🛑 Stop making bad decisions
🛑 Stop making bad decisions
As engineers, our decisions shape the quality of our deliverables. Yet, our decision-making process can be flawed. In this comprehensive guide, uncover the three crucial levels of decision-making
Engineers are paid to solve problems
So we need to make decisions.
The quality of our decisions impacts the quality of our deliverables.
However, our decision-making process is often flawed.
Are you aware of how are you making decisions? You’ll find out reading
⭐ In this post, you’ll learn:
The 3 levels of decision making
Strategies to make high-quality decisions
3-levels of decision-making
🔎 Level 1 - Heuristics
Our brain is designed to save energy.
It will rely on proxies to make a decision.
Here are a few:
Whoever works on it first
Leave it up to personal preference
It’s vulnerable to all psychological biases affecting our brain. That’s why all good processes make use of a second pair of eyes.
Democracy
Majority wins.
Democracy helps to take a collective bad decision, not necessarily a good one.
Besides individual psychological biases, we add those impacting group decision-making.
Tenure/Seniority
Usually, these people are seniors because they create a positive impact by making good decisions.
But leaving everything up to them means it’s again a personal preference.
All decisions centralized in a single person means they can dedicate less time to each decision, relying more on personal preference
Sunk cost
Make decisions based on past decisions.
Often it’s good to align yourself with the company’s direction. But that’s not an excuse to skip evaluating the decision and alternatives.
Don’t think that losing $1M after losing $100M in the past is not “too bad”
Think that you are losing $1M. Period.
☯️ Level 2 - Shallow tradeoffs
You study the proposal and alternatives.
You document the pros and cons.
You advocate for making a certain tradeoff and that’s why you chose one of the options as the proposal.
Seems a stronger argument.
Except when someone comes and advocates for one of the alternatives.
📊 Level 3 - Data-driven decisions
This level protects you against “opinions”.
You need to quantify.
For example: The number of requests per second to a service, the number of new features you’ll need to onboard in a year, or the latency budget based on upstream/downstream dependencies.
Use the data to make an argument relating it to these references:
Official documentation, golden paths from the technology owners
Best practices documented in your company, organization, or team
Lessons learned from post-mortems
Last week I talked with someone who got frustrated because people challenged what he believed was the best solution.
And the challenge was “I’d rather do A than B”
My advice: Add references to your proposal.
This way it’s not your opinion, it’s a fact.
If they still want to challenge, they’ll need to find evidence backing their proposal at least to match the level of detail in yours
The lesson to learn is not to remain closed to changing your mind.
It’s to be a truth-seeker.
Make better decisions because you base yourself on real data. And only change your mind when new evidence arrives.
Decision-making strategies
Besides recognizing that good decisions are data-driven, there are more aspects to the decision-making process.
👔 Define the roles and expectations
Who approves? Who provides input? Who to inform?
Having an approver doesn’t mean that person does as they want. It means it’s clear who has the final word. This is especially important when there are multiple parties and they can’t agree.
Ensure you are communicating information at the right level and cadence to all of them. Decisions often stall because people feel like they are missing things.
⚒️ Break down big decisions
Identify when conversations stall, run in circles, or people don’t know what’s the decision to make.
Make the most immediate small decision that unblocks many others.
This way you can parallelize decisions fully unblocked.
Last week I was in a design review of a new service.
It was big, had tons of details, and had no alignment.
Too many details meant the conversation was dispersing and didn’t reach any conclusion.
What unblocked things was identifying the first decision to make and meeting again.
I’m sure you have heard of Steve Jobs or Mark Zuckerberg wearing always the same colors. They focus on the most important decisions.
Apply the same. Don’t decide on the metrics before deciding on the overall architecture. The architecture will impact the metrics, it’s your most important decision.
🚪 Differentiate 1-way doors and 2-way doors
A quick Amazon onboarding 101:
A 1-way door decision means you can’t go back, or it’s very expensive
A 2-way door decision is one easy to reverse
Commit fast to 2-way doors because they are easily reversible.
The cost of not making a decision is bigger than the downside of any of the alternatives.
Act with 70% or less information.
And for 1-way doors, beware to break your bias. Find disconfirming evidence, evaluate risks, and make sure you are asking yourself the right questions.
Aim for high-quality and high-velocity decision-making processes.
Note: Jeff Bezos also refers to type 1 and type 2 decisions in some interviews.
💥 Disagree & Commit
Important for 2-way doors. Even if you still want to disagree and bring more data, the cost of inaction is bigger than the cost of choosing another option.
This is critical for high-ego people: You don’t have to agree with their view.
You just have to acknowledge the tradeoff they are making, even if you think it’s not the right thing.
But here’s the important part: You have to commit and be a team player when working on it. Be a team player.
🎯 Conclusion
Having an impact is not that much about the lines of code you type.
But about the process before you know what to type.
Take a step back and reflect on what strategies from this post you can apply at work.
And start now.
👏 Weekly applause
North Star Podcast Ep 1: Understanding the business well is the secret to success by
. A conversation with two principal engineers filled with takeaways. Just a few: 1) Any test and alarm is a minion working for you. 2) Mid-senior transition is about delegating work to others. Senior-Principal transition is about operating in a different time horizon. 3) You learn more from your mistakes than taking the easy route. Don’t stay head down but involve yourself in your vicinityMentorship’s Role in My Staff Promotion by The Developing Dev. An interesting perspective is to get both an explicit mentor with regular check-ins and learn from those you work with implicitly.
5 Lessons I learned the hard way from 6 years as a software engineer by
. A couple of my takeaways were 1) Make your expected work fast and high quality to create time for your other items and 2) The impact is on building relationships, not much on the choice of wordsRetrospective Meetings Demystified by
. Retrospectives are a tool, you don’t need a formal process with hard rules around it.
Great article, Fran and thanks for the mention!
The 1-way vs 2-way door decision is huge. There was one time I was shipping a change that was a big change to development workflows--essentially preventing any new JS file into the codebase, but it was easily reversible. We could just comment out the line that is adding this form of linting.
Although it was a big change to workflows, it was a small change code-wise and easy to reverse. I was able to use that to argue why we should try it, because we could easily go back if people don't like it.
Comparatively, if it was something that had a high cost of reversal, it would have been a much harder sell
I'm a big fan of the 2-way door decisions. During freelancing, I always felt like an added extra that I could tell the client the what-if scenarios and ensure them that we consider longevity and maintainability when picking our technologies and solutions.
As for data-driven, it depends on what data you rely on: GitHub Star count 😃, market research, and estimating your potential user base and needs. Then, pick the technology that makes the most sense financially for your case. Because, in the end, it all comes down to finances.