From the source
to the score.

01 · Research approach

Peer-based research. No vendor funding.

The Enterprise AI Pulse is a peer-based study. Fortune 1000 and Global 2000 practitioners complete the instrument — IT and AI leadership, sourcing and procurement. Vendors do not participate, and vendors do not fund the research.

Cadence is quarterly. The published study design uses a 30-day field window, 30-day analysis window, and 30-day publication window.

02 · Sample design

Who qualifies, who responds, and what is published about them.

Qualifying enterprises. Fortune 1000 and Global 2000 companies. The eligibility floor enforced by the instrument is $1B+ annual revenue and 2,000+ employees; respondents from smaller organizations are recorded for panel development but excluded from index computation.

Respondent roles. The current invitation is for IT and AI leadership—CTO, VP AI, Head of AI Strategy, and equivalent operating titles—and sourcing and procurement leaders, including Head of IT Sourcing, VP Procurement, and CPO. Participants need budget authority or direct influence on AI vendor and infrastructure decisions. The instrument routes questions according to role.

Target sample.n=150 per wave. The first Real-Time Update (RTU) is published at n=30+; index scores are marked “preliminary” until the wave closes at the target.

03 · The four proprietary indices

What each index measures and what goes into it.

The Pulse produces four indices per wave. Each is constructed from a defined subset of instrument responses; each has its own category weighting and its own interpretation scale. We publish the construction formulas openly so the numbers can be checked.

INDEX Heat Index

Near-term adoption momentum across 10 emerging technologies.

Heat measures how close an emerging technology is to entering the enterprise stack. It is constructed as a diffusion-index analog to the ISM Purchasing Managers’ Index: each respondent classifies each of 10 technologies on a six-tier maturity scale, and the population share at each tier is aggregated into a single category score.

The six-tier scale is shared with the core-category status questions: in-use, pilot, evaluating, near-term (6–12 months), long-term (12+ months), not-planned. Higher tiers contribute more weight to the category score.

INPUTSStatus responses across 10 emerging-tech items · 6-tier maturity scale. INDEX AI Maturity Index

0–100 composite of penetration, vendor stack coverage, commitment depth, and governance discipline.

Maturity is the headline single-respondent score that ships on first response. It is a 100-point composite across four weighted dimensions: Penetration (30) — share of lines of business with active AI; Stack Coverage (15) — how many of the 6 core vendor categories have a named primary vendor in production; Commitment (20) — AI as a share of total IT spend, plus directional trend; and Governance & Ops (35) — whether a documented program exists and is enforced, board-level sponsorship, regulatory readiness, and shadow-AI containment.

Stage anchors: 0–25 Exploring, 25–50 Establishing, 50–75 Scaling, 75–100 Realized. Reported alongside the cohort percentile.

INPUTSLOB penetration · 6 core-category vendor responses · AI-as-%-of-IT spend · governance program + risk framework + regulatory readiness. INDEX Spend Index

Peer-benchmarked budget direction by category.

The Spend Index is constructed as a diffusion index analog to ISM PMI. Respondents report the directional change in their category spend versus the prior period — increasing, flat, decreasing— and the category score is computed as (% increasing) + (0.5 × % flat). Output is bounded 0–100; 50 is neutral; above 50 is expansion; below 50 is contraction.

The headline score is a weighted average across six categories. Initial weights reflect category share of AI budget: Foundation Models 25%, AI Compute 25%, Data Foundation 20%, Consulting & SIs 15%, AI Software Engineering 10%, AI Governance & Risk 5%. Weights are published with every wave and revised from empirical spend data beginning Wave 3.

Anchoring: 60–100 expanding, 45–59 neutral, 0–44contracting. An Early Warning signal fires on three consecutive wave declines or on simultaneous contractions in Foundation Models and Compute; thresholds are held constant across waves. The full construction brief — including the rationale for choosing a PMI-style diffusion index over CPI-style basket construction and the LEI-style composite architecture we are designing toward — is available on request.

INPUTSDirectional spend change per category · per-respondent weighting · aggregate weighted across 6 categories. INDEX Vulnerability Index

Vendor switching pressure across the 6 core categories.

Vulnerability measures which vendors in the panel’s installed base are at risk of replacement. For each of the 6 core categories, respondents report their primary vendor, switching intent (not considering, considering, or actively displacing), and any displacement-target vendor. The category-level Vulnerability Score is the share of the panel signaling switching pressure, weighted by stated spend exposure.

The headline reports both the most-vulnerable legacy vendor for the cohort and the displacement-target capturing the most switching intent. Wave-over-wave deltas track whether vulnerability is widening or narrowing per category.

INPUTSPrimary vendor + switching intent + displacement target per core category · weighted by stated spend.

AI Yield Index.The Wave 2 evolution of the Maturity Index. Where Maturity measures where your AI program stands, Yield measures what that maturity converts into — a wave-over-wave delta of business impact across lines of business, combining severity (how disruptive removal of AI would be) and velocity (pipeline-to-production rate). It cannot be computed from a single response, which is why it activates only with your second response, once a baseline exists.

04 · Question design

How the instrument is built.

The Wave 1 instrument contains roughly 150 underlying questions across five sections — demographics and qualification, six core vendor categories, an emerging-tech Heat Index, AI governance, and the Yield section that powers wave-over-wave deltas from Wave 2 onward. Each respondent sees between 35 and 60 depending on deployment status and vendor count; the instrument routes around questions that do not apply. The full question bank is auditable; it is available on request and will be published as a dedicated reference page in a subsequent release.

Maturity scale. Status questions use a consistent six-tier scale across all six core categories and all 10 emerging-tech items: in-use, pilot, evaluating, near-term (6–12 months), long-term (12+ months), not-planned. Using one scale across the instrument makes cross-category comparison mechanical rather than interpretive.

