The CFO Question That AI Has Finally Answered

The question every CFO asks — “What’s the actual return?” — has been difficult to answer cleanly for most of the last three years. Until recently, the honest answer was: it depends on what you’re measuring and how optimistic you’re willing to be.

That changed in Q3 2024. And one freight company nobody talks about made the case better than any slide deck I’ve seen.

The turning point most people missed

Gavin Baker, Managing Partner at Atreides Management and one of the more rigorous AI investors I follow, flagged something important about Q3 2024 earnings season. For the first time, Fortune 500 companies outside of big tech began reporting specific, quantified productivity gains from AI. Not potential. Not projections. Audited numbers that moved stock prices.

The example Baker pointed to was C.H. Robinson.

If you don’t know them: they’re one of the world’s largest freight brokers. Their core business is matching shippers with carriers — connecting a truck hauling goods from Chicago to Denver with a return load so it’s not running empty. At the center of that business is pricing: responding quickly to inbound quote requests with accurate rates and availability.

Before AI, that process took a team member 15 to 45 minutes per request. And they could only respond to 60% of inbound quotes — the rest were left unanswered, which in freight brokerage means leaving money on the table directly.

With AI, they now respond to 100% of requests. In seconds.

Their stock rose approximately 20% following that earnings report. Not because of a product launch or an acquisition — because their cost structure shifted and their revenue capture improved simultaneously, in the same quarter, in a measurable way.

What makes this a clean signal

The reason C.H. Robinson is such a useful example isn’t just the magnitude of the gain. It’s the structure of the task.

AI performs best when the outcome is verifiable. Did we win the quote? Did the rate clear? Did the shipment get booked? These are binary questions with no ambiguity. The same logic applies to customer support ticket resolution, financial reconciliation, document review, compliance checks — any domain where there’s a clear right answer that can be assessed after the fact.

This is why freight brokerage turned out to be an early showcase. The feedback loop is tight. You quote, you win or lose, you learn, you improve. AI has been operating in that loop at C.H. Robinson, processing over a million quotes and a million orders through automated systems. The productivity compounding is real and trackable.

Contrast this with more open-ended applications of AI — summarizing documents, generating first drafts, answering questions — where the quality gain is genuine but harder to tie directly to revenue. Both matter. But CFOs can point to C.H. Robinson and say: this is what it looks like when AI directly hits the top and bottom line.

What investors are actually seeing

The VC community has been watching a cohort of companies that deployed AI aggressively 18 to 24 months ago. The pattern emerging is consistent: similar or greater revenue being produced with meaningfully fewer people than the same companies had two years prior. Sales, support, and parts of product development are all categories where the headcount-to-output ratio has shifted.

More telling still: among the largest public spenders on AI infrastructure, Return on Invested Capital (ROIC) has been rising since they began scaling their AI spend — not falling, as critics of the capital buildout would expect. Part of this is efficiency on existing systems. But a significant portion reflects genuine output gains that are now showing up in earnings.

This is not the speculative case for AI. It’s audited financial statements.

The gap between knowing and doing

The challenge I see most often when I work with organizations is not skepticism about whether AI can deliver ROI — most executives accept that it can, at this point. The challenge is translating that general belief into something specific enough to act on.

C.H. Robinson didn’t deploy “AI.” They identified the exact process that was creating the most friction in their business — quoting speed and coverage — and rebuilt it around AI-assisted automation. The specificity was the point. A generic AI rollout across the same organization would likely have produced a fraction of the result. It’s the same reason off-the-shelf deployments stall, a point I’ve made in Plug-and-Play AI Is a Myth.

This is the consistent finding when I look at companies that have generated measurable return: they started with a single, well-defined workflow where the outcome was clear and the current performance had a quantifiable ceiling. They didn’t try to transform everything at once. They picked the highest-friction bottleneck, proved the return, then expanded.

Freight brokerage in 2024 had a natural starting point — quoting — where the constraint was purely human processing time. The question for any business is: where is your equivalent?

In financial services, it tends to be client-facing document preparation or compliance workflows. In healthcare administration, it’s often prior authorization or patient intake processing. In professional services, it’s research and first-draft generation at the analyst level. The categories differ, but the logic is the same: find where speed and throughput are the binding constraint, and where the outcome is verifiable.

That’s where the ROI lives. It’s not in the future. C.H. Robinson proved it in Q3 2024.

We’re Exponential Partners. We help executives and their teams identify where AI creates measurable value in their specific business — and build the systems to capture it. If you want an honest conversation about where the return actually is for your organization, reach out.

References

1. C.H. Robinson Q3 2024 Earnings Call, October 2024.
2. Gavin Baker / Atreides Management AI ROI commentary, 2024.
3. FreightWaves — C.H. Robinson delivers on AI. freightwaves.com