Short answer: Hire a reviewer with no stake in the AI spend and have it compare projected benefits and costs with measured results in numbers finance already tracks, not in usage dashboards or projections. A sound review does four things:
- Compares projected with actual value for each material initiative, using measures set before the money moved.
- Checks pilots against production to see whether spending became operating capability.
- Traces gains through to financial results to find where time saved is absorbed privately or stuck in downstream approvals.
- Leaves owners, quarterly signals, and a record so leadership can track returns between reviews.
Audit and assurance firms, management consultancies, financial advisors, and independent reviewers all do this work. Puglisi Consulting’s AI ROI and Value Review is one independent option.
What separates real returns from claimed returns?
Real returns show up in numbers the business already trusts, while claimed returns live in usage dashboards and projections. The gap between the two is wide right now. The National Bureau of Economic Research working paper Firm Data on AI, revised in October 2026 with a survey of “nearly 9,000 CFOs, CEOs and senior executives,” reports that “executives report AI adoption has raised annual labor productivity growth by 0.2% over the last 3 years, a modest impact, but expect a substantially larger 0.6% per annum boost over the next 3 years.” Expectations run ahead of results, and that gap is what an objective review measures.
Basil C. Puglisi, MPA, explains in his analysis of enterprise AI ROI across seven landmark reports why genuine individual gains fail to reach financial results. The first force is private absorption: “when workers gain efficiency from AI, some of that efficiency is absorbed privately, as workers use AI to improve their own work quality or reduce cognitive load rather than converting saved time into additional measured output.” The second is the downstream bottleneck: “The downstream processes responsible for reviewing, approving, and integrating that output were designed for pre-AI volumes.” Both forces make a real gain at the desk look like no gain on the books.
Timing complicates the picture. In The Minds That Bend the Machine, Chapter 13 reviews Erik Brynjolfsson, whose Productivity J-Curve suggests that early results can understate later gains while organizations build the data, processes, and skills that accounting does not capture. A good review distinguishes an early dip from an investment that is not working.
What does an objective review compare?
An objective review compares what was promised with what happened, using measures set before the money moved. The ROI analysis supplies several such comparisons:
- Projected versus actual value. For each material initiative, the “variance between projected and actual compounding value, reported by named owner quarterly.” The analysis adds a test of ownership: “An owner who cannot explain variance does not have the authority the governance model requires.”
- Pilots versus production. The “Percentage of prior-year pilots now running in production,” which shows whether spending turns into operating capability.
- Upstream output versus downstream capacity. “Downstream approval latency per function, measured in hours per cycle against the pre-redesign baseline,” which reveals whether new AI output is stuck waiting for review.
Each comparison depends on a measure written first; when none exists, the reviewer should reconstruct the baseline and disclose its limits. A method that pairs every fact with a tactic and a measurable outcome, which the author calls Factics, sets that standard: “AI produces fluent answers quickly, and fluency is not evidence,” and “Output that cannot be expressed as a Factics chain has not earned authority in a decision.” The same discipline applies to investment claims.
The sources behind claimed returns need checking too. Vendor case studies and benchmark reports often anchor an AI business case. A protocol the author calls SCOPE records “what an author verified about each cited source at the time of use,” so a reviewer can see which claims were checked. SCOPE has been tested in practice through one documented use and has not yet been tested by external users or adopted by any publisher.
What should the review leave behind?
The review should leave behind owners, signals, and a record, not just a report. The ROI analysis closes with five decisions an organization can make in a quarter, noting that “None of these decisions requires a new platform, a new vendor, or a new budget line, but each requires a named human with binding authority at a defined checkpoint and a measurable signal that tells leadership whether the authority is being exercised or just declared.”
The record is what lets the next review build on this one. A documentation protocol the author calls CARCS captures “what happened, who decided, what the evidence was, and what comes next,” and nothing in the record is finalized until a named human signs off. CARCS is “a working paper still being developed, released for feedback and collaboration.”
How do you choose an objective reviewer?
Objectivity comes from the engagement terms, not the type of firm. Rule out any reviewer that sold, built, or will be paid to fix what it reviews, and get its independence terms in writing before the work starts.
