' AI adoption?
The supplied findings do not include a validated external benchmark, so any comparison would rely on unverified figures; it is safer to report your own baseline first and compare only against your own future results.
What if the baseline survey shows low awareness of AI governance policies?
Low awareness is itself a useful early finding: it flags a training or communication gap for the CAIO to address, and referencing a voluntary framework like NIST's AI RMF can help frame the next steps for leadership.
Sources: AI Risk Management Framework | NIST
