Understanding the Art of Not Obsessing Over Data
Written on
The year 1957 marked the introduction of Ford's ambitious ten-year plan for the Edsel, a car projected to sell 2 million units in its debut year, backed by extensive market research. The outcome? A staggering loss of $250 million, despite the wealth of data informing the project.
One theory explaining this misstep is the "moving target" hypothesis, suggesting that although Ford excelled at data collection and execution, they neglected to adjust their insights once production began. Consumer preferences shifted, and Ford failed to recognize these changes due to their overwhelming reliance on data.
Sometimes, it might be wise to disregard data altogether.
For over ten years, I have been deeply invested in data initiatives, decision-making processes, and machine learning. Yet, there were instances when I chose to dismiss conventional wisdom, best practices, and even the opinions of others—these moments often led to significant breakthroughs.
Throughout my journey, I co-founded a business intelligence startup (which ultimately failed), served as an analyst, participated in a fast-paced data startup, managed a data team as a product manager, and through my writing, engaged with numerous inspiring data entrepreneurs.
A recurring theme has emerged over the last decade:
> Companies and their leaders often focus on data when they really shouldn't. They become so absorbed in details that they lose sight of the bigger picture. As a result, they draw the conclusion that data, AI, and ML are either too complex to yield business value or simply not essential.
The crucial skill that businesses and managers must cultivate is the ability to resist the impulse to "do data." They must learn to disregard excessive data concerns.
Remember that effortless kid from school? The one who seemed uninterested in anything, including the attention of peers? Ironically, they often attracted the most interest.
You ought to emulate that person! It’s not about apathy; it’s about control and discernment regarding what deserves your attention. This perspective allows for a broader understanding of the situation.
Data can easily ensnare you in a vortex of complexity, leaving you overwhelmed and ineffective.
Here are several areas where you might want to reconsider your investment of energy:
1: The Obsession with Best Practices and Others' Successes
Ford diligently followed established best practices to ensure the Edsel's superiority over competitors, believing that market acceptance was assured.
Yet, these so-called best practices proved misguided; the competition hardly mattered.
Every month, Gartner highlights a new "hot trend," while the number of data writers grows exponentially. Numerous companies tout their best practices for data usage, claiming to transform organizations and promote data-driven cultures.
While I acknowledge their experiences may be valid, it doesn't necessarily qualify them as best practices.
> Best practices are simply those methods that others can successfully replicate for similar outcomes.
To date, I haven't seen any example of a company adopting such best practices and achieving tangible business results. Ultimately, what matters is increasing growth and revenue.
Many organizations have claimed to implement "Airbnb's best practices" but rarely share how these initiatives impacted their bottom line. The reality often indicates a significant gap between effort and reward.
A reliable strategy for evaluating current data trends and practices is to ignore them half the time.
2: The Focus on Adversity
Mark Manson, in "The Subtle Art of Not Giving A F*ck," emphasizes that not caring doesn't equate to indifference; it means prioritizing the essential.
This perspective is particularly relevant in data work. A recurring phrase in my career has been, "This isn't feasible."
- Collecting web data for resale? Not feasible for a small startup (though we achieved it).
- Embedding analysts into departments? Not feasible.
- Integrating machine learning into legacy systems? Not feasible.
- Shifting a data team to create a platform for others? Not feasible.
- Managing products within a data team? Not feasible.
Data resembles oil in that while it's often seen as valuable, extracting it requires substantial investment.
> Data is challenging to monetize, but it can be invaluable and fuel an entire organization.
When faced with difficulties, the natural human tendency is to resist change. This resistance intensifies when confronted with something as daunting as data.
To succeed with data, you must dismiss concerns about adversity—not in a reckless way, but with a clear understanding of your objectives.
You may not win friends in the process, but you shouldn't actively create enemies.
3: The Preoccupation with Data Quality
Traveling from London to New York takes about 8 hours and 15 minutes—an astonishing duration considering we've had supersonic travel for decades.
The Concorde, while an engineering marvel, ultimately failed due to its inability to turn a profit. Focusing solely on quality can lead to missing the target.
I once led a project aimed at establishing an internal tracking system to assess product feature effectiveness on revenue. After weeks of refining data capture formats for quality, I realized that any basic tracking method would have sufficed.
The lesson learned? Data quality isn't a one-size-fits-all measure; it depends on the specific task at hand.
