Novice Investor #2 - Calendly, Computer Generated Images and How to Analyse a Model in Under 30 Minutes
Welcome to issue #2 of the Novice Investor Newsletter. Massive thank you to everyone who has signed up to receive their monthly dose of spam. If you like what you read, please share with a friend.
This month was a productive one. With my weekends restricted, personal learning and work have filled up the empty time. I feel like my productivity is going to take a hit once we are all allowed socialise again. Until then, I am perfectly happy taking advantage of the quiet time.
At Kennet, we announced two investments. We led a $17m investment into Provar Testing, a Salesforce focused test automation company. We also led a $13m investment into Grip, an intelligent networking software for virtual, hybrid and live events. We are excited to help both companies grow over the next few years.
On a personal side, I have moved once again. This time the move was only 5km to a longer term let in Pimlico. Here is a look at my new neighbourhood. If you live nearby, let’s grab a coffee!
Interesting Companies
Sky Engine is a company I have known for a few years and is one I find pretty cool. In the world of computer vision models - data, in the form of images, is a strong defensible moat. Big players like Google, Facebook and Microsoft have an unfair advantage given their incredible data stores. While startups or corporates must partner or purchase to get enough images to train models.
Another big issue with developing computer vision models is the need to tag every image. If you are training a model to identify a cat, you will need training data with tagged images of cats. Tagging images is typically a manual and laborious task. Think Captcha when you are deciding which images contain a traffic light. All this adds to the cost of model development and hurts gross margins.
Sky Engine have a solution to both problems. Their platform creates computer generated images to train computer vision models. As they create each image, they know exactly what each pixel is, so no need to manually tag. Model builders then have access to unlimited tagged data to train their models. They can even focus on weak points in the model and create images to train for difficult edge cases e.g., distinguishing between a cat and a tiger.
Investor Tips
Back of the Envelope Model
When making investment decisions, numerous factors play into the decision process. One aspect for later stage investors, the financials, is always paramount. Put simply, if a fund invests $10m, they need to understand what it is going to be spent on and how long the money will last if things do not go exactly to plan. This is usually articulated in an excel financial model.
These financial models are often detailed and complex. Personally, I never trust an excel sheet I have not built. All too often I have spent a full day getting to grips with a complex model and find numerous hardcoded or hidden assumptions, which throw off my logic. Instead of investing too much time getting comfortable with and running scenarios on a complex model at an early stage in the process, I rely on a simple trick to solve the issue.
Essentially, I build an extremely basic model myself and flex the assumptions there. Here is a screen record of how I do it.
Productivity Tech
This is probably the quickest time-to-value when it comes to productivity tech. A complete no brainer in my opinion. Calendly is a super simple tool that automates meeting scheduling. It links to your calendar. You set when you are available to take meetings and it overlays your current calendar to block times you are already booked for.
When setting up meetings, instead of crosschecking your calendar and typing emails with various time slots that suit you, simply hit them with the “please schedule through this link”. They pick a time slot that suits and Calendly automatically sends out a meeting invite with a zoom link to both participants.
It will take 15 minutes to set up and will definitely save you at least 30 minutes per week if you schedule a lot of meetings. Plus, if you are happy with an unbranded Calendly landing page, it is free.
Only downside is the risk of coming across a bit arrogant asking someone to book a meeting with you via your Calendly link. This is a risk I am happy to live it with given the efficiency gains. Apologies in advance for my arrogance.
Books & Podcast
The Hitchhiker's Guide to the Galaxy
Following last month’s tedious, but extremely helpful read of Getting Things Done by David Allen, I decided to go easy on myself and indulge in some fiction. I managed to read the book in about three days. The big learning for me this month - books are far more enjoyable to read when they are not self-help related.
I very much enjoyed reading this book. In truth, I am a closeted Sci-fi fan. Plus, Douglas Adams’ sense of humour struck a chord and gave me plenty of laughs. The book is not full of life changing learnings, but at least I now know why some startups include the number 42 in their name. Highly recommend.
DoorDash - Acquired
Sharing another Acquired Podcast. I thought about mixing it up, but this episode was super insightful.
It highlighted how positive unit economics are not always necessary for a great investment. DoorDash had a long history of poor unit economics in its growth phase i.e., spending $10 acquiring a customer who only ever spends $5 through their platform. Alfred Lin of Sequoia, Series A investor in DoorDash, backed the idea that the company can continue to expand aggressively to acquire market share with poor unit economics. Once the company has enough volume through the platform, it can set prices lower than any other competitor, while still generating a marginal profit. Smaller competitors will not be able to compete and fail, essentially creating a monopoly and winning the market.
At the time of the Series C in 2016, no one else in Silicon Valley believed in the idea. Sequoia was forced to lead an internal follow-on round. However, Sequoia had already seen this strategy play out in China. Their portfolio company, Meituan had already become the food delivery market leader in China and experienced an uptick in unit economics once it reached critical mass. Luckily for Sequoia, the same thing happened in the US for DoorDash and the earlier bad unit economics flipped.
It is a very high-risk investment strategy. Extremely capital intensive, high competition and usually a winner takes all outcome. DoorDash’s $65bn market cap is not a bad prize though. Investing in bad unit economic companies chasing a theoretical pot of gold is probably not an investment I will be making anytime soon. From a learning side, it does prompt me to think twice about some opportunities where there is a compelling reason for unit economics to improve. The reason will just have to be extremely compelling.
Best of YouTube
Cool history, overview and latest news on the Vision Fund. Probably comes across a little negative in my opinion. The WeWork investment very much rocked the fund and its public perception, but I think the long-term net impact of the fund is positive for the technology sector.