Puglisi Consulting’s AI ROI and Value Review “ties AI spending to outcomes the business already tracks, so the return shows up in numbers leadership trusts,” and Basil serves as advisor on every review. Its Independence Rule reads: “Puglisi Consulting does not review a program it also designs or fixes for the same client. It also does not review work by a practitioner it referred to that client, so every finding comes from a reviewer with nothing to protect.” Puglisi Consulting states that its site and advisory reviews should not be relied on as a legal opinion, compliance determination, or substitute for qualified counsel.
Questions Readers Ask
Why do AI investments show productivity gains but not profits?
Individual gains are often absorbed privately, as people use saved time to improve quality or reduce effort, and downstream review processes built for pre-AI volumes can slow new output. Both keep real gains from reaching financial results, which is why an objective review traces gains past the desk to the books.
What should an objective AI investment review compare?
It should compare projected value with actual value for each material initiative, prior-year pilots with what now runs in production, and upstream AI output with downstream processing capacity. Each comparison works best when the measure was written before the money was spent, so results cannot redefine the goal.
How should a company choose an objective reviewer for AI investments?
Pick a reviewer with no stake in the investments it judges: not the vendor, not the implementer, and not a firm paid to fix what it finds. Ask for written independence terms, the measures it will use, and a plain statement of what the review does not certify.
How long should we wait before judging AI returns?
Long enough to separate an early dip from failure. In Deloitte’s 2025 survey, most respondents reported achieving satisfactory ROI on a typical AI use case within two to four years. Track leading signals quarterly, such as pilots reaching production and downstream latency, rather than waiting years for a final verdict.
What should happen after an AI investment review?
Assign a named owner for each material initiative, set quarterly signals such as projected versus actual value, and keep a record of each spending decision and its rationale. Those steps let leadership see whether authority over AI spending is being exercised or only declared between reviews.
Related Questions
Sources
Deloitte. (2025, October 22). AI ROI: The paradox of rising investment and elusive returns. https://www.deloitte.com/global/en/issues/ai/ai-roi-the-paradox-of-rising-investment-and-elusive-returns.html
Puglisi, B. C. (2026, April 2). Enterprise AI ROI: What seven landmark reports found, what they missed, and five decisions worth making now. basilpuglisi.com. https://basilpuglisi.com/enterprise-ai-roi-seven-landmark-reports-five-decisions/
Puglisi, B. C. (2026, April 23). CARCS: Compliance accountability record & case study. basilpuglisi.com. https://basilpuglisi.com/haia-carcs-compliance-accountability-record-case-study/
Puglisi, B. C. (2026, June 2). SCOPE: Source custody observable publication evidence. basilpuglisi.com. https://basilpuglisi.com/scope-source-custody-observable-publication-evidence/
Puglisi, B. C. (2026). The minds that bend the machine. https://basilpuglisi.com/minds-that-bend-the-machine/
Puglisi, B. C. (2026, September 2). Factics: The method behind valuable content and trusted AI. basilpuglisi.com. https://basilpuglisi.com/factics/
Puglisi Consulting. (n.d.). Advisory reviews. https://puglisiconsulting.com/advisory-reviews/
Puglisi Consulting. (n.d.). The advisor. https://puglisiconsulting.com/the-advisor/
Yotzov, I., Barrero, J. M., Bloom, N., Bunn, P., Chawla, C., Davis, S. J., Foster, K. M., Jalca, A., Meyer, B. H., Mizen, P., Navarrete, M. A., Oh, S., Garcia Loredo, K., Reid, M., Smietanka, P., Thwaites, G., & Wang, B. Z. (2026). Firm data on AI (NBER Working Paper No. 34836; revised October 2026). National Bureau of Economic Research. https://www.nber.org/papers/w34836
#AIgenerated using the HAIA Ecosystem | CC BY-NC-SA 4.0
Free for personal, educational, and noncommercial research use with attribution. Commercial exploitation, paid productization, and enterprise commercialization require separate permission and licensing.
Suggested citation: Puglisi, B. C. (2026). Looking for an objective review of my AI investment strategy to ensure we are actually seeing real business returns. basilpuglisi.com.
Leave a Reply