#76 - Cut costs with AI - without losing your team or your CFO

A couple of weeks ago I wrote about the MRC 2026 report’s message for fraud teams: adopt AI, cut costs, and get more efficient.

Most fraud leaders read that and feel two things simultaneously - pressure, and dread.

Today I want to unpack why, and give you a framework for how to actually handle this conversation.

Why is this so hard?

Let's be honest about what this conversation actually involves.

You're being asked to build a business case that could end with your team being smaller. Possibly you included.

That's not a comfortable position. I wouldn't be comfortable in it either.

But the self-preservation piece isn't even the hardest part.

The harder problem is org-wide trust. AI hasn't proven itself in most organizations to the point where leadership feels safe rapidly losing institutional fraud knowledge without a credible backup plan. 

And that skepticism is fair. At least for now.

Then there's the third issue: even teams that genuinely want to move forward often don't know which tasks in fraud are actually automatable, versus which ones only look automatable from the outside.

Side note: If a vendor has ever told you their AI can “fully automate” your fraud review queue, ask them what happens to the edge cases. Then watch what they do next.

These three obstacles are real. Naming them isn't making excuses - it's knowing what you're actually up against before you try to solve it.

Where do you start?

Most teams skip straight to evaluating AI tools.

That's backwards.

The right starting point is boring, costs nothing (other than your time), and almost nobody does it: map your costs.

Break down every vendor by what it actually delivers. Then map every team member's time by task - even a rough breakdown works:

High-volume case review. Rule tuning. Vendor management. Escalation handling. Model monitoring. Complex investigations.

This is the only way to identify where efficiency gains are realistic and what the effort would actually involve. The answer isn't always AI - sometimes it's just a smarter process.

At this point you might be asking yourself: “I now know what needs to be automated with AI but do I build it myself or take a vendor?”

There's a heuristic I find useful for these build versus buy decisions: build if the underlying data powering the solution originates on your own system. Buy if it originates with the vendor.

If you want to automate internal data work that is currently happening over excel sheets - build that.

If you want to automate an investigation process utilizing fraud signals and scores you get from a vendor? Buy that, preferably from the same one.

It's not a perfect rule, but it stops teams from building things they'll never fully own and buying things you can deliver yourself faster and cheaper.

How do you make the case?

Here's where most fraud leaders lose the room.

They walk into a CFO conversation talking about efficiency: “we'll save X analyst hours on case review.” The CFO does the math. It's a nice number, but it's capped.

Here's the thing about optimizing for efficiency: your ROI will always be limited by what you currently have. If you have a team of three FTEs, you can only “optimize” three FTEs.

The frame that actually lands is different - You're not trying to cut costs. Instead, you're trying to decouple OPEX from growth.

Right now, more transaction volume means more review hours, more API calls, more analysts. It scales linearly.

What you're offering isn't a smaller headcount line, but a better financial model. One where growth doesn't automatically mean proportional spend on fraud operations.

That's a different conversation. And CFOs respond to it differently.

But here's the thing:

The obvious pushback is still coming. “Great, so who gets let go?”

The honest answer is no one has to - but you need a real plan, not a vague promise about repurposing people to “higher value work.” That phrase has been in every AI pitch deck for three years. You need to name the tasks.

They fall into two buckets.

The first is work that requires deep business context AI can't replicate: optimizing risk policies for a new product line, evaluating vendor partnerships, designing fraud strategy when entering a new market, the cross-functional work with product and engineering that runs on judgment and institutional knowledge.

The second is technically demanding work that's harder to hand off than it looks: writing fraud rules from emerging attack patterns, calibrating ML score thresholds across customer segments, running incident root-cause analysis, designing the workflows that sit between automated outputs and final decisions. 

Side note: The teams I’ve seen make this shift successfully have one thing in common - they told their team what the plan was before announcing any tooling changes. And they gave them a real path forward, not just reassurance. I’ve written about why hiding it backfires.

If your team can shift toward those two areas, you have a real argument. One that your team can actually get behind, not just tolerate.

And if adoption goes well, there's a third path almost nobody talks about: passive downsizing through natural attrition.

Most fraud teams see single-digit annual turnover. If you model that rate, you can show your CFO a path to down-sizing headcount over two to three years.

Without layoffs, without losing critical knowledge overnight, and without betting everything on a rollout going perfectly from day one.

It's not glamorous. But it's credible. 

And credible beats optimistic in a budget meeting every time.

The bottom line

When your CEO asks you to adopt AI and “cut costs” (and we all know what that means) - they’re right. The lever is real.

But the lever doesn't pull itself.

The teams that will make real progress are the ones that start with the cost audit, not tool evaluation. That know how to speak to the CFO when making the case. And that bring their team along with an actual plan.

I hope that with this framework you now have a clearer sense of how to go about it.

Are you having this conversation at your organization? Hit reply - I’d genuinely like to know where teams are getting stuck.

In the meantime, that’s all for this week.

See you next Saturday.

P.S. If you feel like you're running out of time and need some expert advice with getting your fraud strategy on track, here's how I can help you:

Free Discovery Call - Unsure where to start or have a specific need? Schedule a 15-min call with me to assess if and how I can be of value.
​Schedule a Discovery Call Now »​

Consultation Call - Need expert advice on fraud? Meet with me for a 1-hour consultation call to gain the clarity you need. Guaranteed.
​Book a Consultation Call Now »​

Fraud Strategy Action Plan - Is your Fintech struggling with balancing fraud prevention and growth? Are you thinking about adding new fraud vendors or even offering your own fraud product? Sign up for this 2-week program to get your tailored, high-ROI fraud strategy action plan so that you know exactly what to do next. ​
Sign-up Now »

Enjoyed this and want to read more? Sign up to my newsletter to get fresh, practical insights weekly!

<