How To Fix the $406-Million Data Problem

The promise of AI is undeniable, but the road to success is rarely straightforward. Too often, companies rush to invest in shiny new tools, only to discover that the real stumbling block isn’t the algorithms or the tech stack. It’s the data. The paradox of the digital age is that we’ve never had more information at the ready, yet most businesses are bogged down by gaps, silos, and messy pipelines. Instead of fueling innovation, data problems quietly drain millions in lost opportunities, stalled projects, and bad decisions.

In this issue, we’ll look at how to turn that mess into momentum. You’ll learn how to fix costly data issues without blowing your budget, and the five biggest mistakes to avoid when rolling out AI agents. We’ll also explore why building trustworthy AI depends on more than raw data. And to bring it all to life, we’ll take you behind the scenes of Formula 1, where teams transformed data overload into a winning fan experience.

On a budget? You can still fix the $406-million data problem

Let's talk about the silent killer in your business: bad data. It’s not just an inconvenience, it's a massive financial drain, especially when it's fueling your AI. How big of a drain? One survey found that AI trained on low-quality data costs large companies an average of $406 million per year, or 6% of their revenue.

So, why aren't more companies fixing the problem? Because data cleaning can be just as expensive as the problem itself. But it doesn't have to be. We've compiled a list of five cost-effective ways to clean your data, so you can stop wasting money and start trusting your AI.

We’ve done the math: Here are the 5 most common AI agent pitfalls 

Launching a few AI agents doesn’t make you an agentic enterprise. In fact, treating AI as a series of point solutions misses the bigger picture. The point of AI agents is to improve productivity, customer satisfaction, and retention, and create new revenue streams. But too many orgs rush to build agents without a clear purpose and a defined business value. Based on more than 80 AI agent deployments, we’ve identified the five most common pitfalls, and how you can sidestep them.  

The data is there, but the context is not: Enter metadata

You have more data than ever, and that’s not necessarily a good thing. Data is the raw building material for AI, but metadata is the blueprint. Data provides the facts; metadata is the crucial context that tells AI what the data means (is that string of numbers a phone number or contract value?), where it came from, and how it connects to other information. This is the key to building the best AI. Here are the critical differences between the two data types, and tips on how to use them together to turn raw data into a strategic advantage.

How F1 is using data to tune up the ultimate digital fan experience 

Less than 1% of Formula 1’s massive fanbase will ever attend a race in person, which means F1 needed to turn its digital viewership into a loyal, personalized following. The problem was 100 siloed data sources that made it nearly impossible to see a fan as an individual, not just a row in a spreadsheet. So, F1 tackled that data chaos head-on by bringing all its fan data together. This has not only made the fan experience more consistent and personal, but it’s given F1’s commercial team the data it needs to attract sponsors and partners, and build loyalty across every turn of the fan journey.

This newsletter was curated by Lisa DiCarlo Lee, Contributing Editor, Salesforce.