Building a Bottom Up Business Model for Your Startup
Building a Bottom Up Business Model for Your Startup
Do you understand the key levers that drive your business? (#30)
TAM. SAM. SOM.
They’re all important acronyms.
But they’re also nonsense (especially for early stage startups.)
TAM = Total Addressable Market
SAM = Serviceable Addressable Market
SOM = Serviceable Obtainable Market
I’ve talked about these in the past, which you can read below:
Most startups come up with their market size via a top-down method.
You research the market and often land on the “highest level market” you possibly can. For example, you might claim you’re part of the MarTech market (which is ~$330-$500B) and use that or some variation of that as your TAM.
Oh boy.
It’s too high level and absurd, because MarTech is a massive, convoluted bundle of many things. You have to niche down from there. But even a sub-category of a giant market like MarTech is often too big. And yet, I still see every pitch deck with a SOM-SAM-TAM combination…and I skip right by.
It’s only when I see a bottom-up business model that I’m intrigued.
What is a Bottom-Up Business Model?
A bottom-up business model doesn’t focus on defining the market. Instead it focuses on how your startup will get initial traction and scale. It’s built on a host of underlying assumptions that are actually important to understand, because they reveal a lot about how you think about your business.
Note: No one actually believes your projections
Every startup presents a 3-5 year plan demonstrating amazing hockey stick growth. Oh that beautiful “up and to the right” curve. It’s irresistible. It’s also complete B.S.
Investors know your projections aren’t going to work out as presented, but that’s OK. If they’re done in a bottom-up way with transparent assumptions, your business model, traction and growth plans become worthwhile discussion points.
How Do You Build a Bottom-up Business Model?
First, you need to know what business you’re in. I know that might sound obvious, but it’s important.
Are you B2B, B2C, B2B2C or something else?
If you’re B2B are you targeting small businesses, big enterprises, somewhere in the middle?
How are you making money? Are you charging right away or not?
Freemium? Monthly subscription? Donations/Tips? Advertising? Etc.
Is it a marketplace? If so, are you monetizing transactions and/or something else?
The list of questions here is quite extensive. You have to understand the underpinnings of your business model and how you’ll make money.
Note: IMO, a business model isn’t only how you monetize, it’s the whole damn thing. It’s how you acquire and activate users/customers. It’s how you engage and retain them, because that’s usually what drives monetization. It’s how you up-sell them. Etc. Don’t think of the business model narrowly around how you make money—think about everything that happens before and after you make money.
A great way to do this is to map your business:
The map describes the whole business, including all of the interactions or touch points with users/customers, and the pieces of the business that you haven’t yet developed (i.e. if you’re early stage you might not have a robust customer success function, or sales team, etc.)
In the post above, I share an example of a SaaS Business Model from Lean Analytics:
That’s a business map—a systems diagram that describes how the entire business functions; i.e. a business model.
So my recommendation is you stop reading this and go map out your business. I’ll be here when you’re ready.
…
…
…
The next thing we need to do is fire up a trusty spreadsheet.
A quick truth about spreadsheets:
Spreadsheets lie.
OK, that’s not true.
The people who use spreadsheets lie.
That’s better.
You can make a spreadsheet do pretty much anything you want (with numbers.) Is your recurring revenue not scaling quickly enough? Lower that pesky churn number. Need to grow a bit faster? Increase that free to trial conversion number. Boom!
The key to using a spreadsheet properly is be very transparent about all of the assumptions you have, so that you can figure out where the sensitivity in your model lies. Even if you’re stretching the truth a bit, it’s easy to adjust the assumptions (especially as you learn new things) and build increasing confidence in what you’re doing.
When you first start building a business model bottom-up, think of it like a “back of the napkin” plan. Don’t try and build the fanciest, most comprehensive P&L. Just start experimenting with the key assumptions and metrics that’ll drive your business.
If you’ve got the systems diagram, you have the basic user acquisition and activation flow—you know how you’re supposed to acquire users/customers, get them signed up (and possibly paying), and turn them into happy customers. So your initial assumptions get built on that.
