💡 Business Brainstorms 💡- My favorite ideas of the week
“By far the worst company to emulate in Startup Make-Believe is Google / Alphabet. Google has a number of structural advantages stemming from its Search monopoly that mean that it’s able to be almost completely divorced from reality, and that has allowed it to grow in strange and unique ways.
Google Search is so profitable that you literally could not stop that 1,000mph freight train if you tried. As a result, everything from Google’s hiring policies, HR customs, technical infrastructure, sales culture, and more are able to be completely divorced from your reality. Google has gourmet meals, perfect build systems, and pays at the 99th percentile – we need to do the same! Well, Google also has an all-powerful money printer, would you happen to have one of those too? Your startup probably literally has more in common with the Wendy’s down the street.”
Quote above is from a great post on Startup Make-Believe.
Google and other big companies make a ton of money and do a lot of weird stuff. But that doesn’t mean it’s the weird stuff that is responsible for their success. Rather, they can afford to do weird stuff because they print so much money no matter what.
Another good example would all the weird stuff Ray Dalio promotes as the keys to Birdgewater’s success when in reality all of it is completely unrelated to the real money-making machine.
“All the best companies are data-driven, and we need to be data wizards so that we can make decisions “correctly” / “better.” Let’s invest heavily in data tracking and analytics to make sure that we have a best-in-class data program.
The company that gets emulated the most here is Meta. Meta has built an extremely strong data-driven culture over the years, and it’s truly inspiring to see how they’ve used it to align and scale their business. Everyone knows what matters to them, and how to prove that they’re making an impact (move the number, make more money).
But you’re not Meta. Meta’s business is complicated to operate but conceptually kind of simple – people open app, people see ad, we make money. More importantly, as a startup your data is small and probably incomplete, if not actively incorrect. This means that being doggedly data-driven will be useless in many cases and sometimes actively counter-productive.”
Quote is from the same blog post on Startup Make-Believe.
Yes, in theory being data-driven is awesome. In reality, hardly any company has enough data to come to conclusions at a statistically significant level.
Your A/B test might seem to paint a clear picture but that doesn’t matter if it’s not statistically significant.
For example, let’s say you send 100 cold emails using 3 different variants and only one of them generates positive reply. Now you’re of course tempted to turn off the losing variants and focus on the winning one. The problem here is that after, say, 500 additional emails sent, the results will be completely different. The “clear trend” you see after 100 emails sent is most likely just a statistically fluke. After, say 500 emails sent, another variant could very well be the winning one.
#3.5 💡 Today’s edition is brought to you by Rewind
Rewind is your truly personalized AI assistant. Automatically record, transcribe, and summarize your meetings without an awkward bot joining the call. Then later, ask Rewind to recall details or write drafts for you using all the context from your interactions. Try it free for 30 days!
(you can sponsor this newsletter here)
"Most people will solve problems that they understand how to solve. Roughly speaking, they will solve B+ problems instead of A+ problems. A+ problems are high-impact problems for your company but they're difficult--you don't wake up in the morning with a solution to them, so you tend to procrastinate... If you have a company that's always solving B+ problems, you'll grow and add value, but you'll never create the breakthrough idea because no one is spending 100% of their time banging their head against the wall every day until they solve it."
Quote is from a lecture Keith Rabois gave a few years back on How to Operate.
This is definitely something I’m experiencing myself. There are always tons of tiny tasks you can work on to keep yourself busy. But that also means you’ll never get to the big, truly meaningful stuff unless you make a conscious effort to block out time for it.
Checking tiny tasks you know how to solve off your todo list is fun. Going back to the exact same problem day after day without making visible progress is not.
The lesson also applies at an organisational level. It is always easy to start solving adjacent problems for your customers, instead of taking on the big truly gnarly problem and solving it really well.
Another thing to keep in mind is that ff there are no easy problems handily available, people have a tendency to create their own problems.
I coded an AI agent that continuously scours the internet for trends and paint points, then brainstorms opportunities at the intersection of them, and categories and ranks them. A few examples below. Will share more in the coming weeks. You can get lifetime access to the full dataset here.
Service Business Opportunities:
AI diplomacy + Autonomous agent cooperation: Develop a conflict resolution platform using AI that simulates complex negotiation scenarios for training diplomats and corporate negotiators, enhancing their strategic planning and cooperation skills.
LLM + AI Ethics: Develop an advisory firm specializing in ethical implications of deploying large language models, providing consultation on bias mitigation and responsible usage.
AI safety + Self-improvement algorithms: Develop a consultancy that offers services to audit and certify the safety measures of self-improving AI systems for businesses before they go live. Ensure that as these AIs evolve, they maintain ethical standards and don't deviate from their intended functions.
Safety tech debate + Public-private partnership: Launch a consultancy firm specializing in advising startups and government agencies on navigating the legislative and public opinion landscape around emerging vehicle safety technologies.
Advanced drunk driving prevention tech + AI behavior analysis: Develop an AI-powered software platform that transportation companies can integrate into their fleets to monitor and prevent impaired driving incidents in real-time, reducing liability and enhancing safety.
Assistive Technology + AI Image Description: Create a startup that develops AI-powered wearable cameras which provide real-time descriptive audio of the user's surroundings, specifically tailored for visually impaired individuals. This could help them navigate spaces more independently and safely.
Carbon capture technology + Sustainable consumption: Launch a startup focused on integrating carbon capture tech into everyday consumer products, educating and incentivizing users to contribute to climate solutions through their purchase habits.
Bulbous bow + Marine pollution reduction: Create a retrofitting service for ships that installs custom-designed bulbous bows, improving fuel efficiency and reducing emissions, targeting older fleets looking to extend their operational life and cut costs.
Sodium-ion batteries + Energy storage: Residential energy storage systems that capitalize on the safety and cycle life of sodium-ion batteries, offering homeowners a reliable and environmentally friendly alternative to current lithium-ion home batteries.
Workflow engine + API integration: Build a custom integration service that connects popular SaaS apps with self-hostable workflow engines for enterprises looking to maintain control over their data while automating processes.
Anti-adblocker techniques + Privacy-focused video consumption: Develop a privacy-first, ad-free video streaming platform that leverages a subscription model. Focus on providing high-quality streams without tracking user data, appealing to privacy-conscious consumers tired of battling adblocker countermeasures.
Data visualization + Machine learning: Develop an AI-powered tool that simplifies creating intuitive and impactful visualizations for non-experts. Use machine learning to automatically select the best type of chart or graph based on the data provided.
Cloud Compute Credits: Create a marketplace for unused cloud compute credits.
Open-source research + Advanced AI models: Create a platform that crowd-sources the development of cutting-edge AI models, providing open access to researchers and monetizing through premium enterprise services.
e/acc (Effective Accelerationism): Sell merch (stickers, T-shirts) that allows people to signal that they subscribe to the e/acc ideology.
As always, if you’re enjoying this brainstorm, I’d love it if you shared it with a friend or two. You can send them here to sign up.
Have a great week,