How to Stop Automating Your Way Into Bad Business Decisions

I recently worked with a founder who said his marketing was “completely automated.” AI wrote the copy, scheduled the posts and optimized the budget. He was thrilled until his “successful” campaign drove zero qualified leads.

Sound familiar? Here’s what happened: He used SEO tools to find trending keywords, then fed them into a generative AI to produce content. The problem? He focused on what competitors did, instead of what his customers cared about. Great sounding content, wrong audience.

Today, our decisions are increasingly shaped by machine-generated information that’s divorced from reality. The hardest part of decision-making isn’t gathering data. It’s knowing which data to trust.

Related: How to Use Automation (and Avoid the Pitfalls) as an Entrepreneur

The self-referential internet problem

Every algorithm learns from history, but what happens when that’s just repurposed ideas? Google’s AI overviews and featured snippets sit above everything else, determining what we see. Meanwhile, content farms publish AI-generated articles optimized to feed that same algorithm. The result is a self-referential internet where biases compound.

I learned this the hard way. After selling my first ecommerce business in 2004, I spent two decades building marketing systems for startups and small businesses. Back then, we worried about data scarcity. Now? I’m cleaning up messes created by data pollution.

Often, automated sentiment tools start to misread nuance because their language models ingest AI-written text that lacks authentic human tone. The result is synthetic insights, and consequently, bad business decisions.

2 traps smart founders fall into

You’ve likely heard of psychological biases like confirmation or anchoring bias. Here’s a modern rendition:

1. Algorithmic authority bias

When an AI or search engine makes a recommendation, we instinctively assume it’s correct. But Google doesn’t rely on accuracy alone. The algorithm checks for Experience, Expertise, Authoritativeness and Trustworthiness, or EEAT, which may have imperfect components. Don’t treat AI content as truth just because it looks good. Validate output against reputable sources.

2. Synthetic confirmation bias

Chatbots make it dangerously easy to confirm what you already believe. Ask an AI, “Why is my product perfect for millennials?” It’ll generate supportive reasons based on its analysis of published content that supports your idea, even if those opinions are wrong.

You’ve just created what behavioral economists call a reinforcement loop. It rewards overconfidence instead of reality-testing. Research published in Nature reveals that human-AI feedback loops amplify biases significantly more than human-to-human interactions, and we’re blind to it.

Related: The Top Fears and Dangers of Generative AI — and What to Do About Them

The bias firewall: 3 steps to sharper decisions

Try this three-step bias filter to avoid automating your way into bad decisions.

Step 1: Diagnose the data source

Before trusting a metric, ask: Where did this data originate? Was it collected from real customers, scraped from the web or generated with AI? A few minutes of checking URLs and authorship can significantly improve data quality. Ask “Where did this number come from?” If the answer is “I don’t know,” then you haven’t done your job.

Step 2: Triangulate the truth

Compare at least two independent data sources or tools before making a decision. If they disagree, dig deeper. If they align, your confidence increases. This is how researchers reduce error through validation. Many founders skip this step because one dashboard feels like enough. It’s not.

Step 3: Run a sanity-check simulation

You don’t need fancy software to stress-test a decision. A spreadsheet with best- and worst-case scenarios can suffice.

With one recent client, this simple test showed that a traffic surge turned out to be bot traffic. Filtering the bad data saved them thousands in ad spend.

Each of these steps forces what psychologist Daniel Kahneman calls slow thinking. Try this deliberate, rational process to counteract your tendency to trust fast, automatic judgments.

From individual thinking to team culture

Technology may introduce bias, but leadership perpetuates it. The antidote is cultural, and it starts with how your team talks about data.

Encourage respectful dissent: If everyone nods at the dashboard, no one’s thinking critically. Challenge people to ask, “What if this is wrong?”

Use pre-mortems: Before launching a campaign or product, ask the team to imagine it failed spectacularly. What went wrong? You’ll uncover hidden assumptions faster than any amount of data analysis. Frameworks like SCAMPER (Substitute, Combine, Adapt, Modify, Put to another use, Eliminate, Reverse) can help teams systematically challenge assumptions and explore alternative scenarios.

Make data storytelling a habit: Be able to explain how data was sourced and cleaned before sharing results, to expose the chain of assumptions behind every chart. Use visualizations and data storytelling best practices so everyone understands your data.

Over the last 20 years, I’ve learned that the best marketing depends not just on good data, but great stories. When your team can explain why the data matters and where it came from, you’ve built a bias-resistant culture.

Next time you interview a candidate, try asking, “Tell me about a time data told you one thing, but your instinct said another.”

The answer reveals their level of critical thinking.

The new information pollution

A decade ago, the challenge was data scarcity. Today, it’s data pollution.

Bad data alongside AI-generated articles and reviews can confuse insight with noise. Even genuine analytics can be skewed by contaminated input data or opaque model logic. For founders, this means we can’t outsource discernment. Where tools crunch numbers, humans question meaning.

That’s why ongoing curiosity matters. AI models are only as ethical and accurate as the people guiding them. Technical skills are valuable, but critical thinking about data quality is priceless.

Related: The Big Risks You Need to Avoid When Using Marketing Automation

The competitive edge of clear thinking

Automation will continue to improve. So will synthetic content. But here’s what won’t change: the competitive advantage of founders who know when to pause and ask, “Is this real?”

The founders who win aren’t the ones with the flashiest AI tools. Instead, they combine machine precision with human skepticism.

Your move: Audit one major decision this week. Trace the data source, test the assumption and decide consciously. If you catch yourself blindly trusting a dashboard, good. That’s the moment you become a better entrepreneur.

I recently worked with a founder who said his marketing was “completely automated.” AI wrote the copy, scheduled the posts and optimized the budget. He was thrilled until his “successful” campaign drove zero qualified leads.

Sound familiar? Here’s what happened: He used SEO tools to find trending keywords, then fed them into a generative AI to produce content. The problem? He focused on what competitors did, instead of what his customers cared about. Great sounding content, wrong audience.