Modeling seed stage growth
How to avoid the usual thing that no one believes anyway.
The problem
We see a lot of early stage company pitch decks. Inevitably, they all contain a slide that looks like this:
No one (authors or readers) believes an early stage company’s ability to predict revenue growth. When we look back at historical company materials (e.g. if a company pitched us a year ago) and compare what happened to what was predicted, the data shows how inaccurate these predictions are.
That’s not a failing of the founders, market or business model; it’s simply impossible. It would be like aliens landing on earth, giving them a thermometer, and 6 hours later asking them to predict long-term climate trends. If neither founders nor VCs believe these predictions, why do almost all decks include this slide? It’s an example of negative transfer: because more mature companies do this, we think this should transfer to early stage companies as well.
It doesn’t. There is not enough data nor ability to predict tipping points in a startup’s life. Tipping points for a startup include when they first find product-market-fit, and when they first find a repeatable sales motion.
What to do instead
To represent this movement across tipping points, we should use a model that looks like a state machine or Markov chain. Here’s a (drastically oversimplified) early stage company representation:
For the sake of brevity and simplicity, let’s ignore a few things (topics for future posts):
The transitions between states are not single moments in time (“Yesterday I didn’t have PMF and today I do”).
It’s not a simple linear path: one often goes upstream to revisit prior states, etc.
The exact “graduation criteria” for moving from one state to another aren’t specified here.
Investors should focus on evaluating the activities startups perform in each state. It’s impossible to predict how long startups will live in each of the states. That’s why “revenue vs. time” forecasts are useless.
We can evaluate the activities that happen in each state, and come up with a point of view on how likely it is that the company will ever advance to the next state. For each of the states in the above diagram, here are some things to evaluate.
Looking for PMF
In this state, the company is a science lab. We should evaluate the quality and speed of their experiments and the company’s learning rate:
How are they running PMF (product market fit) experiments? How many do they ship each week?
How do they efficiently extract as much information from each as possible?
How much does each experiment cost (in time, money, etc)?
What does a writeup / summary from a few sample PMF experiments look like?
Are they getting closer to finding PMF? How do we know?
What is the success criteria in this stage? How do they define PMF, and for whom?
What customer signals do they use (and how do they collect them) to answer the above?
What makes less sense to ask is: “How long until they have PMF?” That’s like asking a scientist: “How long until you make a discovery that will get you a Nobel Prize?”
Additional work for the investor includes: given everything you know about the problem they are tackling, the target customers and their unmet needs, as well as the skills and backgrounds of the founders, how confident are you that the above activities will ultimately result in a post-PMF company? How does that answer vary over different time horizons?
Learning initial sales motion
Once the company has found early product market fit, we need to understand if we’ll be able to build an “atomic unit” of sales. That is, given a fixed (small) investment in sales and marketing, can we turn that into revenue (gross margin)? Key questions in this stage include:
How long does a deal take to close?
Where are new customers coming from?
What is the cost to acquire a new customer (CAC), and their long term value (LTV)?
How happy are these paying customers? How likely are they to stick around, spend more, and recommend the product to their peers?
What is the gross margin for these first customers?
These questions should be answered first for founder led sales, then answered for the first few sales team hires.
Scaling sales motion
Do we believe we can repeat this process and essentially have a knob we can turn up or down, where the company can convert sales and marketing spend into gross margin? This is the stage at which we can start predicting revenue over time.
Things to understand:
How quickly are we able to grow the sales team (account executives (AEs), etc)?
How quickly do new AE cohorts become ROI positive (their payback curves)?
What percentage of new sales hires are able to make quota, exceed quota, and how quickly does it take them to ramp up?
When AEs fail to meet quota, what are the reasons, and what can we do to fix them?
As the number of customers increases, is CAC growing too quickly? What is the CAC payback period and LTV/CAC ratio?
As the company looks at customer cohort data, how is churn trending over time? Is it declining as the product improves? What are the reasons for churn?
As customers stay with the product for longer, do they increase or decrease their average spend with the product?
Evaluating sales organization performance is a long post in and of itself. The point is that only after you’ve seen enough sales cohorts to accurately predict sales scaling does it make sense to start forecasting revenue over time.
Conclusion
Early stage startups have no ability to forecast their revenue growth over time. This comes only after finding product market fit and collecting data on how the sales motion scales. As it is very hard to predict how long startups will live in their early exploratory states, investors should focus on evaluating the quality of activities at the early stage, not ask for false precision on a metric like “revenue growth over time”. Investors should come up with a point of view regarding, given the founding team, market, and resources (including or not including a potential new investment), how likely is it that the company will graduate to the “Scaled Sales Motion” stage?
Each of those states deserves a (long) post by itself. This is meant as a quick overview to convince you to put zero faith in early stage “revenue vs. time” forecasts.
We hope this has been helpful. If you’re a founder and this methodology resonates with you, and you are well on your way to finding product market fit and/or scaling sales, get in touch! I’m alex at moxxie dot vc.