Why Your Biggest Risk Isn't What You're Measuring

Tal Lev-Ami is co-founder and CTO of image and video platform Cloudinary, which is trusted by more than 11,000 brands and 3 million users.

getty

​The late Andy Grove, former CEO of Intel, famously said “only the paranoid survive.” In his seminal book on leadership with that same title, he urged readers to be vigilant for what he called SIPs (Strategic Inflection Points), when massive change occurs and the rules of business shift “fast, furiously and forever.” We're living through an era where SIPs arrive relentlessly—and yet, paradoxically, the risks that most often derail companies aren't the ones anyone sees coming. It’s no wonder then that risk management has wended its way to the top of the CTO agenda.​

In twelve years of building a high-growth SaaS company alongside visual media infrastructure for enterprise companies, I see companies putting ever more resources into reacting to those big, visible and measurable SIPs. Yet I've watched the same pattern repeat: it's not the big, visible risks that derail companies—it's the small inefficiencies no one thinks are worth managing, slowly compounding into multi-million dollar crises.​

Ain’t Broke, Still Fix It

The bane of risk managers everywhere, "the prevention paradox," causes these small, gradual inefficiencies to snowball into full-fledged crises. When risks are prevented, by definition, nothing happens, so the outcome is neither visible nor measurable. So ironically, successful risk management tactics often don’t get reinvested in, especially when margin pressures are tight.​

Small Glitches, Big Consequences​

As a CTO, I see the consequences of this paradox in action everywhere. A really widespread problem is companies overlooking accumulated inefficiencies in their cloud storage. That’s why a whole cloud optimization industry has cropped up to help companies recover millions in efficiency leaks. However, until you discover that those inefficiencies exist, you have no idea you’re actually haemorrhaging money. ​

Another IT pitfall I’ve seen is confusing build-time and runtime automation when using AI agents. While an agent might perform well 99% of the time, a single, unexpected catastrophic failure can be disastrous. That’s why when implementing automation at scale, it’s best to use agents robustly to build, and then simplify the execution phase as much as possible to ensure reliability.​

The Biggest Risk: Opportunity Cost​

But these are only likely to be the tip of your iceberg. Let me share a massive risk I’m seeing with global ecommerce retailers: opportunity cost. There is so much revenue they potentially miss out on due to small, unaddressed technical issues. And these issues can live in totally unexpected places, including in relation to how they manage images and video online. ​

One of the most crippling margin-eroders for online retailers is the cost of returns. It’s not just the cost of the lost sale itself, but the additional shipping costs, the restocking, acquisition costs and recycling. US retailers are projected to see $849.9 billion in returns in 2025 according to NRF, which represents 15.8% of their annual sales. ​

Aside from the 9% of "bad actors" who intentionally buy to return, a key reason for returns is that items simply fail to meet a customer's expectations. This commonly results from inaccurate color representation, poor fit or discrepancies between the physical item and its depiction on the website. Many of these problems can be prevented by simply addressing image and video quality issues. ​

If you still aren’t convinced, consider this: according to Capital One Shopping Research from Jan 2026, online return rates average 24.5%, compared to 8.72% for Brick and Mortar—nearly 3x as high.​

However, well before a shopper gets to this stage, technical glitches can deter them from completing a purchase at all. Issues like slow image rendering on mobile devices or videos failing to auto-crop correctly—such as losing the subject when switching between portrait and landscape views—are often perceived as minor. However, these seemingly minor technical shortcomings can lead shoppers to form a negative impression of the brand's overall quality. This is a classic example of hard-to-measure but potentially crippling opportunity cost.​

Another Silent Margin Killer​

Going further down this rabbit hole, you have the problem of SEO penalty compounding. Poor image optimization means slower page loads, which affects search rankings, which reduces organic traffic, which increases dependency on paid ads, which raises acquisition costs. The connection between "images aren't optimized" and "CAC went from $9 to $29" is nearly impossible to trace. ​

As AI-powered search summaries and shopping assistants become more prevalent, they're increasingly pulling from visual data to make recommendations. If your product images are poorly tagged, inconsistently formatted or lack proper metadata, you may be invisible to AI-powered shopping tools that are starting to mediate purchase decisions.​

So yes, be paranoid. Grove was right about that.​

But channel your paranoia wisely. The CTOs who thrive aren't just the ones monitoring the big visible risks everyone sees coming. They're the ones who've built systems to catch the small ones before they compound into crises no one saw coming until it was too late.​

The risks that will derail your company probably aren't on your dashboard. They're accumulating quietly, right now, in the places no one thinks are worth watching.​

Start watching.​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?