What Went Wrong with Snap, Netflix and Uber? (The Economist)

I’m struggling writing my chapter about why the Hockeystick growth curve is a mirage in today’s post-pandemic unpredictable world. Then I read this. Basically, fast-growth business models “all turn out to face the same main pitfalls: a misplaced faith in network effects, low barriers to entry and a dependence on someone else’s platform.”

1. Network effects are real but have limits: building a network in one city does not necessarily mean better service in another. Incremental capacity does not translate into incremental value (what’s the added value of one more Indian restaurant on Doordash or one more driver on Uber?);
2. Low barriers to entry: On one hand, the tech stack is cheaper and cheaper, allowing copycats to quickly steal traffic. On the other, customer acquisition and retention costs climb because of competition;
3. Dependence on someone else’s platform: Reliance on the gig economy means loss of control over quality and customer satisfaction. Or you may be dependent on another platform, for example if your business relies on Facebook, Spotify, AirBnB, or the iOS/Android duopoly.

The bottom line is that your business model must evolve with the stage you are at. The strategies that get you in the game need to change as you scale, then change again as you achieve your first, second and third levels of consolidation.

Link:: https://www.economist.com/business/2022/10/31/what-went-wrong-with-snap-netflix-and-uber

Data Quality and Quantity for an Industrial AI Project

[Panel Question #2: How can you ensure that you have the correct data in sufficient quality and quantity to carry out an AI project]

Data is the lifeblood of an AI project. In a perfect world, an algorithm would train on all the data that ever existed. However, it still would not be a perfectly trained algorithm.

This question is a hot research topic. The answer depends on several factors, such as the type of classifier, the number of weights, and the data quality. If there is insufficient quantity and quality of data, the model will not achieve the desired accuracy. If there is too much data, the model will tend to overfit and not be accurate in edge situations. This is the most problematic because these are the cases where you most want accurate recommendations.

Continue reading

Transferring Knowledge from Academia to Industry

Photo by Zdeněk Macháček on Unsplash

[Panel Question 1: How to make accessible the knowledge of AI researchers and experts to Quebec’s industries and SMEs?]

Transferring knowledge from academia to industry is a long-existing problem because each operates in separate bubbles. Academia wants to solve cool technical challenges, while businesses want to maximize profit. 

Once a business reaches the point where it is making a profit from a process, this becomes a core capability, which can only be changed by risking the business’s cashflow. Of course, this is the last thing they want. This resistance is the main obstacle to corporate innovation.

There are three issues that come up when considering how to make accessible the knowledge of AI researchers and experts to industry and SMEs:

– The company’s willingness to innovate;

– Maintaining the company’s core vs innovating at the edge; and

– The innovation value chain, or how new practices diffuse through the organization.

Continue reading

Preparing for a Series of Panels

Dima Pechurin from Unsplash https://unsplash.com/photos/JUbjYFvCv00

As more people go back to the office, including me, I’m noticing a sharp increase in the number of in-person events this fall. 

My day job is as a technology advisor for a federal government scientific agency. I provide coaching and consulting services to tech startups, as well as funding support. Although I have a background as an R&D engineer, my hands-on technology knowledge is light-years behind what the new generation is working with today. However, I bring my project management and business development skills, which new founders severely lack.

My office is in a research institute specializing in data science and artificial intelligence. I know enough about these topics to understand what entrepreneurs present to me. And that’s about it.

Continue reading

What Is Your Profit Proposition?

Photo by micheile dot com on Unsplash https://unsplash.com/photos/SoT4-mZhyhE

Your business model describes how you create, deliver and capture value. Much time and effort are spent defining the value proposition: the added utility you create for your customer by solving their job-to-be-done. If the job is important enough and your solution effectively enhances their gains or alleviates their pains, your customer will say “yes” to investing their money and attention with you.

However, founders flounder when considering how to capture this added value through revenue streams. The primary consideration becomes to enhance recurring revenue metrics through subscriptions, commissions and markups. Pricing seeks to underbid the alternatives or is set arbitrarily, hoping the customer will pay with minimal protest. Strategies are established without understanding how the customer perceives the value you offer. This means that you leave cash on the table by not recognizing the real value you add.

Before setting prices, understand your Profit Proposition

Before considering your pricing structure, you need to understand your Profit Proposition. If your Value Proposition is the benefit you offer your customer, then the Profit Proposition is how your customer assesses the utility of your Value Proposition. Understanding your Profit Proposition provides you with a solid foundation to establish a pricing strategy that supports your margin, or the difference between what you charge for your offer and how much it costs you (fixed, variable and marginal costs).

You cannot price effectively if you do not understand how you will generate profit. SaaS companies often have very weak profit propositions because they price arbitrarily, focusing on quickly gaining a customer base to leverage future funding rounds, even when losing money on each transaction. You cannot lose money on each trade, hoping to earn money at scale.

