The Most Effective Creatives Maximize Leverage, Not Hours Worked | by Samuel Flender | Sep, 2022

By Jessie Hobb On Sep 22, 2022

Forget ‘quiet quitting’: 3 strategies for creating more business impact with fewer hours

During my career journey so far (first at JP Morgan Chase, then at Amazon), I’ve met some peers that appear to be constantly stressed out: they put in long hours and work hard, without creating that much real impact. On the flipside, I’ve also met peers that consistently deliver high impact to their organization without appearing stressed at all, all while working reasonable hours.

How can this be? Why do some individuals appear to so much much more effective than others?

At first I thought the answer is luck: certain individuals are simply handed better and more impactful projects to work on, and therefore produce more impact in less time. It’s true that luck plays a role, but I argue here that it’s not the main determinator in how much impact you create inside an organization.

Instead, the answer is leverage, the ratio of impact created relative to time invested. Leverage is the central, guiding metric that effective individuals use to decide where and how to spend their time, says Edmond Lau in his book The effective engineer.

Leverage is a powerful concept: if your leverage stays the same, then you can only produce more impact by working more hours, and your hours are fundamentally limited. Working overtime to deliver more impact is not sustainable. Instead, by increasing the leverage of your work you may be able to increase your impact by 10X or more, all while keeping your hours reasonable and sustainable.

How to increase your leverage as an individual contributor inside an organization is a big topic covered by people such as Lau. In this post we’ll dive into just 3 strategies which I believe to be the most powerful, with a few particular examples from the data science profession:

manage your time well,
optimize for learning, and
validate your ideas early.

Let’s get started.

1. Manage your time well

“I do one thing at a time. This is what computer scientists call batch processing — the alternative is swapping in and out. I don’t swap in and out. […] What I do takes long hours of studying and uninterruptible concentration.” — Donald Knuth

One of the most leveraged things you can do is learning to manage your time well: with a little bit of upfront planning, you may be able to free up several hours of your time each week that you can then invest into more impactful work. If you feel constantly stressed out yet have little impact to show for, that’s a good indicator that perhaps you need to improve your time management.

Here are some ideas on how to manage your time better.

Minimize context switching. Context switching, a term from Computer Science, refers to the act of temporarily stopping one task in order to start another task. You switch contexts, for example, when you stop your project work because of an Email notification or a meeting. Context switching is much more costly than people might think: “A 20-minute interruption while working on a project entails two context switches; realistically, this interruption results in a loss of a couple of hours of truly productive work”, writes Dave O’Connor in Site Reliability Engineering: How Google runs Production Systems. Make a conscious effort to reduce context switching as much as possible, and stay on a single task as long as possible. Don’t ‘swap in and out’, as Donald Knuth would say.

Practice interrupt coalescing. A large contributor to context switches is the prevalence of adhoc requests: “why does this time series have a spike?”, or “why did that metric suddenly drop?”, and so on. In their book Algorithms to live by, authors Brian Christian and Tom Griffith recommend interrupt coalescing, another term originating from Computer Science: instead of dealing with an adhoc request immediately, add it to your backlog. Then, once a week or so, set up a few hours to take care of all of these small requests in one batch. Interrupt coalescing has high leverage because it frees up a disproportionate amount of time (because of the reduced context switching) with a simple tweak in your time management.

Minimize meetings. It’s no secret that not all meetings are useful. Particularly problematic are recurrent meetings that cost you time and context switches each week. Amazon principal engineer Steve Huynh has a practical tip for this problem: one week, simply don’t show up to the next instance of a recurrent meeting that you think is useless, and see what happens. If no one asks where you were, that means no one even noticed you were gone, and you’ve found a meeting that you can permanently drop from your calendar. If someone comments on your absence, you can apologize and say you’ll be back next time.

Set up specific hours each day for focused work. Huynh recommends to consistently focus for 2–4 hours every day on the most important thing you need to get done. Make it a habit to turn off any notifications during that time: remember that context switching is much more costly than you might think. I also found it useful to use a simple stopwatch to time my focused work. Another good idea is to block these hours in your work calendar, so that others don’t schedule an unexpected meeting during that time. The most effective professionals are those that consistently find the time to focus and chip away small pieces of a big problem.

2. Optimize for learning

“I absolutely know it’s hard, but we’ll learn how to do it.” — Jeff Bezos

Learning a new skill, such as a new framework, API, or programming language, is perhaps the highest-leverage activity you can spend your time on. For example, if learning a new framework takes you a few hours of your time but makes one of your regular workflows 10% faster, that’s an investment that is guaranteed to pay off over time.

‘Optimizing for learning’ means that you make deliberate choices that force you to consistently learn new skills. Here are some ideas on how to do that:

Take on projects in a problem domain you are unfamiliar with, don’t just stick to problems that you already know how to solve. Trust that you’ll ‘learn how to do it’, as Jeff Bezos would say.
Join a team where people are smarter than you. Switch teams or companies if you feel that your learning has saturated.
Read books, technical blogs, and research papers. Again, the principle of leverage applies: focus on those resources that can bring you the biggest gain for your studying time. If a paper has no application in your problem domain, the leverage of reading it is more or less equal to zero. Focus on applied papers: a good resource is Eugene Yan’s curated list of applied ML papers.
Shadow a more experienced engineer or data scientist during their workflow. Sometimes it’s just that one little trick that ends up saving you a huge amount of time, and the only way to learn that trick is by shadowing.
Read and try to understand other people’s code. The more code you read, the better you become yourself at coding. Try to understand what’s going on under the hood of common APIs that you use, such as scikit-learn.
Start writing and create your own blog. These days, it’s easy for anyone to start blogging with platforms like Medium or Substack (and I have some tips to get started here). Follow the Feynman technique: the best way to learn something is to teach it.