Here’s a simple example for an enterprise B2B startup:
This software company is selling something for $10k annually. (You could break this down to a monthly number too.)
Based on their top-down market sizing they figured out the total number of potential customers (Y) and through their own knowledge of the market they have an estimate on how many licenses each customer will buy, leading to 1.65 licenses/customer
They assume the sales cycle takes 2 months and they’ll convert 10% of leads into sales (mostly through an outbound approach)
And you can see the growth rate % estimates as well, which decline over time because they’re a % of a higher number. It’s easier to have a high % growth rate early when it’s based off a small number of customers.
Churn is set annually, because this company is only contemplating annual licenses; but you could break this down monthly.
Once you’ve plugged in numbers you think are reasonable, this quick set of variables tell us three important things:
How many customers will I have each year?
How much revenue will I generate yearly?
How many leads do I need to get in order to achieve those targets?
The goal isn’t to get this perfect but to stress test the assumptions.
Do these numbers look insane? Are they even remotely achievable? What would it cost to get there?
In this case, these numbers look a bit aggressive. $12M in Y3 is a classic hockey stick. 6,246 leads in Y3 means getting 520 leads/month, which is ~17/day.
This is the type of math and work that I enjoy doing with founders. As an investor it’s interesting to challenge and discuss these numbers. Again, it’s not because I’m trying to prove a founder wrong, but I want to validate that there’s reasonable thought that’s gone into building the model. This is drastically different from a top-down model where there’s almost no meaningful math that’s done—it’s just “Imagine if we got X% of this gigantic market! Woohoo!”
You can (and should) make this model more sophisticated. But a quick exercise like this is already very interesting and telling.
Let’s look at another example. Below is a B2C play that focuses on converting people into users on a website and then upselling them to a subscription service.
This is definitely more complicated, but it was still fairly easy to put together. We just had to think through the customer journey:
People will come to the website (let’s worry about how they get there later)
They’ll convert into free users
Some of these free users will churn
Some will then convert into paying customers
Some of these paying users will churn
The numbers in blue are where you input data; everything else is calculated. So when you look at this model you have to ask:
How many users can we drive to the website initially? (Again, there’s more complexity here around how to do that and which channels are effective)
Is the “conversion to user” number realistic?
Will people actually pay $12/month? What happens if we could get them to pay more? What happens if they’ll only pay $6/month?
Can we improve churn faster than we’re proposing?
Etc.
Unlike the B2B enterprise example many of these numbers are fairly conservative. And in some cases we have evidence for why we’re using the numbers; i.e. benchmarks we’ve found from other places, or from tests that were previously run before the model was built. When you roll this out over three years, here’s what you get:
There’s still a hockey stick on revenue, but it’s not a terribly exciting business. And ultimately, the biggest lever is the number of site visitors. After three years, this website is only generating ~700,000 site visitors, which simply isn’t enough to grow the business fast enough. Even if you lower churn significantly it doesn’t have an enormous impact because the total number of subscribers is still quite low.
But look what happens when you 2x the Monthly Site Traffic Growth Rate (from 15%, 5% and 5% to 30%, 10% and 10%):
Holy hockey stick Batman. 📈
You need 6.6M site visitors in Y3 though. 😃
This model is still fairly basic. For example, if this is a mobile app that requires a download (from the App Store or Google Play Store) then you’ll need to include that step for calculating users. And you might want to include an activation step as well—ultimately you can model out the entire customer acquisition process to paint a more accurate picture and figure out where the choke points lie. But initially most of that is unnecessary to quickly stress test your core assumptions.
Dealing with Costs
A business model isn’t complete without figuring out the cost side of the equation. Initially this isn’t a priority, because we want to focus on the top-line growth potential of the business. But at the end of the day, things cost money. 💸
My starting point on the cost-side is to figure out Cost of Acquisition (CAC) because it’s directly tied to my ability to acquire users/customers. So it flows logically into the model.
