Real Data, not Ideal Data
Last week I introduced Growth Isn’t One Sided and the growth paradox: companies that focus only on efficiency stall, while those that focus only on innovation can’t scale. This week we’re starting Part 1: The Trouble With Scale. These short articles explore why traditional approaches to growth can backfire. Today’s chapter: why real data beats perfect models.
One Takeaway: Data can help your business growth or hurt it. Having lots of data doesn’t mean you have the right data. What matters is how well your data matches the real world.
The Data Trap
Metrics, dashboards, and analytics surround us every day. These tools can fuel growth by showing us patterns and tracking our decisions. But they can also get in the way of growth. Why? They give us the illusion that we have a clear picture of the world.
Here’s the truth: Just because we know a lot doesn’t mean we know everything.
Smart business leaders know which data helps, which doesn’t matter, and which actually hurts their decisions.
The Myth of Ideal Data
By “ideal data,” I mean what businesses pretend to have in their spreadsheets and decks: complete, accurate, timely information about everything that matters. By “real data,” I mean what we actually have: partial information with gaps, delays, and measurement errors. The difference matters.
Businesses often assume they have “ideal” data. Data that’s complete and flawless. It tells you which certainty markets to enter, which features to build, and what to optimize. Perfect data doesn’t exist.
Instead, we have real data. Real data is messy and incomplete. It has gaps. It doesn’t give us a perfect view of the world. It tells us some things about some markets, with confidence. It’s valuable, but has limits.
The problem isn’t that we have imperfect data (that’s inevitable). The problem starts when we act like our real data is ideal. When we build strategies that assume ideal conditions, those strategies break when they hit reality.
How to Use Real Data Right
Lasting growth comes from using your data in the right ways. Stop chasing or assuming perfection. Start embracing what’s practical.
Here are the key ideas:
Not All Data Is Equal: More data doesn’t automatically mean better decisions. You need to tell the difference between data that helps and data that distracts. Focus on what you can actually do something about.
Data Is Always Old: Your data captures what already happened, not what’s happening now. Without fresh insights, you risk building strategies for a world that no longer exists.
Numbers Need Stories: Spreadsheets show trends and patterns. But they can’t tell you why those trends matter. You need people close to the problems to add context that computers miss.
Quick Self-Check: Is Your Data Helping or Hurting?
Ask yourself these questions:
Relevance Check: Are you making important decisions based on data that’s 6+ months old? If your key numbers haven’t changed to reflect recent market shifts, you might be optimizing for yesterday’s reality.
Signal vs. Noise: Do you spend more time talking about what you can measure than what actually matters to customers? If your team meetings focus more on dashboard numbers than customer results, you have a measurement problem.
Decision Paralysis: Are you avoiding decisions because you’re waiting for “better” data? If you often hear “Let’s get more data before we decide,” you might be chasing perfect data instead of acting on real data.
Reality Check: When was the last time someone challenged your “best” metrics with real-world observations? If your front-line team never questions your data, you’re missing important gaps.
Real Examples from the Tech World
The gaps between data and reality show up everywhere:
Amazon: Jeff Bezos famously said, “When the data and the anecdotes disagree, the anecdotes are usually right.”1
Amazon once had metrics showing customer service calls were answered within 60 seconds. Yet customers kept complaining about long wait times.
To investigate, Bezos called the support line himself. He experienced a 10-minute wait. Their “trusted, gold-standard metrics” didn’t reflect reality. Not because the data was wrong, but because it was measuring the wrong thing
During my time at Uber and Lyft, I saw similar gaps:
Uber: In 2016, UberEats launched in Las Vegas but skipped certain dense suburban neighborhoods. Before expanding, the team looked at app opens to see where demand existed. The data suggested no one wanted delivery in those areas.
But these neighborhoods had tens of thousands of potential customers (myself included) with few restaurant options. People weren’t opening the app because no service was available there yet. Because there was no service, the data collected by the app wasn’t relevant to provide proof of demand.
Expanding required judgment and local knowledge, not just internal trackable data.
Lyft: Las Vegas has major events multiple times each week. At these events, rideshare apps can underestimate demand. This created bad customer experiences.
Here’s what happened: Groups of friends would leave together, each wanting their own ride. One person would check the app, see high prices, and tell the others. The friends wouldn’t bother checking their own apps. Why stop looking at Instagram if you knew what your Lyft app was going to tell you.
Lyft’s system saw one possible ride when there were actually three or four rides. This happened hundreds of times at once, creating less than ideal automated pricing decisions.
In these situations, local operators had the ability to make manual adjustments when they saw data the app couldn’t capture.
The lesson: When wrong (less than ideal) data signals happen thousands of times daily, small errors add up fast. Companies need data to make decisions. But they also need systems that let people challenge the “gold standard” metrics.
Beyond Tech Companies
Every industry faces this challenge:
Retail: Sales data might look strong, but customers in stores struggle to find popular items because they’re poorly placed.
Logistics: Computer programs focus on shortest routes, while local drivers understand real traffic, seasonal road closures, and detours.
Manufacturing: Factory efficiency numbers might suggest peak performance, but workers on the ground see problems that don’t show up in the data.
The Bottom Line
You need data, but data alone isn’t enough. Data can never speak for itself, and it’s always looking backward. It’s risky to believe dashboards show the complete picture.
Successful leaders recognize their metrics have limits. They look for gaps between data and reality. Your data systems must include human judgment in decision-making. Ignore this and you’ll create flawed strategies, miss opportunities, and slow your growth.
Understanding your data is just the first step. You also need systems that can act on that data while staying flexible when reality doesn’t match your numbers...
This is Chapter 1 of Growth Isn’t One Sided, a mini-book I’m sharing weekly on how to grow your business without breaking it. [Read the Introduction: How Lyft and Uber Taught Me to Think Differently] | Subscribe for weekly insights
https://www.startuparchive.org/p/jeff-bezos-recounts-the-time-he-called-amazon-s-customer-service-number-mid-meeting-to-prove-a-metri