🎲 From assumptions to data-driven decisions

Founders Factory Startup Bulletin #19 (December)

Welcome to the Founders Factory Startup Bulletin—“Created for founders, by founders”.

Each month, we bring you a round-up of startup and investment stories, key learnings from founders, and insights from the Founders Factory team.

Welcome to the 296 new subscribers who joined us last month! Be part of our 16,197-strong community of founders, investors, and startup enthusiasts—subscribe here:

If you’ve ever played poker, there’s a point in the game where you have to make a call about what the other players have in their hand. There’s no real way of knowing—anything you do is based on a ‘hunch’.

We’re often told not to ‘assume’. But as humans, so much of what we do is based on assumptions. When things aren’t obvious, we skip forward and make assumptions.

In life, this can lead to mishaps and misunderstandings. But in business, this can be fatal. Running your business on opinions and assumptions is akin to sticking a finger in the air to work out which way the wind blows. This is why so many startups are, are aim to be, data driven. In 2022, data can power and automate so many of the decisions founders have to make, more efficiently and more accurately.

Data science is more than a buzzword—it’s a golden rule. But what are we actually talking about when we refer to data science? We posed this question to our Head of Data Science Ali Kokaz, who shared some of the most powerful data science methods and how you can use them in a startup environment.

Also in this month’s newsletter:

• Our top recommended reads

• Highlights and news from our portfolio

• Opportunities for founders

---

Before we start—shout-outs!

We’ve got two shout-outs to kick off this month’s newsletter. First, Gary Izunwa, founder of Tangent: an employee referral marketplace connecting talent from lower socioeconomic backgrounds (63% of the UK population) to employees in tech companies for referral opportunities. Secondly, Kazuki Nakayashiki, cofounder of Glasp, who are building a social web highlighter for active readers and writers.

Want to earn a shout-out too? Refer the Startup Bulletin to your network:

➡️ 10 referrals =  shout out to our 16k subscribers 

➡️ 25 referrals = a 30 min 1-2-1 with our team

---

🛠 Six data science methods—and how they really work

by Ali Kokaz, Head of Data Science at Founders Factory

AI or machine learning is often used as a catch-all term to describe a range of data science capabilities. What it’s actually referring to is a broad range of methods, tools, and techniques that all analyse and transform data in very different ways. So what are we actually talking about when we talk about data? 

Here’s my non-exhaustive list of data science tools that can transform startups, and examples of how they’re deployed in the real world:

1. Predictive analytics

Usually what people think of when they talk about “machine learning”—using algorithms that can learn patterns in order to predict something. In the $277 trillion property industry, regression algorithms are used to drive pricing strategies. It’s also popular in fraud management, where machine learning can identify both genuine transactions that have been declined, as well as undetected real fraud (amounting to around $12 billion saved).  

2. Recommender systems

Paralysed by choice? AI has your back. Recommender systems are used to recommend relevant products by understanding product characteristics or previous user patterns. Netflix (or other entertainment platforms) are king at this—Netflix’s recommendation engine is estimated to drive around 80% of what you watch. It’s the same with ecommerce: around 35% of what customers spend on Amazon is due to recommendations.   

3. Computer vision

What if instead of building complex visual models of real world environments, instead we built a high capability system that can understand and process visual content? This is exactly the question Tesla asked, when opting for a computer-vision system for their autonomous vehicles, over the popular LIDAR system used by most companies.

Computer vision’s application stretches beyond just AVs: it can include object detection, counting, classification, recognition, image/video processing, tuning and creation. 

4. Natural language processing (NLP)

Ever wondered how Google knows what you want to search for, almost better than you? The answer is natural language processing (NLP), used for understanding meaning and patterns in speech or text. This powers tools like Siri and Alexa. 

This has become a popular tool among hedge funds, who use sentiment analysis (a form of NLP) to take advantage of sentiment or opinion to predict financial markets. 

5. Clustering & segmentation

A popular tool to discover patterns and groupings in large quantities of data. This is key in market research or customer understanding, allowing you to find broader patterns across large groups of people to inform your decisions. 

It’s also become a popular tool in sport—the renowned ‘Moneyball’ concept uses clustering to find patterns to help inform recruitment decisions. 

6. Time series

Often known as ‘forecasting’—methods or algorithms that specialise in analysing data that evolves over time, or are inherently time focused. We became very acquainted with this during the COVID-19 pandemic, when time series forecasting models were used to map out how waves of infection might pan out. It’s more commonly used in commerce, where algorithms are used to build stock forecasting to reduce waste. 

Understanding how data science can transform your startup and your product starts with analysing the right tool that works for what you’re doing. But this is only the start—evaluating the level at which you’re currently operating (see the pyramid below), and knowing how to improve, will help you go from an organisation based on opinions, to one based on data-driven decisions. 

Read Ali Kokaz’s full guide on Making Data Science Work for Your Startup 

---

📚 Our top recommended reads

• Decentralization (Not Boring)—investor Packy McCormick presents his ‘theory of everything’, on a trend that he’s saying not just in finance but in all walks of life

• The creative legacy of Gifs: Past, present, and future (It’s Nice That)—a deep dive on the viral media format that is the scourge/saviour of social media

• Generative AI is here to stay, but so are artists (Maddyness)—origins of the topic that’s on everyone’s lips, and what it really means for creatives

• Narratives (Stratechery)—finding threads in two of the month’s most popular stories: Elon Musk/Twitter  & Sam Bankman-Fried/FTX

---

💸  News from the Founders Factory portfolio

• Worldr announced their $11m seed round, led by Molten Ventures, with continued support from IQ Capital and Playfair Capital. Founder Max Buchan is building a zero trust architecture for workplace communications tools like Slack and Microsoft Teams

• NFT gaming platform Eterlast officially launched, raising $4.5m. They’ve already launched gaming experiences for boxing and rugby

• We hosted our inaugural ff3 community event—’Web3: The End of Business-as-Usual’. This featured a talk from founder/author GC Cooke, as well as Web3 expert panel

• Blueprint announced their partnership with marketing unicorn Klaviyo, helping customers tap into their SMS marketing tool across the EMEA region 

• In Founders Forum’s 16 ClimateTech Startups to Watch, four of our portfolio companies featured—Materials Nexus, Solivus, Matter, and Metalchemy

---

See you next month 👋

Interested in reading more of the same insights? Check out the Founders Factory blog, and previous newsletters. Want to learn more about Web3? Read & subscribe to ff3