A few days ago, Seshu Motamarri from 3M asked a great question on LinkedIn:

It’s one of those questions that sounds simple until you realize how many hallway debates it’s sparked between operations, safety, and finance.

I jumped in with a short comment:

“One way to think about past incidents is to treat them as inputs, not savings. They help you model the Expected Annual Loss (EAL) = how often incidents happen × how costly they are. This should tell you what such an event really costs today. Then I model it again with automation. With automation, the failure rate drops: fewer incidents, lower severity, and more stable outcomes. So now, imo, delta isn’t a hypothetical ROI, I frame it as cash-flow stability, and that tends to land well with finance leaders.”

That short exchange kept playing in my head.
Because this is exactly where engineering logic meets financial logic and where we often lose the thread in automation conversations.

When I model this, I start by reminding myself that a warehouse is a system with people, equipment, and processes all interacting with each other.
Safety incidents are simply outputs of that system.
Automation? It’s just a new input that can shift those outputs over time.

Before we talk about improvement, we need to understand what the system actually costs us today.

I treat safety incidents like failures in an engineering model.
Each has:

So: Expected Annual Loss (EAL) = Frequency × Cost per incident

For example, a client averaged 19 recordable incidents a year over the last 36 months.
Each one costs about $29,400 fully loaded, including medical, investigation, legal, and the morale hit.

That puts their baseline EAL at roughly $559,000 per year.
It’s the check they don’t physically write every year, but the hit to the bottom line is very real.

Now imagine we introduce automation that removes high-risk manual touches - say palletizing robots - cutting out about 68% of the activities that cause these incidents.

That changes my model to:

EAL (new) = 6 × $22,100 = $133K / year

So we’ve dropped from $559K → $133K. A meaningful reduction in volatility.

This is where I’ve learned to slow down the conversation with finance.
It’s not about headline ROI or payback periods.

It’s about cash-flow stability. Fewer unpredictable losses, steadier throughput, and less time spent firefighting.

The delta of $426K per year becomes a risk-adjusted cash flow in the NPV model.
That adjustment alone shortens payback by about 0.8 years on a $4.2M CapEx.

And because finance teams think in variance, I show them a quick Monte Carlo view:

That framing usually lands.

Bonus: the same cell that removed the incidents also improved throughput by ~31%.
Double credit.

When we start treating safety improvements like reliability improvements, everything gets clearer.

You can trace the impact, from floor incidents to financial confidence, without overselling or guessing.

And that’s when automation stops being a ‘nice to have’ and starts becoming a stability tool.

How do you think about this in your world?
Do you fold risk reduction into ROI models or keep it in its own lane under safety?

I’d love to hear how others are bridging this gap between operations, safety, and finance.

If someone forwarded you this email, then you can sign up here to continue receiving the latest in the warehouse robotics space.