Lastly, Psychologists have introduced the (overly simplified, yet still educative) dichotomy of fixed vs growth mindset. People with a fixed mindset tend to avoid challenges because they’re afraid to fail. People with a growth mindset embrace challenges because they welcome the opportunity to learn, even if they fail at first. Make it a habit to adopt a growth mindset: even if you don’t yet know how to do it yet, you’ll figure it out along the way.

By the way, Jeff Bezos said the words quoted above when he proposed the idea that Amazon should build a digital book reader, which later became known, of course, as the Kindle. Amazon had not built electronic gadgets before, and therefore colleagues were skeptical of his idea. His response, “I absolutely know it’s hard, but we’ll learn how to do it”, is great example of a growth mindset.

3. Validate your ideas early

“What’s the scariest part of this project? That’s the part with the most unknowns and the most risk. Do that part first.” — Zack Brock, engineering manager at Meta

Perhaps the concept that epitomizes the concept of leverage is that of the minimum viable product, the MVP. The key idea behind the MVP is to validate your value hypothesis (i.e., the hypothesis that the proposed product or piece of software has economic value) as early as possible with as little investment as possible. Designing a good MVP is therefore an activity with high leverage: a little bit of upfront investment has a disproportionate amount of business impact, namely the ability to confirm whether we’re on the right track.

A good example for this concept is the story of the shoe e-commerce platform Zappos. Instead of actually building the physical and digital infrastructure needed for such an application, the founder Nick Swinmurn simply went to his local mall, took photos of the available shoes, and uploaded them to his site. When a customer would make a purchase, Swinmurn would go back to the mall, purchase the shoes, and ship it to the customer, giving a small commission to the store owner. Zappos turned out to be a success, and was later acquired by Amazon for $1.2B. But importantly, Swinmurn was able to confirm his value hypothesis early on with very little upfront investment: just the cost of a digital camera.

Let’s also take a look at a counter-example. Lau, author of The effective engineer, recounts the story of Cuil, which was supposed to be the ‘Google killer’: the founders of Cuil hypothesized that they would be able to beat Google by offering a search engine with a bigger search index. (A search index is a lookup-table that maps search queries to URLs: a bigger search index means higher recall at the expense of longer querying time.) The Cuil team spent years building their new search engine, but in the end it flopped. The UI was too slow and buggy, and it turned out that users prefered low latency over a bigger search index. Josh Levy, inventor of Cuil, said he should have validated his product sooner:

Don’t delay. Get feedback. Figure out what’s working. That’s by far better than trying to build something and the trust that you got everything right — because you can’t get everything right.

So how can you validate your ideas early as a data scientist? Here are a few practical tips.

Use your peers as a sounding board. Make it a habit to pitch your ideas to your peers. Usually, people are happy to hear about a new problem or idea and give their opinions. It brings them out of their usual rut. Having informal coffee chats with your colleagues is a great way to initiate such an informal exchange of ideas.

Write a 1-pager before doing any actual work. In that document, outline the problem you’re trying to solve, how you want to solve it, and what’s your proposed success metric. Ask people to review the 1-pager: if there’s a serious flaw in your proposal, you’ll want to know before you actually do the work and potentially waste your time.

Solve the scariest part of the project first. Think about which piece of the project has the most unknowns. In ML projects, this is often the integration of the model with the backend infrastructure: can the model actually take automated actions at inference time? Solving this part first is a great way to validate your ideas early.

Deploy your ML model as soon as possible. It is often tempting to keep optimizing your model offline and try out ‘one more thing’ before deploying to production. However, offline performance does not guarantee online performance, and therefore the leverage of spending more time on improving offline performance may be quite low in practice. Data Scientist Damien Benveniste recommends time-boxing every step in the ML development process, so that you don’t feel tempted to go down the rabbit hole of offline optimization.

Conclusion: instead of ‘quiet quitting’, maximize leverage instead

Leverage, the ratio of impact created to time invested, is the central, guiding metric that you should use to decide how to spend your time. You can improve the leverage of your work, among other things, by (1) learning to manage your time well, (2) optimizing for learning, and (3) validating your ideas early.

As I was doing the research for this article, ‘quiet quitting’ became a new buzzword, popularized by engineer Zaid Khan in a TikTok video. The idea behind quit quitting is to ‘coast’ and do the minimal amount work that’s expected to keep your job, so that you can have more personal time for yourself. Khan explains:

“You’re not outright quitting your job, but you’re quitting the idea of going above and beyond. You’re still performing your duties, but you’re no longer subscribing to the hustle culture mentality that work has to be your life. The reality is it’s not — and your worth as a person is not defined by your labor.”

From a point of view of leverage, which we take here, the premise behind quiet quitting, namely that you need to work long hours in order to go “above and beyond”, is misguided. Instead, the study of leverage teaches us that we can go above and beyond while working fewer, not more, hours. The key is to maximize the leverage of these hours of work, not the number of them.