Continue reading

Managing Uncertainty is H.A.R.D.

Image by Maarten van den Heuvel via Unsplash.com

Founders bet on success. That is their job. However, this is a poor bet because 90% of startups fail to survive beyond three years. This is because so much needs to go right: funding, people, customers, product development, timing. If only one aspect goes wrong, the whole venture can collapse. 

The key to successful scaling is managing uncertainty while maximizing execution. Unfortunately, managing uncertainty is problematic because scaling is full of unknowns. And it is the unknowns you must fear – then overcome by making them visible and actionable.

The most popular type of risk analysis is SWOT: Strengths, Weaknesses, Opportunities and Threats. Although SWOT helps to get founders to look outside their bubble, it is not an adequate risk assessment tool for two reasons:

  1. SWOT tends to reinforce biases. SWOT is based on what you know or what you think you know. Whether because of optimism, hubris or ego, founders put too much faith in the positive: their Strengths and their Opportunities. They do not pay enough attention to their Weaknesses and their Threats.
  2. SWOT is not actionable. It does not lend itself to making decisions, assessing the level of risk, the impact of weaknesses or threats, or assigning responsibilities to reduce risk.

Uncertainty happens because you are moving outside of your comfort zone into a state where everything is unknown. You have never experienced this before. The more you are outside your zone, the more uncertainty you will experience. If you are a technical founder, business aspects will cause you to crash. If you are a business founder, it will undoubtedly be the technical unknowns that will cause you to stumble and fail. No amount of training can prepare you because the threat comes not from what you know but from what you do not.

Continue reading

Let’s Kill The Myth of Startup

Why do we teach founders that all startups must scale exponentially? Who decided that the hockey-stick growth curve is the only viable outcome for an emerging venture?

Photo by Annie Spratt on Unsplash

The era in which an app can be built in a basement and rapidly scale to millions of users with little marginal costs is long over. To achieve hockey-stick growth requires massive amounts of cash and even more good luck. I doubt that the basement-to-riches growth curve is even possible in today’s reality. The rules of the game are not set up to favor the startup entrepreneur. 

Continue reading

Startups need to consider risk from the start

Cybersecurity is more critical than ever. I see too many startups who rely on cloud computing and make no mention of security in their plans or budgets.

Issues such as data security, responsible AI, environmental impact, and hiring diversity and equity must be addressed from the beginning. Retro-engineering these considerations later are more expensive and can put your whole venture at risk.

Nvidia was hit by a hacking collective who compromised their corporate e-mail systems. Interestingly, the hackers don’t want cash. They want the company to change the code in their video cards to enable cryptocurrency processing. Lesson learned: hackers can force you to make unwanted changes to your product.

About the NVidia breach:


Comment by Andrew Orlowski in The Telegraph (7 March 2022)


(Cross-posted to my LinkedIn page)

When metrics are dangerous

When are metrics dangerous? When they lead to making decisions that put your venture in peril.

I was talking with a founder who was enthusiastic about his progress. Revenue up 300% over last year! Team size up by 500%! He committed to raising $1.5M in a seed round with numbers like that.

Then I started to dig deeper. What was last year’s revenue? $100k. What is this year’s expected revenue? $300k. Yes, he tripled his revenue. But is this the signal to invest all of his energies and bet the company on a seed raise?

Photo by Arie Wubben on Unsplash

Metrics are beneficial and also very dangerous. The purpose of metrics is to help sort signals from noise. If you look at the wrong metric, you will miss critical signals. Or worse, interpret false signals that lead you to lose everything.

When revenue is below $1M, the percentage revenue increase is meaningless. Indeed, all metrics at this very early stage are noise because they have no reference and no history.

At this initial stage, the founder is at the very beginning of H1 (Credibility). The focus of H1 is finding problem-solution fit, and specifically, the value innovation. Problem-solution fit focuses on identifying what specific job-to-be-done you propose to solve in a way that adds 10x value to the target user. This means mapping each step of the interaction to see the one subcomponent that is “broken”. You want to understand precisely why a buyer chooses you – what makes your solution more valuable than their other options, including doing nothing.

Continue reading

Can the idea scale?

A recent Freakonomics podcast featured John List, author of The Voltage Effect, asking, “why do most ideas fail to scale”?  

The core idea is “how to achieve voltage – the power to take an idea from small to large scale” robustly and sustainably.  He answers that successful scaling is not the work of “one great man” but rather how you coordinate large numbers of people and stakeholders to collaborate effectively.

A specific example I found interesting was his work with Uber. Their innovation was not necessarily the technology, even though building a quality platform that replaced human taxi dispatchers is challenging but does not require new tech. The innovation was their business model: inventing a way to externalize fixed workforce costs – the drivers – radically reducing the firm’s cost structure and making scaling much more straightforward. It also could skirt around regulators and licensing authorities that would have slowed its growth. 

Continue reading