What is the Deloitte Enterprise AI Navigator?

The Deloitte Enterprise AI Navigator is a platform Deloitte launched in March 2026 to help large enterprises measure and improve AI ROI. It includes an AI opportunity scanner, business impact evaluator, process redesign module, and agent creation hub. Deloitte built it in response to a finding in its own 2026 State of AI in the Enterprise report: 79% of large companies have deployed AI, but only 36% report measurable business impact. The Navigator is designed to close that gap at the enterprise level.

Why would this matter for a small firm if Deloitte's tool is for big companies?

Because the 79%/36% gap isn't limited to enterprise. Small professional services firms are facing the identical problem: they've signed up for ChatGPT, Copilot, or Claude, used them a few times, and can't point to a specific dollar figure or hour savings that justifies the cost or the mental overhead. Deloitte's tool addresses this systematically for enterprises; small firms can apply the same logic with a lightweight three-question framework adapted for their scale.

How do I know if my firm's AI is actually paying off?

Ask three questions: (1) Which of our AI tools is saving the most team time, and can we quantify it in hours per week? (2) Does that time savings translate into more client capacity, faster delivery, or reduced overtime — or does it just disappear? (3) If we removed that tool tomorrow, what would we have to hire for or stop doing? If you can't answer all three, you don't yet have a clear ROI picture. The tool saving you the most time that you could least afford to lose is your highest-value AI.

What types of professional services firm AI use cases have the clearest ROI?

The AI use cases with the most quantifiable ROI for small professional services firms are: (1) meeting summarization — measurable in minutes saved per call and accuracy of captured action items; (2) document drafting — measurable in first-draft hours and revision cycles; (3) billing capture — measurable in recovered revenue (untracked hours × billable rate); (4) client communications — measurable in response time and follow-up completion rate. Use cases with fuzzier ROI include 'research assistance' and 'brainstorming' — valuable, but harder to quantify.

What should a firm do if its AI tools aren't producing measurable returns?

Stop buying and start focusing. Most small professional services firms that aren't seeing AI ROI have one of three problems: (1) they're using AI for tasks where the ROI is genuinely small (AI for brainstorming doesn't save much for a 5-person firm); (2) they haven't defined a specific workflow — they're using AI ad hoc without a repeatable process; or (3) they're using the right tool for the wrong task. The fix is to pick one specific, high-volume task, define the AI-assisted version step by step, and measure it for four weeks before adding anything else.

AI ROI for Small Firms: A Practical Audit

Published: March 14, 2026 | By: The Crossing Report | 6 min read

Summary

Deloitte launched a new platform in March 2026 to help its largest enterprise clients figure out if their AI spending is actually paying off. The reason: Deloitte's own research found that 79% of large companies have deployed AI, but only 36% report measurable business impact. That gap isn't an enterprise problem — it's the same gap most small professional services firm owners are sitting in right now. Here's a lightweight version of Deloitte's framework that a 5-15 person firm can run in an afternoon.

The Problem Deloitte Is Trying to Solve (for Their Clients)

In March 2026, Deloitte launched the Enterprise AI Navigator — a platform with four modules: an AI opportunity scanner, a business impact evaluator, a process redesign module, and an agent creation hub. The tool is explicitly designed to help large companies move AI "from cost to value."

The reason Deloitte built it: their own 2026 State of AI in the Enterprise report found that while 79% of large companies have deployed AI, only 36% report measurable business impact.

That number is worth sitting with. These aren't small firms experimenting with ChatGPT. These are companies with dedicated AI teams, significant budgets, and access to Deloitte consulting. And more than six in ten of them can't point to a clear return.

For small professional services firm owners — accounting, law, consulting, staffing, marketing agencies — this is either reassuring or alarming, depending on how you look at it.

Reassuring: You are not uniquely failing at AI ROI. The largest, best-resourced companies in the world are struggling with the same measurement problem.

Alarming: If enterprise companies can't measure their AI returns with dedicated resources, what are the odds a 10-person firm is doing it right without any?