> Data quality is defined by whether the data is in the right form for its intended use.
Focusing excessively on data quality can divert attention from taking actionable steps.
The only scenario where you can justifiably concentrate on data quality is when you truly believe your organization is data-driven. Otherwise, prioritize action over perfection.
4: The Importance of BI & Analytics
> “Begin by seizing something which your opponent holds dear; then he will be amenable to your will.” (Sun Tzu)
There exists an established method for building business intelligence and analytics within organizations, typically starting with hiring a small data team to gather information into a data warehouse, establishing a modern data stack, and producing reports and dashboards for users.
However, this often leads to a classic oversight: failing to recognize the broader context. Many companies hire data teams without a clear understanding of their data strategy, resulting in expenditures on technology and reporting without achieving genuine business intelligence.
One should disregard modern data stacks and conventional views of "BI" and "analytics." Most people overlook the core definition:
> “Business intelligence (BI) encompasses a range of strategies and technologies that organizations use to analyze information and convert it into actionable insights for informed decision-making.” (CIO Voice)
The focus should always remain on actions and decisions. If you aim to implement BI and analytics effectively:
- Don’t begin with technology.
- Avoid starting with a data team.
- Refrain from commencing with analysis.
Initiate your efforts with your marketing and sales teams, product managers, and decision-makers. Equip them with data tools and training to facilitate informed decision-making.
Only when demand for data outstrips your existing capabilities should you consider investing in a data team and advanced technology.
Most data teams skip this foundational work because it’s challenging and messy. However, neglecting this step leads to wasted resources.
Achieving effective BI and analytics requires a departure from conventional thinking. Some label this approach as “data democratization,” yet it’s essential to recognize that you can empower employees to utilize data immediately without a centralized system.
5: The Misconception of Data as a Product
> “Standing in the middle of the road is very dangerous; you get knocked down by the traffic from both sides.” (Margaret Thatcher)
It's common advice to treat data as a product, with discussions surrounding curated datasets and data products. Yet, the reality is more straightforward: there are products and there is data—nothing more.
For product managers and business leaders, the distinction matters little. It’s essential not to get bogged down in the concept of "data as a product," as it often leads to confusion.
Instead, focus on actual products, as they directly influence your organization's success. While incorporating data into your offerings is advantageous, simply introducing data haphazardly will lead to inefficiency and confusion.
6: The Shift from Big Data to Smart and Small Data
The era of big data is fading; today, companies are increasingly leveraging small and wide data.
Some experts emphasize "good data" as a primary goal. To illustrate, consider the oil industry:
- Finding new oil fields reflects the "big" strategy.
- Identifying the best locations to pursue corresponds to the "small" strategy.
- Maximizing value from existing resources represents the "smart" strategy.
However, the latter two strategies have a natural limit that depends on discovering new oil reserves.
> If you think you’re focused on smart or small data, remember the ultimate objective: big data.
While maintaining a balance between various data strategies is crucial, don’t lose sight of the need for continual input and expansion.
7: The Reality of Data Teams
Transitioning from a machine learning engineer to a data product manager required significant adjustments to my mindset, particularly in overcoming biases toward technology and data.
> The truth is that data (science/ML) teams are not the ones delivering products; only product teams can do that. Data teams typically exhibit technical biases and may lack a business-oriented perspective.
To create products that integrate seamlessly into your organizational strategy, prioritize assembling competent product teams rather than focusing on data teams.
A well-rounded product team should include:
- A business-oriented product manager (an engineer may not be the best fit)
- A data engineer
- A data scientist
- A software engineer
- A UX/frontend engineer
While team composition may vary, a complete product team is essential for developing successful products and effectively managing data-heavy tasks.
8: The Key Takeaway About Data
Make a conscious choice to stop fixating on data. Focus on the product instead. If you identify a way to incorporate data effectively, invest in that without hesitation. If not, don’t force it.
The underlying principle is straightforward, yet my experiences indicate that it’s challenging to implement.
But then again, worthwhile endeavors often are difficult, aren’t they?
Consider subscribing to my free newsletter, "Three Data Point Thursday." It has become a trusted resource for data startups, VCs, and data leaders, aimed at enhancing your business with data and AI.
Interested in data engineering? I share my top six articles each week in my free newsletter, "Finish Slime."