Vendor pick-lists. Each core category exposes a vendor pick-list composed from three sources: (a) known market-share leaders, (b) adoption-threshold candidates surfaced by practitioner signal in prior waves, and (c) custom / internal / other as first-class options. On governance categories, in-house / home-grownis surfaced explicitly — because a large share of Fortune 1000 AI governance is in-house today, and collapsing it into “other” would distort the Adoption Index.

Editorial pass. The instrument itself is reviewed by an internal editorial pass each quarter. The review examines every wording change from the prior wave, confirms that additions or removals are justified by practitioner signal, and preserves comparability for the headline indices. A per-wave changelog is published alongside results.

05 · Data governance

What happens to a respondent’s data.

All responses are aggregated before publication. Individual responses are never shared with vendors under any commercial arrangement. Respondent identity is never disclosed to sponsors.

Anonymization occurs before any downstream analysis. Respondent identifiers are stored separately from instrument data, and the analysis pipeline operates on pseudonymized records. When a segment has fewer than five respondents, results for that segment are either suppressed or merged into a broader cohort so that a single respondent cannot be identified.

Retention. Respondent-level data is retained for up to 24 months for longitudinal quality checks and index calibration, then deleted or de-identified. Aggregate indices and anonymized segment cuts are retained indefinitely. Arcana does not sell respondent data to third parties and has no mechanism to do so.

Withdrawal and access. Respondents may withdraw their responses within 14 days of submission (and before the wave’s publication date), and may request a copy of their individual responses or a complete deletion of their record. Contact ken@arcana-research.com with the respondent code displayed in the survey header. Withdrawal removes respondent-level data from the pool; aggregate indices already published stand, since they cannot be reconstructed at the individual level.

Research ethics. The published methodology commits to informed consent (Page 1 of the instrument), voluntary participation (Skip is first-class), confidentiality (aggregation before publication, never vendor-shared), and the right to withdraw. No deceptive framing, no hidden-purpose questions, no personal or sensitive data collected beyond what’s needed for the research. This research follows the ICC/ESOMAR International Code on Market, Opinion and Social Research.

06 · Team

Who designed the instrument.

Ken Male, Chief Research Officer. Thirty years in IT research leadership. Joined the founding team at Giga Information Group (founded by Gideon Gartner; acquired by Forrester Research) and later held senior roles at Gartner, Apptio, and Jupiter. In 2001 he founded TheInfoPro, where he pioneered peer-based IT research devoid of analyst spin or bias — an approach that became known as the “Voice of the Enterprise.” TheInfoPro was acquired by S&P Global Market Intelligence. Ken designed the four Arcana indices.

Ian Kar, Managing Partner. Platform architecture, instrument design, and go-to-market. A decade identifying and building at the frontier of emerging technology markets. Co-founded Fintech Today, a research and community platform for fintech operators and founders — acquired in 2022. Earlier: product manager at Acorns; writing on fintech and enterprise technology has appeared in Quartz, Newsweek, American Banker, the World Economic Forum, and Stanford Business Review, and has been cited by Bloomberg Businessweek, CNBC, Barron’s, and TIME.

Research advisory. NPI provides price intelligence and panel development support. Named sourcing and sector advisors contribute to instrument review; disclosures for individual advisors are available on request.

07 · Publication schedule

When the numbers become public.

Quarterly cadence. The published design uses a 30/30/30 cycle: 30 days in field, 30 days in analysis, and 30 days to publication. Contact the team for current field and publication dates.

Historical wave dates and pending sample-composition announcements have not been carried forward as current status. The research team can provide the latest publication schedule and composition disclosures.

08 · Limitations

What the instrument cannot measure.

Self-reported status. All responses are self-reported. Respondents may over- or under- state production readiness; the consistent six-tier maturity scale is designed to reduce ambiguity, but it does not eliminate it.

Wave 1 sample and confidence. Wave 1 targets n=150. Wave 1 aggregates should be interpreted with the confidence intervals that will be published at wave close. Category-level comparisons and sub-segment cuts with fewer than 30 respondents are not reported at the headline level.

Geographic concentration. The panel is weighted toward North American and English-speaking global enterprises in Wave 1. Geographic expansion is planned as the panel grows; until then, regional cuts should not be interpreted as globally representative.

Category completeness. The instrument covers six core categories and 10 emerging-tech items — 16 categories total. This taxonomy is not exhaustive; it is revised quarterly based on practitioner signal. Categories that fall out of use are retired; new categories enter the Heat Index before graduating to core.

09 · Conflict-of-interest disclosure

How Arcana is paid, and by whom.

Arcana does not accept vendor sponsorship for the Pulse study. Vendors do not fund the instrument, do not review it before publication, and do not receive respondent-level data under any commercial arrangement. No vendor has pre-publication review rights. No vendor has influence on index weights or thresholds.

Arcana’s revenue comes from two sources: (a) enterprise membership paid by the employers of respondents and (b) licensing of aggregate analyses to institutional buyers — venture capital firms, private equity firms, and strategy teams at non-vendor corporations. Membership and licensing are at the aggregate layer only.

Ken Male and Ian Kar individually disclose relevant holdings and advisory roles. Individual disclosures are available on request.

10 · Audit trail and contact

How to verify, question, or request the underlying material.

The research methodology is published so its design and assumptions can be inspected. Contact the research team for the applicable instrument version and wave changelog.

Questions, audit requests, and requests for the full instrument, the Spend Index construction brief, or individual advisor disclosures should go to ken@arcana-research.com. Press and institutional inquiries may also reach Ken Male directly at ken@arcana-research.com or Ian Kar at ian@arcana-research.com.