The Question
When you're trying to do something well — build a company, write better, exercise consistently, raise a child — the first instinct is to look for what's worked before. What are the best practices? What does the literature say? What do successful people do?
This instinct is reasonable. It saves time. It prevents you from making mistakes that someone else already made and documented. It signals to the people around you that you've done your homework.
It's also where a lot of intelligent people go quietly wrong.
Not because the best practices are false — many of them are well-observed and carefully tested. But because the act of borrowing a solution carries a hidden assumption: that your problem is the same problem the solution was designed for. Most of the time, it isn't. And the gap between "similar enough" and "actually the same" is where borrowed wisdom becomes expensive.
The Argument
Best practices are optimized for the median case, not your case.
The startup playbook that works reliably for B2B SaaS in 2024 — build a tight ICP, focus on outbound, hire sales before marketing, land-and-expand pricing — didn't exist as a coherent playbook in 2005, because the market structures that make those practices work didn't exist yet. By the time a practice is called "best," it has been selected from a distribution of outcomes across a particular set of companies, industries, and time periods. That selection process bakes in context. The context becomes invisible.
The same problem appears everywhere. The "evidence-based" nutritional guidance that recommends X often reflects studies of populations that skew toward middle-aged Western men — which is to say, the evidence is real but the population it was gathered from may not resemble you in the ways that matter. The management frameworks developed at GE in the 1980s — forced ranking, rigorous performance calibration, chainlike reporting structures — encoded assumptions about hierarchy, information flow, and the nature of manufacturing work that often don't hold in a twenty-person software company in 2024. They were best practices for a specific kind of industrial organization. When applied elsewhere, they produced specific kinds of damage.
George Box's famous line — all models are wrong, but some are useful — is usually quoted in the context of statistics, but it applies here too. A best practice is a model. It compresses a complex historical reality into a reusable heuristic. That compression is precisely what makes it useful at scale. It is also precisely what makes it dangerous when applied without thinking.
The deeper problem is that best practices tend to come stripped of the conditions that made them work. You get the prescription without the diagnosis. You're told to do stand-up meetings every morning, but not told that this practice was designed for teams with chronic coordination problems and information silos — teams where daily synchronization genuinely reduced waste. If your team of four sits together in one room and talks constantly, a daily stand-up is not a best practice. It's cargo cult behavior dressed up in respectable clothing.
W. Edwards Deming spent decades watching American and Japanese manufacturers adopt quality practices without understanding the system-level conditions that made those practices effective. Copying the tool without understanding the system produced organizations that performed the ritual without achieving the result. His frustration was not with the practices themselves — he had developed many of them — but with the wholesale transplantation of solutions from one context into another without any examination of fit.
The Counterpoint
But rejecting best practices wholesale is also a failure mode, and it's worth naming clearly.
The person who insists on reinventing every wheel is usually reinventing it worse than the original. There is real value in accumulated knowledge. Medical protocols, building codes, aviation checklists — these are best practices that represent hundreds of thousands of hours of failure analysis and careful refinement. The surgeon who decides their intuition supersedes evidence-based protocols is not practicing discernment. They are practicing arrogance, and their patients bear the cost.
There is also a particular trap that affects smart, independent-minded people: the reflex to see "best practice" and immediately ask "but does this apply to my situation?" can become its own avoidance mechanism. If you can always construct an argument for why the standard approach doesn't fit your context, you can always justify starting from scratch. Starting from scratch feels like thinking. Often it's just delay.
The real skill is not following or rejecting best practices. It's knowing when each is appropriate. That requires understanding both the practice and the context, which is harder than either reflexive adoption or reflexive rejection. The entrepreneur who follows the SaaS playbook in an entirely novel market, where none of the playbook's underlying assumptions hold, is going to burn time and money applying solutions to problems they don't have. But the entrepreneur who ignores every playbook on principle because they believe their situation is unique will spend three years learning things that are freely available to anyone who reads carefully.
The test is not "is this a best practice?" The test is "does this practice fit my situation?"
What To Do With It
For the next significant decision you face — in work, in how you manage your time, in how you build or maintain something — do the following.
First, identify the best practice that most people would apply here. Not to follow it uncritically, but to make it explicit. "The standard approach to X is Y." Name it clearly.
Second, ask three questions about it:
What context was this practice developed in? What kind of organization, at what stage, in what market or environment? The answer is usually discoverable with thirty minutes of reading. Most best practices have a documented origin — a company that invented them, a researcher who studied them, an industry that codified them.
Does my situation match those conditions? Not approximately — specifically. Are the underlying assumptions present? Is the problem the same problem? If the practice assumes high information asymmetry and your team has radical transparency, or if it assumes a long sales cycle and you have no sales cycle, the mismatch matters.
What is this practice trying to achieve at a fundamental level? Sometimes you discover that the goal of the practice is exactly your goal, but the method needs to be adapted. Other times you find that the goal itself doesn't apply to your situation at all.
If the conditions match, use the practice. You don't need to rediscover gravity. If they don't match, use the practice as a starting point — a first hypothesis — not as a blueprint. Something developed for similar problems in similar contexts is still a better starting point than nothing, as long as you hold it loosely.
The best practices you'll trust most, over time, are the ones you've interrogated seriously. The ones where you asked "why does this work?" and found a good answer that held up in your context. That understanding is harder to acquire than copying a playbook. It also doesn't expire when the context shifts.
Box's point was not that models are useless because they're wrong. It was that knowing how a model is wrong is what makes it useful. Apply the same standard to the advice you take. Know how it's wrong, and it becomes far more valuable than advice you follow without understanding.
If this hooked you
Where we'd send you next.
|