Xero's product does make money - but only just
Xero, the company that has led the charge (at least outside the US) in disrupting business management and accounting services, recently announced their results for the 12 months to 31 March 2016.
From an accounting viewpoint, Xero had a net loss of $82.5m, compared to a net loss of $69.5m the year before.
This begs the question on how this can equate to a product that makes money. That's some pretty good creative accounting!
The answer to this outlines the fundamental differences between the technology / subscription based business models vs the classic industrial age business models built around the manufacturing process. Xero also does a lot of additional disclosures around their business model which makes this sort of analysis possible.
Under the accounting rules, Xero did indeed lose $82.5m in the 12 months to 31 March 2016. For that reason, any time I refer to profit this is in line with the accounting rules. When I talk about how much money the product makes I will use the term "yield" instead.
To start with, let's look at the costs in their financial statements:
From an accounting viewpoint, their costs were $341m for the year. Cost of revenue is similar to "cost of goods sold" in manufacturing terms.
Those costs include the transfer of cash spent on product development to the balance sheet (known as capitalisation). This creates an intangible asset that then needs to be amortised, which is a lot like depreciation on property assets.
They also include the cost of share options issued, normally to employees. These need to be counted as an expense item however they are not cash related.
Once we adjust for these two, we have a total spend of $294m for the 12 month period.
Next, we need to look at their business model metrics (page 19):
The average monthly revenue / customer is $30. Given the total life time value is $2,103, that means a customer on average stays for 70.1 months ($2,103 / $30), or 5 years & 10 months.
That is a long time to keep a customer, who pays you every month, without fail. This gets to the first point of the underlying value of technology businesses, especially those with recurring revenue models. Xero's product is particularly "sticky" as it's a core part of the tax compliance process for small business, which is something you have to do as a business owner.
The average churn rate is calculated at 1.4%, being 1 / 70 months. What this means is 1.4% of their customer base churns each month. The lower your churn rate, the longer you keep the customer and the more valuable they become when you look at product "yield".
It is taking them 14.5 months to return their cost of customer acquisition (sales & marketing activities to acquire a customer), hence their cost of acquisition per new customer ("CAC") is $435. More on this later.
Next, we need to look at the number of customers they have:
Their accounts disclosed total new customers added of 242K. The active customers / month is an estimate based on an average from the beginning to end of the year.
The cost of customer acquisition is the largest and most underestimate bill in any technology company that has reached scaling & growth stage. In this case, total spend on acquisitions is 242K customers * $435 = $105m. That is 36% of their total spend of $294m, it's a big bill.
The table below is an estimate of where the $294m was spent.
If you're going to keep customers for an average of 6 years, focusing on retention is essential. However, the cost of retention is very, very hidden in the profit & loss statement. I can't calculate it directly from Xero's results however have inferred it from the following two areas:
- total spend on sales and marketing is $145m. The remainder not related to acquisition is therefore related to retention
- the portion of product development not capitalised will relate in part to research that can't be capitalised under the accounting rules and bug fixes / small features. Product updates are a key retention strategy.
As you can see, customer acquisition has the biggest price tag on it, followed by product development.
In technology companies, product development is a sunk cost - you build once and deliver an infinite number of times. It's a bit like a building you rent out, the rental creates a yield on the sunk cost of building the property.
The difference to property is the fact that in technology, you build once and deliver an infinite number of times. Your reach is only limited by the number of customers you can access via whichever channels you use.
In my calculations, I have $71m allocated to product development as a sunk cost. This is an estimate, it is not something you can directly pull from their financial statements.
The last cost category is the cost of "running a business". This is finance, HR, office rent & in Xero's case, the cost of being a publically listed company.
Based on the average number of active customers of 596,000, we can calculate the annual amount & thus the average amount spent per month on per customer on delivery, retention and running a business.
This now allows us to put together the "unit economics" that determines the product yield on the sunk cost of developing the product:
The yield is the life time value, less cost of acquisition, less the other monthly costs * 70 months. Over a 70 month period, this customer will yield $514, or 24% of revenue.
(It's important to remember this is an estimate - I may have overestimated sunk product development costs which would reduce this number.)
Based on this calculation method, Xero does make money. The yield % rate, however, is very thin. In many tech models this would be above 40%.
Why is this an issue? Because it doesn't take very much to change the yield rate. For example, let's say the cost of delivery doubled. How would this happen? A lot of these costs for Xero are the costs of the feeds from the banks. Let's say they find a way to increase the cost of those:
Is this case, the cost of delivery has doubled which has taken the yield rate down to effectively breakeven, hence no yield on that sunk cost of product development.
A second example is them having to drop their price to $20 / customer / month in order to protect their customer base from a customer. There is a reason Quickbooks Online announced a very low cost alternative in Australia:
In this case their yield is negative, hence they would be losing money for every new customer they take on. That's not pretty!
Xero's strategy is fairly clear. They are going for the highest acquisition rate they possibly can allow them to protect their monthly subscription price & to give them leverage with the cost of acquiring data from the banks.
The cashflow potential of this model starts making sense when you look at their cost base for 1.1m users, assuming they acquire 250K users in that 12 months period.
In this scenario, their revenue is also $396m, hence their cashflow breakeven point is approx 1.1m users.
At 1.5m users they have a surplus of $65m via this calculation method.
Google has used a similar approach and now have $73B in the bank.
There's a reason why Xero's market cap is approx $2B, which is comparable to MYOB's market cap when they IPO'd last year, for a company that was substantially more profitable when you look at the accounting measures of profit.