- Huntington Beach, CA, July 2025Before I met Serge, I’d never tried even one sip of coffee in my entire life. Then Serge (the best data engineer I know) blackmailed me, saying he’d only pick me up from the airport if we stopped at his favorite coffee spot for a cup.
To my surprise, I loved it. I loved it so much that I quickly became a coffee snob. I explored all the different ways to brew coffee until I finally found my beloved Technivorm Mocha Master, a handmade $400 machine from the Netherlands that delivers a perfect cup every time.
Then Serge showed me a $34 plastic device that, according to him, makes the best cup of coffee ever.
As cool as my Technivorm is, it’s a good example of a terrible IT condition: Shiny New Toy Syndrome. IT people catch this illness when they create a technical solution that nobody asked for, something I call “solutions looking for problems.”
Because of AI, the syndrome is more common than ever. Author Kavita Ganesan tells the story of a law firmthat asked her to help implement a new AI solution. They wanted to add standard legal language to contracts automatically, but she advised them to find a data engineer to improve the process instead of using a large language model (LLM). She writes:
“Even though I saved the attorney a year’s worth of fees from hiring me, he was upset. It turns out, [he] wanted to market his tool as an AI product. I unknowingly put a dent in his plan.”
That attorney wanted to market a Shiny New Toy to his clients, even though less exotic options were available. He went straight to his favorite solution instead of first defining the business problem.
Use Cases in Disguise
As a consultant, I constantly remind myself to avoid suggesting a solution for a problem before preparing a good, detailed problem statement. But people often think they’re writing a problem statement when they’re really just describing their favorite “use case,” making it easy to forget all the problems they should acknowledge before picking a solution.
For example, you might hear someone say, “The database needs an upgrade,” without stating the real problem: the data processing jobs aren’t finishing on time. There could be many more solutions to that problem than a software upgrade.
Here are some helpful tips for writing a good problem statement:
State what’s not working right now, and link that issue to its root cause.
- State the history of the problem and help people understand its significance by quantifying the impact. Measuring the problem (i.e., how often it occurs, its impact on cycle time, etc.) always helps.
- Ask “why” five times every time someone states a problem. That helps pinpoint the root cause.
- Don’t say anything about the solution. That comes later.
That helps everyone agree on the right tool to use when deciding on the solution.
Avoiding the Trap
Serge didn’t suffer from Shiny New Toy Syndrome. When he joined my company, he focused on improving the accuracy of the company’s revenue forecast for our small, high-volume customers. That was a very specific, age-old problem that everyone was familiar with. Serge collaborated with our brilliant data scientists to train a machine learning model using data from our data warehouse, which produced much more accurate forecasts. We combined it with the human-forecasted data for the larger customers to create a consolidated demand picture.
Ironically, my company's smartest, most innovative data engineers used simpler machine learning techniques instead of a more complex AI solution. They used the best tool to fix the data we needed the most help with.
He also found great, inexpensive coffee-making solutions because he started with the problem: finding better coffee to drink.
Buy now on Amazon