"Garbage In, Garbage Out" Is a Lame Excuse

If we start with bad inputs, we’ll end up with bad outputs. Consider these scenarios:

• The data pipelines keep failing, so nobody uses the dashboard.
• The decision metric is broken, so our experiment results are unreliable.
• The source data was flawed, so we can’t trust the analyst’s recommendations.

You get the point.

When people use the phrase "garbage in, garbage out" in the context of data, it often sparks a negative, even fatalistic reaction. It’s treated as the final word on the matter - a full stop in the discussion. It’s a shrug, or a resignation. C’est la vie. Shhit happens.

But that’s a lame excuse. Don’t fall into that trap. I want to challenge you to think differently. “Garbage in, garbage out” shouldn’t be the punchline at the end of the conversation - it should signal the beginning.

The Origins

The first published uses of the phrase “garbage in, garbage out” were observed in the 1950s. It entered the Oxford English Dictionary as an acronym, GIGO, in 1993. It has its roots in early computing, when developers realized that flawed inputs led to flawed outputs, regardless of the sophistication of the computations. As computers were used to solve complex mathematical problems, the importance of accurate data became evident - leading to GIGO being a cautionary tale for all those who work with data.?

We used the phrase in the past, we use it now, and it will keep its place in our vocabulary for years to come. As we move into an era of AI, GIGO is more relevant than ever. Business leaders are sensitive to the fact that data inconsistencies can cause major errors in AI-driven processes. As data teams’ roles are expanding to cover both data and AI, I suspect that the “garbage in, garbage out” refrain will become more frequent and acute. Buckle up.

Cake: An Analogy?

Imagine I bake you a chocolate cake. I start with a proven recipe that’s been passed down from my mother. But this time, my ingredients are less than ideal - my eggs are rotten, my flour is infested with beetles, and I mistakenly use carob instead of chocolate. Even if I frost it nice and serve it to you on a fancy plate, there’s no disguising the fact that the cake is ruined. You won’t enjoy it, and it may leave you wondering whether you want to eat my cooking again.

It’s the same with data. When a data team starts with flawed inputs like inaccurate data or incomplete information, the end product will fail to deliver, no matter how much effort goes into the analysis or presentation. As a result, business stakeholders lose trust in the insights provided, which in turn erodes the perceived value of the data team’s work. This lack of trust can snowball, leading to reduced influence within the organization, missed opportunities for strategic decision-making, and ultimately, a diminished role for data in driving business outcomes. Like a bad cake, even the most sophisticated models and dashboards can’t hide the flaws of poor data.

How it Comes Up

Do you hear “garbage in, garbage out” in your business? Pay attention to who’s saying it and what they really mean.

Within the data team: Imagine a data scientist saying, "Garbage in, garbage out." What they might really mean is, "I’m paid to do the fancy math, but if the data I’m given is flawed, it limits the quality of my work. This is frustrating, and it’s not something I can fix on my own. Someone else needs to address this." It’s a sign of exasperation, but it can also come across as a form of blame-shifting. I get it. Poor data quality disrupts your work, but a mindset of simply passing the buck won’t lead to meaningful improvement. Don’t play the victim.

Within business leadership: Now consider when a marketing VP receives an analysis that’s clearly off the mark and says, "Garbage in, garbage out." What they really mean is, "Data team, this isn’t working. How can I trust or value what you give me when half the time what you give me is crap? I need reliable insights to make informed decisions. Go fix the issues, and don’t let this happen again. But I won’t hold my breath." It’s a harsh rebuke, reflecting frustration with ongoing data quality issues that undermine the value of the data team’s work. Hey, marketing VP, we’re doing the best we can over here.

Among data producers: And then there are the people who create or manage the source of the “beetle-infested flour.” They are likely silent. They might not even know about the disturbance. If they do know, they might not fully grasp the downstream impact of the flaws. The disconnect between data producers and data consumers is a key part of the problem.

What Can We Do About It

Within the data team, there are several actions we can take to fight GIGO. We can conduct code reviews, run rigorous validation checks, manage data quality proactively, set up alerts on critical data sources, and increase bandwidth devoted to data cleaning and preprocessing. Some of these actions are easier than others, but most fall under the umbrella of operational excellence. The real challenge is transforming “best intentions” into ingrained, robust processes. We also need to consider solutions that go beyond the basics, such as rearchitecting systems to be more resilient. Maybe our cake should be gingerbread, not chocolate - or perhaps we don’t even need to serve dessert at all.

For business leadership, the first step is to give the data team the benefit of the doubt. Recognize that GIGO isn’t just a data team problem; it’s often symptomatic of a larger issue within the organization. The data team is part of the solution, but we’re not the sole owners. Help by bringing other leaders into the conversation, and give us funding so we can tackle the problem effectively.

As for data producers, it’s time to own up to it. Make the effort to understand why data quality matters and how it impacts the business. Proactively collaborate with your data consumers to address GIGO issues. It’s not just the right thing to do - it’s critical for the success of the business.

What’s the BEST we can do about it?

The best we can do is foster collaboration and accountability across all parties involved:

• “Yes, chef, I used rotten eggs. Here’s how I’m fixing it.”
• “This cake tastes awful. How can I help?”
• “The beetles got into the flour again. I understand why we need the cake. Let’s solve this problem together.”

Next time you hear someone say, “Welp, garbage in, garbage out,” don’t let those words echo as a lame excuse. Instead, turn it into the beginning of a meaningful conversation about how to improve the situation.

See Also

• 3 Key Strategies for Improving Data Quality
• Data Quality Is a Mess and It's All Your Fault
• Code Reviews: An Uncommon Practice in Data Analysis?