Congrats, you got one of my financial models. Now you are thinking, how the heck do I forecast my startup in a financial model template? Yup, everyone is going to wonder this, so if you are feeling lost, you’re in common company!
In this blog, I’m going to take you through at a high level the broad steps you should be taking. The most important things to remember, though are:
Figure out your financial goals that you have to hit before you start. Make the model support those goals!
Keep calm and carry on! Don’t aim for perfection at first run. Iterate.
The earlier the stage you are, the less you know. Focus more on the metrics and growth rates you need to tell a fundable story. VCs know your #s are bullshit, so focus on ‘logical and believable’.
Step n°1 |
Start with pen and paper and set your goals for the financial model template
Don’t just start throwing in numbers. You want to work smart not hard. So here’s a big tip, start with pen and paper.
You want to know your answer before you start. A model is not the place to find answers, it’s a place to learn the implications of things.
When I was in M&A, your MD would say ‘this company is worth $500m. Make the model show that.‘ Actually, that’s a lie. They would make you work for days and then they would say that. In M&A it’s all about buying and selling stuff, so valuation is the objective. When you are in startup it’s about your growth rates.
As an investor, I care about two things, the growth rates and size of the numbers. Only then do I want to see what supports those numbers.
If you are super early, then maybe you can get away with a 12-month projection, but for the most part, my opinion is 3 years is the ideal. 5 years are fine, but you have no way of really thinking that far forward. Yes, 3 years is a crap shoot too, but you can sort of get your head around it.
There’s only a 2 year difference, why is that harder? Well, we overestimate what can be done in the short term and drastically underestimate what can be done in a longer term… so the issue is you really want your numbers to start looking crazy.
Step n°2 |
Don’t be fancy. Be messy
Modelling is a process. Don’t start out trying to be perfect. Just get shit done. Chuck in whatever assumptions and then make them better over time. You are effectually triangulating.
How annoying is it to spend a day putting in super accurate numbers only to realise the numbers just don’t work and you need to change your operating model? If you did ballpark numbers real quick, you can see what’s going on and keep triangulating without surprises.
Triangulating is also more fun. You see immediately what’s happening as you make some key assumptions and start thinking about the effects of all your decisions. Only once you think things make sense should you start nerding out and putting in the details.
Step n°3 |
If you couldn’t care less about email marketing then don’t!
My models aren’t for kids. They’re big boys’ toys. If you haven’t bought one of my models and you’re reading this to learn, you might think I’m joking or naive. I’m not. These things are huge.
Let’s pick on say marketing. There are a bunch of sheets on, for example, email and social marketing. Unless you really believe this is your core marketing channel (which it isn’t for most) then you might think spending a second is a second too long.
The whole model is pre-populated with numbers. I’ve looked at benchmarks to find reasonable default assumptions. So if you are looking at email marketing, you could just input the size of your current email list and leave everything else as is. Yeah, once you finish the model you need to know what assumptions ‘you’ made, but that’s it.
But why bother with this email stuff at all? Well, most startups have to buy customers. Paying users suck. Well, how do you decrease your effective CAC? By diluting down with free users. Some ways of doing that are by blogging, organic, emails etc. So a 20% ‘free’ traffic is a 20% lower paid CAC. So instead of $100 a user, you are paying effectively $80. Um, yeah. That’s good. So a little messing about with these sheets and you can argue lower CACs.
Anyway, if you really don’t care about this stuff and want to be basic, then just chuck in zeros. You don’t really have to set everything to zero. You just need to zero out the key drivers. So for emails, that’s your starting database and any conversion rate. If you don’t think super fast, then it is faster to just set every assumption to zero.
Step n°4 |
Chuck in your core numbers which aren’t really going to change
Depending on your stage, there are things you know. If you are later stage then you know a whole lot. Your numbers are not going to be total bullshite. But the earlier you are the more known unknowns there are and generally the harder it is.
But, regardless of stage, I would still do a similar process when building a model whether you are late or early stage. Start with what matters and then deal with stuff that is less important as you go. Frankly, it’s just more fun and rewarding to progressively build a model iteratively and see results.
Step n°5 |
Understand the basic logic
My models are based on the following bottom-up logic:
You do marketing and that creates registered users or trials
Those trials can % convert to paid on a schedule of your choosing
You segment customers into buckets to get averages and model distinct avatars
You make revenue from those customers on a first-time basis. For SaaS that recurs on a monthly or annual basis. For ecommerce there is a first-time purchase
Your customers will churn over time
Some or all customers will repurchase
There are costs of servicing the customers (CoGS)
There are costs of running the business (Staff etc)
All the exceptionals like bad debt are dealt with in the P&L sheets which makes it easier to manipulate numbers without dealing with them in the main logic engines
add_shopping_cartContinue reading for free (70% left)