Start with the scope of the evidence
These sources combine enterprise surveys, software-provider datasets and a survey of U.S. workers. They measure different things and should not be treated as one benchmark for a firm of 10 to 250 employees. The figures below retain their source populations and reporting periods.
AccruePilot's suggested review questions are practical prompts, not conclusions established by the studies.
AI spend statistics at a glance
These findings describe different samples and measures. They are context for a review, not a benchmark or savings target for your firm.
| Finding | Population and limit | Publisher and release |
|---|---|---|
| 74% included AI cost reviews; 43% used usage or token budgets | 314 U.S. leaders at companies with at least US$1 billion annual revenue. | KPMG, September 24 |
| 31% reported accurate AI-software visibility; 59% reported increased wasted AI spend | 512 technology professionals worldwide. The latter measures respondents reporting an increase, not dollars wasted. | Flexera, June 24 |
| AI-native application spending increased 108% | Zylo's analyzed dataset. This is not a forecast of spending growth at a small firm. | Zylo, January 29 |
| 78% reported unexpected consumption or AI pricing charges | A separate survey of 218 IT leaders reported by Zylo. | Zylo, January 29 |
| 52% used AI in their role; 30% used it a few times a week or more | U.S. workers' self-reported use. This does not measure subscriptions or spending. | Gallup, July 20 |
Table sources checked October 28, 2026. Each study's context and practical review question appear below.
How are enterprise leaders reviewing AI costs?
In KPMG's U.S. Q3 2026 AI Pulse survey, 74% of respondents said their organizations included cost reviews in AI approval processes, compared with 61% in the previous quarter. 43% reported implementing usage or token budgets.
The survey covered 314 U.S. C-suite and business leaders at organizations with annual revenue of at least US$1 billion. Responses were collected from July 24 to August 25, 2026. These are enterprise findings, not a survey of small professional-service firms.
Source: KPMG US, Q3 2026 AI Pulse release, September 24, 2026. Reviewed October 28, 2026.
Review question: Who approves a new AI purchase, and where are the approved cost, purpose and next review date recorded?
Visibility into AI software is incomplete
Flexera's 2026 State of ITAM Report found that 31% of respondents reported accurate visibility into AI software. 59% said wasted AI spend had increased over the previous year. The report reflects a worldwide survey of 512 technology professionals.
Source: Flexera, 2026 State of ITAM Report press release, June 24, 2026. Reviewed October 28, 2026.
Flexera also reports that maintaining an accurate software inventory was an existing responsibility for 78% of IT asset management teams, while AI tracking was a planned responsibility for 47%.
Source: Flexera, 2026 State of ITAM results and responsibility chart, June 24, 2026. Reviewed October 23, 2026.
Review question: Can your team produce one current list of AI subscriptions, with an owner, renewal date and billing model for each?
AI spending and software billing need separate checks
Zylo's January 2026 SaaS Management Index release reports that spending on AI-native applications increased 108% year over year across its analyzed dataset. The release describes data covering more than 40 million SaaS licenses and US$75 billion in managed spending.
In a separate survey of 218 IT leaders included in the release, 78% reported unexpected charges connected to consumption-based or AI pricing in the preceding 12 months.
Zylo also reported an average 36% of SaaS licenses unused when measured against industry-recommended utilization levels. This is a software-license measure across its dataset, not the percentage of AI spend wasted at a typical small firm.
Source: Zylo, 2026 SaaS Management Index announcement, January 29, 2026. Reviewed October 28, 2026.
Review question: For each tool, have you checked both the seats being paid for and any usage-based charges that can appear alongside them?
Does employee AI use measure subscription spending?
No. Employee use and subscription spending are different measures. Gallup's Q2 2026 reporting found that 52% of U.S. workers used AI in their role, with 30% using it a few times a week or more. This measures employee-reported use. It does not establish how many paid subscriptions a firm has or what those subscriptions cost.
Source: Gallup, Organizational AI Adoption Jumps Six Points, July 20, 2026. Reviewed October 28, 2026.
Review question: Do the people responsible for the subscription inventory know which tools staff use for work, including tools appearing on expense claims?
Use dated price evidence
Vertice recorded a 16.4% SaaS inflation rate in June 2026. Its later update reported 14.3% from early September data. The June observation should not be presented as the latest rate or as a 16.4% month-over-month price increase.
Source: Vertice, SaaS inflation rate, June to early September 2026 observations. Reviewed October 23, 2026.
Review question: Have you compared the actual renewal quote, included features and notice period with the existing agreement? For a vendor-specific example, read the Copilot Business usage-billing explanation.
Turn the questions into your own evidence
A U.S. accounting, consulting, legal, agency, recruiting or advisory firm needs its own records to decide what to keep, resize, test or retire. Start with a full billing cycle, normally 12 months of relevant statements and expense reports, then reconcile the list with current contracts and the people using each tool.
- Record the vendor, plan and billing model.
- Check paid seats against current users.
- Record renewal dates and cancellation notice periods.
- Assign one person to own each review.
- Record approved usage limits and confirm the available controls in the vendor's system.
- Write the decision, approver and next review date.
A percentage from another organization's survey cannot replace this work. A review may confirm that existing subscriptions are appropriate; it may also identify questions to resolve. AccruePilot makes no savings or return-on-investment promise.
How should a small firm use these statistics?
Can we use a survey percentage as our savings target?
No. The studies cover different organizations, datasets and periods. A percentage from one of them does not establish your firm's unused spend or potential savings. Use your own invoices, agreements and usage evidence to make the decision.
Are these AccruePilot customer results?
No. These are findings from the named third-party sources. AccruePilot's review questions are practical suggestions, not study results or evidence of outcomes from using the toolkit.
Do these studies establish an AI budget per employee for a small firm?
No. The studies reviewed here do not establish a budget benchmark for professional-service firms of 10 to 250 people. Build a scenario from your paid seats, usage, required base subscriptions and actual commitments.
Does Flexera's 59% figure mean 59% of AI spending is wasted?
No. It is the share of respondents who said wasted AI spend increased over the previous year. It does not measure the proportion of AI spending wasted.
Does an employee using AI mean the firm pays for an AI seat?
No. Gallup measures reported use at work, which can involve paid, free or bundled access. Confirm the payment source and subscription terms from your own records.
Does an unused SaaS license mean wasted AI spending?
Zylo's 36% figure concerns unused SaaS licenses measured against industry-recommended utilization levels. It is not an AI-only spending measure, a percentage of money recoverable, or a finding about your firm.
Does AI-native spending growth forecast our firm's AI budget?
No. Zylo's 108% figure describes year-over-year AI-native application spending growth in its dataset. It does not forecast the budget of a small professional-service firm. Build your budget from your own paid seats, usage, required software and agreement.
Start with the free checklist
The AI Renewal Checklist is a free, no-signup procedural checklist. Work through its steps using your own records. It does not collect a subscription inventory or analyze your accounts.
Use the free AI Renewal Checklist Read the AI subscription audit guide
Use the free renewal and decision templates to keep the evidence, owner and next action together.
The AI Spend Control System, Founding Edition provides a local browser Control Center, Excel workbook and 12-page field guide for recording a manual review. It costs US$59 once, with no subscription, for internal use by one purchasing U.S. organization. Resale, redistribution or repeatable client-service delivery requires a separate written commercial license.
The four review choices, approver and notice deadline belong in your own records or the free companion templates. They are not dedicated fields in the paid Control Center or workbook.
Source checks are dated beside each citation. This page was updated on October 29, 2026. Source figures are attributed to their publishers and do not represent AccruePilot customer results. For a correction, contact accruepilot@proton.me.