Different situations have different levels of error tolerance.
Most professionals work in environments with an extremely low tolerance for mistakes. Accordingly, most employers spend a lot of time training people to complete perfect work.
This is perfectly logical! Mistakes have serious consequences at big companies, for three main reasons:
Scale: Mistakes compound in large, complex systems; a tiny error rate in all the cogs of a machine will shut the whole thing down.
Brand: If you're the representative of a big, established brand, any mistakes can damage the brand's hard-won reputation. Even trivial errors can go viral if they are seen to represent the brand.
Promises: If you make a promise to someone & break it, they will remember it. This is true for customers, suppliers, partners, and more. As a company grows, the number of promises increases exponentially.
Big companies train people to avoid mistakes. They use heavy processes to ensure that mistakes are caught quickly. They also limit the scope of each individual’s work to minimize the potential impact of each mistake.
These methods of mistake minimization come with a big tradeoff: they reduces each employee’s productivity and slow things down.
People who grow up in a “big company” system internalize the norms and practices that they learned in that environment. Unfortunately, those practices are bad if you’re working at a startup; in a startup, productivity matters more than perfection.
In this post, I’ll lay out why your team should make more mistakes, and how you can get them to change.
Why mistakes matter less at startups
If scale, brand, and promises are why mistakes matter at big companies, they are why mistakes don’t matter at startups.
Startups are typically small, so there’s implicitly more error tolerance from each part of the system. Mistakes can only compound so far.
Startups are usually unknown to most people. If you make a mistake on behalf of the company, it’s very unlikely to go viral on Twitter. No one will care, because no one knows who you are!
Most startups haven’t made promises to many people. Of course, you have to keep any promises you’ve made to customers, but you probably don’t have that many customers. You definitely don’t have a lot of contracts with suppliers, vendors, or partners.
All of this means that mistakes cost less at a startup. They still cost something, but at a startup, something else costs more: being slow.
The only real edge that a startup has over incumbents is speed. Startups must build a culture where speed is paramount, and accepting mistakes is a key part of that culture.
Why you should make more mistakes
If your team is working hard and moving fast, mistakes should be celebrated. It’s a great sign that the team is pushing themselves to their limit and learning a lot.
In general, the number of mistakes you make are inversely correlated with the time you spend on something. But it’s often good to be further up the curve than feels comfortable.
Most people spend a lot of time working on something after it’s been complete to ensure there are no mistakes. This is drilled into people at big companies. That’s the point farthest to the right.
Unfortunately, it’s hard to prevent mistakes by spending more time on something. We’ve all been here: you’ve read through a document 4 times to ensure nothing is wrong, only to miss a typo in the header, or have a formatting error when you convert it to a PDF.
Because of this, almost everyone would be better off migrating to the midpoint. At this point, you’re trading off a tiny increase in mistakes for significant gains in productivity.
But people have a hard time doing this, because the pain they’ve felt from being caught in a mistake far outweighs the benefit of getting more work done. This is a classic example of loss aversion.
In a startup, your optimal position is actually even farther left. At this point, you’re going to be cranking work out, but making mistakes pretty frequently.
There are four main reasons why this is where you want to be:
1. Perfect work <> valuable work
The value of the work you do has little correlation with the number of mistakes you make. If you do great analysis that uncovers a valuable insight on the product, no one cares if your graph has formatting errors or a few words in the doc are misspelled. If you get to the meat quickly, you don’t need to be perfect.
It’s very easy to do “perfect work” that isn’t valuable. Most people spend most of their careers doing exactly that.
2. Working faster means more valuable work per hour
If you plot time vs. value of work, it’s an inverse of the mistake curve; the majority of the value of the work typically comes early.
This is a riff on the Pareto principle, and is typically referred to as the “80/20” rule (80% of the value comes from 20% of the work you do). It was drilled into me at Bain - which is ironic, because Bain is culture that places a HUGE value on perfection (like all client service companies).
By giving yourself permission to work faster and make mistakes, you can get a lot more done.
Let’s say you’re writing outbound emails. You have a choice of writing 100 emails but getting 2 or 3 people’s name wrong, or writing 50 emails and getting everyone’s name right. Which do you choose?
At a startup, it should be option A, every time. Most people won’t notice you got their name wrong, few will care - no harm done! But you sent twice the number of emails, which could have 2X the impact.
Even if a task doesn’t “scale up” like this (i.e., you can’t do 2X more of that task for 2X impact) you always have additional, high-value things on your plate - so saving time on one task means you’ll complete another.