Let’s say you’re using a channel such as Facebook, Instagram or Google AdWords to attract customers.
You include an assumption on the Click-through Rate (CTR) and Cost Per Click (CPC)
You can also include an assumption on what % of traffic will come from these sources
Now you know how much traffic these channels have to drive to hit your website traffic targets (in order for the rest of the model to work); and by extension you know how much it’ll cost
If you need 5B people to see your ads, to drive a small % of them to your website, and it costs you millions to do it, you’re in trouble. So stress testing CAC + channel traffic early on is a good idea.
There are of course, many other costs to consider. The biggest cost is usually humans. The most interesting part is stress testing the number and type of employees that will be needed as your business grows.
Let’s go back to our B2B enterprise example for a moment. I mentioned that most of those leads are intended to be outbound. We need ~17 leads/day at scale. How many salespeople are needed to execute on that? How many calls have to be made? How many outbound emails have to be sent? The numbers probably start to break down…
Finding Benchmarks
Most of the numbers I shared above are fabricated. At least partially. 😆 You don’t want to build a bottom-up model with completely made-up numbers, because the minute someone asks you a clarifying question you’ll panic and run out of the room (or look foolish for not having some degree of confidence in them.) Or, you might defend the numbers aggressively and look equally foolish for doing that.
No one expects you to have perfect numbers—that’s why we call them assumptions. But assumptions need a basis in reality, otherwise they’re fantasy.
So how can you find benchmarks to use?
Most recently I’ve been using Perplexity. It’s a powerful AI-based search engine that generates good, verified results. Here are a few query examples I’ve run in the past that gave me interesting info:
What is the typical conversion on an outbound cold call for enterprise B2B software?
What’s the typical conversion rate from free to paid on a financial mobile app?
What’s the typical conversion rate on a website that gets people to download an app in the App Store in the health and wellness industry?
You can also find a lot of benchmarks shared on Twitter or LinkedIn, and more companies are sharing their numbers as well (on blogs, newsletters, etc.) It still takes a lot of research, but it’s possible to find attributable benchmarks that you can plug into your model.
Comparing a Bottom-Up and Top-Down Model
Once you’ve built a reasonable bottom-up model, you in effect have your Serviceable Obtainable Market (SOM), which is the portion of the SAM that can be realistically captured and served. The SAM is the complete market you could get with your (existing) product and/or service.
How do the bottom-up and top-down SOMs compare? Are they reasonably close? If so that’s a good thing, because it’ll give you more confidence in the bottom-up numbers.
Does the SOM you calculated bottom-up, say within 3-5 years, look reasonable compared to the top-down SAM and TAM you researched? While you should never say, “All we need is 1% of this market and we’re going to win BIG!” it is interesting to see what market share your bottom-up SOM represents compared to the SAM or TAM that you researched. If that percentage is low, it should increase your confidence in the bottom-up numbers, which is a good thing.
Ultimately when going through a market sizing exercise, you should do both a top-down and bottom-up version. Focus on the bottom-up one more than the top-down, but use a comparison of the bottom-up to the top-down to see if you’re heading in a reasonable direction.
Get copies of the business models 👇
I’ve put together two sample business models for you, which you can download for free.
They’re pretty simple, but hopefully they’re helpful.
Check out the Focused Chaos Referral System!
Hey folks! Thanks for reading this latest newsletter post. Substack recently released a Referral System, which I’m now using.
All you have to do is click the SHARE button below (and then share this post via email, social, etc.):
As you refer people to Focused Chaos, you can earn rewards that I’ve made available:
First reward: After 3 referrals you’ll receive an email with a link to download the Analytics Lessons Learned e-book from Lean Analytics, which includes 13 case studies.
Second reward: After 10 referrals you’ll receive an email with an option to send me your pitch deck for feedback.
Third reward: After 25 referrals you’ll be able to book a 30-minute chat with me on any topic you’d like.
It means a great deal to me when people share Focused Chaos and provide feedback. Thank you in advance!