In the next few weeks I’ll be working through a modelling exercise to illustrate how this technique can help make product decisions/tradeoffs. In the first installment the goal is taking it from square one and build a static 5 year P/L for a product.
While the main audience is product managers/owners I think most scrum masters would also benefit from understanding this. If you are trained in finance you are likely about to cringe at some of the simplifications I’m making here, but keep in mind the goal is not to deliver the financially 100% accurate model. We’re looking to provide a good enough tool for people to understand the impact and help decision making.
While you could certainly take a stab and put in some number of sales you are going to make each year, understanding where your sales come from will allow you to drive your results in a more meaningful way.
Throughout the series try to focus on techniques, assumptions and train of thought rather than the actual numbers. I’m using MS Excel for everything, and will make a spreadsheet available with each post. For this first part equivalents of MS Excel (google drive, openoffice, … ) will probably work too, though I don’t guarantee that will still be true later in the series. Feel free to leave comments or get in touch if you want more details on certain areas.
The base setup
The entire series is based on the idea of modelling a simple webshop selling a digital yearly subscription. So don’t expect any complexity around multi product setups, complex sales journeys, affiliate deals, … . Think of it as a one pager with a big “buy now” button followed by a credit card transaction.
I’ve modeled 5 years of this product’s life, but obviously you can just as well pretend that those are 5 quarters or 5 decades or any other time interval you prefer. As long as you update the assumptions to fit with your time interval the structure holds up.
There’s a matching excel spreadsheet you can download here and which you should probably keep an eye on while reading this post. In the excel the convention is that cells with a blue background are assumptions you enter into the model.
Part 1: Modelling customers
The first thing to understand in this model is the customers, which I’ve modelled in 3 stages: Real Contacts, Real Shared pool and Valid contacts
New contacts x Quality % = “Real contacts”
The base assumption is that every year you make 100,000 new contacts. I intentionally left this a bit vague, but it could be anything from speaking at events to buying a mailing list, cold calling, … . The first step is to assign some form of quality to those contacts. If you’re buying a mailing list some addresses are going to bounce, if you talk to people directly some of them will just not be interested at all, etc. I assumed here that out of your 100,000 contacts 80% are actually real people who might be interested.
“Real contacts” + growth from sharing = “Real Shared pool”
The next step is to think about word of mouth, or Facebook shares, tweets and other forms of contact expansion. My assumption here is that 1 in 5 of your valid contacts shares your information. It’s likely going to be more like 1 person sharing it with 10 people and 49 not sharing at all but you get the point, for every 5 real contacts you make you reach one “for free”.
This leads us to the next element, overlap.
“Real Shared pool” - (Overlap in the list adjusted for compound growth) = “Valid Contacts”
I’ve split overlap in 2 distinct parts. The first is “natural” overlap in your contacts. If you speak on events you might reach to the same person multiple times, they might be subscribed to your email list with multiple addresses, 2 people might share with the same person, … . On your total number of contacts your overlap will grow over time as you reach more people. For instance if you were speaking on the same industry event in your city for 2 years in a row, chances are you’ll reach largely the same people. And if you did that again the year after the fraction of new people would probably be even lower. In this model I assumed you are not doing anything spectacular in terms of finding new markets or reaching distinctively different audiences but simply operate at a 25% compounding growth overlap. So if you wanted to model this you start with the initial overlap 25% and than in year 2 you add 25% to that (31.25%). The year after that it’s another 25% on top of that (39.06%) and so on. You quickly start to see where this is heading.
All this adds up to the number of valid new contacts you can make every year, people who might be potentially interested in becoming customers. As you can see the yield on your 100,000 contacts goes down year on year under these assumptions because of the increasing overlap.
If you assume an average subscription % over the year it’s pretty easy to see how many new customers you’ll get every year.
From those, adding in a churn rate (~ People who don’t renew their subscription) you can also easily get your recurring customers.
Adding up new and recurring customers gets you a total number of customers for the year (obviously in year 1 you have 0 recurring customers)
Part 2: Building a simple P&L
There are a fair few accounting details when it comes to building accurate P&L but for the sake of this article I’m keeping it simple. All costs you make are fully attributed to the project in the year in which they are incurred. Feel free to go wild here and include loans, interest payments, depreciation of assets, different payment structures for employees, … . The assumptions I made are pretty straightforward but will get you a pretty long way to creating a basic setup.
a P&L really is just (sum of revenue) – (sum of costs) and that’s your profit or loss for the year
sum of revenue
The price (all ex-VAT) of our product for a year is $90 for new customers and $120 for recurring, so the assumption is you run a 25% discount for new subscribers. Think of it as a “first 3 months free” type deal.
So revenue for a year is simply (new price times new customers) plus (recurring price times recurring customers)
sum of costs
Transactions
I’m also assuming there’s a cost per transaction of 2% of the value for handling the creditcard payment. Since it’s a % of the transaction value it doesn’t really matter if people pay $10 every month or pay for the full year in one go. This would obviously be different if you paid a fixed fee per transaction or if you had some sort of agreement that changes the % when you reach certain volumes of transactions.
Signups
For every new contact you acquire (regardless of quality and excluding those “free” ones you get from people sharing) you pay an average cost of $0.10. And for every one who subscribes (new or recurring) you have a $10 cost, maybe you’re sending them a magazine or some welcome gift or so.
Operations
The last few costs are just running your operation. At a base level, you have some staff with a salary cost that increases every year and also need an office and some equipment. You could just as easily put in a cost for development by a 3rd party and a whole lot of other services. Don’t get too bogged down in details when doing this kind of exercise for your product, keep focus on the big ticket items. You are not an accountant who needs precision to the last cent.
Reading this graph is pretty straightforward. When the P/L line is above the horizontal axis you are making a profit for the year.
Part 3: The results
So far we built a very basic, static, model but it’s already showing a lot and you have a base to go play out different scenarios. Fiddle around with the assumptions to see how they impact your P&L.
The first observation is that you are reaching a plateau if you don’t find a fresh pool of customers. Compare the P&L graph to the customer graph and you’ll quickly see the relation.
Try playing with the overlap and overlap growth to see how different levels affect that plateau (i.e. more or less “fresh” contacts in your pool). You could also look at what happens if your user share more. Or imagine what would happen if your churn rate dropped to 1%. Or maybe the contacts you make are 90% quality. You could do similar exercises to understand what happens if you change costs or increase/decrease the price of your product. Try modelling this out to year 10. You’ll notice that by year 8 you have reached your entire contact pool, and by not making new contacts your profit goes down quickly turning to a loss in year 10.
You can’t control new customers, but you can certainly affect the parameters to this model and give yourself a better chance of getting them
The key to this model is that by working from those assumptions you can work out (crudely) what the impact is of changing a few parameters which you typically can control or influence in some way. You can go and focus your effort on those that make most difference or look at historic data and run experiments to understand if your assumptions are realistic. It also means you can just update your sheet as you move along and get real data.
All this is a lot more valuable than just entering a number of sales you expect to make and “work really hard” to get it.
Next time we’ll take a closer look at how to work in variation in conversions.