3. It’s hard to catch mistakes yourself
The biggest mistakes are the most fundamental; they’re the sort of things that invalidate the work altogether. The sooner you share work, the sooner others can validate the work and catch foundational mistakes.
Like I said above, we’ve all had the experience of a reviewing a document 5 times only to miss big typos. Even worse, we’ve had the experience of running a deep analysis, writing out all the implications, and then discovering that we used the wrong formula in the middle of presenting our findings, which ruin the whole thing.
At startups, typos are okay, but wasted effort isn’t. Don’t waste time by holding your work back!
4. Expectations scale with time
The expectations for work are the inverse of the time you spend on the work. If you spend days on a project, others will expect it to be perfect. If you spend hours on it, they’ll assume there are some mistakes. You always want to be on the right side of expectations.
This is true for customers as well as for your colleagues. If a customer spots a bug and you fix it quickly, they’re more likely to forgive rough edges in the output. They’ll also be more likely to flag bugs next time, because they’ll get a response.
How to get your team to make more mistakes
So you’ve read this, you agree, but your team is still working too perfectly & too slow. How do you fix it?
Here’s a shortlist of tips that you can use as a manager:
Give them permission to make mistakes: Explicitly discuss the tradeoff of mistakes vs. productivity with them (maybe by sharing this post), and give them permission to have errors
Don’t criticize mistakes: When you see mistakes in their work (like typos, bad formatting, etc.) don’t correct it. Just let them go, if they don’t matter!
Note: If there’s a scenario where they actually matter, you should tell them, but be clear about why they matter. Also emphasize that you only care about mistakes in certain scenarios
“Time box” work: Don’t give people unlimited time to complete a task. Say “I’d like to see your 4 hour version of this analysis, then let’s talk”. Alternatively, if you don’t want to prescribe specific amounts of time for work, give them tight deadlines.
Give them too much to do: If you give your reports a ton of tasks to juggle, it forces them to make prioritization decisions and kill unnecessary tasks (like fixing formatting)
If you’re a perfectionist, these probably sound stressful. But that’s a good thing!
Change is always stressful. As a manager, you need to push your team to drop the framework that served them well in former jobs, and work with a new set of norms that are suited to the current environment.
There’s one thing final thing you do want to be careful of: making sure your team is making mistakes, and not being stupid.
Make mistakes, but don’t be stupid
Mistakes happen when you do your best to achieve an outcome with the information you had, but it didn’t work. People typically make more mistakes when:
They’re moving super fast
They’re doing something outside of their expertise/skillset
They’re making bets with limited information
They’re juggling a large number of different tasks
In a startup environment, you need people to be doing all four of these. If they’re not, your startup is not going to be around for long.
But that’s not an excuse to be stupid. People are stupid when:
They don't use data to make decisions
They don't evaluate the impact of their actions
They do things with a lot of risk and little reward
They change something in a system that they don’t understand
Sometimes it’s hard to tell which is which. Looking back at my time at Statsig, I’ve made a lot of mistakes and I’ve been stupid a lot. Some examples:
Mistake: Pouring time, energy, and money into optimizing paid conversions via search ads, rather than building our organic presence
Mistake: Moving our GCP billing account from the company billing account to “my project” billing account
Stupid: Deleting our ‘latest’ cluster as I was trying to automate Azure billing reporting, because I didn’t understand our Azure implementation
One good way to differentiate mistakes from stupidity is by looking at what you learned. Mistakes should tech you something more than “I shouldn’t do that again”.
In the examples above, both mistakes made the company better. By putting time, energy, and money into paid search, we learned about the efficacy of that channel and were able to make the decision to wind it down. By accidentally switching our billing accounts, we learned about that weakness in the system - and subsequently set up controls.
The stupid example taught me nothing. I don’t manage our Azure deployments, so it didn’t make me better at my job; I just learned that I shouldn’t delete clusters again. Our engineering team still teases me about it.
How I stopped worrying and learned to love mistakes
Mistakes are one of the great joys of working at a startup.
Most places try to train the mistakes out of you, until every piece of work is perfect. There’s value in that - it is important to know how to do perfect work. But it’s equally important to recognize when perfection doesn’t matter; when you can expand your error rate to increase impact.
This philosophy opens the doors to a whole world of creative work that isn’t possible in most places.
A high tolerance for mistakes means freedom to pursue crazy ideas, work on things without asking for permission, and absolutely crush to-do lists. It’s an amazing feeling.