Why the Measurement Problem Is Real (and Not Your Fault)

Most AI tools in professional services save time in ways that are hard to capture. When Fathom summarizes your client call and you spend 10 fewer minutes writing notes, that 10 minutes doesn't automatically become 10 minutes of billable time. It becomes 10 minutes of thinking time, or checking email, or moving slightly faster into the next task.

The time savings are real. The revenue impact is diffuse.

This is the core measurement problem. AI tends to return value in a form that's distributed across many small efficiencies rather than concentrated in a single measurable outcome. And distributed efficiency is hard to see — until something breaks or you remove the tool.

The 36% of companies that do report measurable impact tend to share one characteristic: they attached AI to a specific, high-volume task where before-and-after comparison was possible. Transaction coding with 90% accuracy vs. 70% manual accuracy. Contract review time of 2 hours per document vs. 6 hours. Monthly close in 3 days vs. 8 days.

The firms that can't measure their returns are typically using AI broadly and inconsistently — a little drafting here, some research there — without ever connecting a specific AI use case to a specific metric.

The 3-Question Value Audit (Small-Firm Version)

You don't need Deloitte's four-module platform. You need three questions, answered honestly, about each AI tool your firm is currently paying for.

Question 1: Which tool is saving the most time — and how much?

Name the tool. Name the task. Estimate the hours.

Examples:

"Fathom saves my team about 15 minutes per client call on note-taking. We have about 8 calls a week. That's 2 hours."
"Claude saves me about 45 minutes on the first draft of client proposals. I write about 4 proposals a month. That's 3 hours."
"Copilot saves me about 20 minutes per engagement letter. We do 6 per week. That's 2 hours."

If you can't name a specific tool, a specific task, and a rough time estimate, that tool is not producing measurable ROI. It might still be worth keeping — but you can't justify it beyond "it seems useful."

Question 2: Does the saved time go somewhere useful?

Time savings only produce ROI when they translate into something your firm can measure: more client capacity, faster delivery, reduced overtime, or lower staffing cost.

Ask: When we save that time, what actually happens to it?

If the 2 hours Fathom saves goes to preparing better for the next call → that has value, even if it's hard to measure directly.
If the 2 hours disappears into Slack and email → the tool is saving time, but the firm isn't capturing it.
If the 3 hours saved on proposals lets your team take on one more client per month → the ROI is immediate and clear.

The highest-return AI implementations are always the ones where someone decided in advance what the saved time would be used for.

Question 3: What would you have to hire for (or stop doing) if you removed this tool tomorrow?

This is the clearest test of real value.

If removing Fathom means someone has to spend an hour a day writing meeting notes again → Fathom is worth the cost of whoever's time that is.

If removing Copilot means you go back to spending 90 minutes on engagement letters instead of 30 → Copilot is worth 60 minutes of your hourly rate per letter.

If removing Claude from your research process means you'd have to hire a research assistant for 10 hours a week → Claude is worth at least 10 hours of an assistant's salary per week.

Tools that fail this test — where you couldn't tell a meaningful difference if they disappeared — are the ones that aren't producing ROI, even if they feel useful.

The Action This Week

Set a 20-minute timer. List every AI tool your firm is currently using or paying for. Run each one through the three questions above.

High value: Tool addresses a specific task, saves quantifiable time, and the saved time goes somewhere useful. Keep it.

Unclear value: You use it, but can't estimate time saved or can't say where the time goes. Set a 30-day experiment — define a specific use case, track it, and revisit.

Low value: You have access to it but your team doesn't use it regularly, or you can't identify what problem it solves. Cancel or pause.

The goal isn't to use more AI or less AI. It's to use the AI that actually does something for your firm — the way 36% of companies have figured out how to do — instead of staying in the 64% that pays for it without measuring it.

Even Deloitte's Biggest Clients Can't Figure Out If AI Is Paying Off — Here's How Small Firms Can