Questions appear to being asked in VC circles about sky-high Software-as-a-Service company valuations – including one suggestion i’ve seen that it should be based on customer acquisition cost (something I think is insane – acquisition costs are far higher than i’d ever feel comfortable with at the moment). One lead article is this one from Andreessen Horowitz (A16Z) – which followed similar content as that presented on their podcast last week.
There are a couple of observations here. One is that they have the ‘normal’ Enterprise software business model misrepresented. If a new license costs $1000, then subsequent years maintenance is typically in the 20-23% of license cost range; the average life of a licensed product is reckoned to be around 5 years. My own analogue for a business ticking along nicely was having license revenue from new licenses, and support revenue from maintenance (ie: 20% of license cost, for 5 years) around balanced. Traditionally, all profit is on the support revenue; most large scale enterprise software vendors, in my experience, assume that the license cost (less the first year maintenance revenue) represents cost of sales. That’s why CA, IBM and Oracle salespeople drive around in nice cars.
You will also find vendors routinely increasing maintenance costs by the retail price index as well.
The other characteristic, for SaaS companies with a “money in this financial year” mindset, is how important it is to garner as many sales as is humanly possible at the start of a year; a sale made in month 1 will give you 12 months of income in the current financial year, whereas the same sale in month 12 will put only 1 months revenue in the current fiscal. That said, you can normally see the benefits scheduled to arrive by looking at the deferred revenue on the income statement.
Done correctly, the cost of sales of a SaaS vendor should be markedly less than that of a licensed software vendor. Largely due to an ability to run free trials (at virtual zero marginal cost) and to allow customers to design in an SaaS product as part of a feasibility study – and to provision immediately if it suits the business need. The same is true of open source software; you don’t pay until you need support turned on for a production application.
There is also a single minded focus to minimise churn. I know when I was running the Individual Customer Unit at Demon (responsible for all Consumer and SME connectivity sales), I donated £68,000 of the marketing budget one month to pay for software that measured the performance of the connectivity customers experienced – from their end. Hence statistics on all connectivity issues were fed back next time a successful connection was made, and as an aggregate over several 10’s of 1000’s of customers, we could isolate and remove root causes – and hence improve the customer experience. There really is no point wasting marketing spend on a service that doesn’t do a great job for it’s existing users, long before you even consider chasing recruitment of new ones.
The cost per customer acquired was £30 each, or £20 nett of churn, for customers who were spending £120/year for our service.
The more interesting development is if someone can finally break the assumption that to sell Enterprise software, you need to waste so much on customer acquisition costs. That’s a rubber price book and golf course game, and I think the future trend to use of Public Cloud Services – when costs will go over a cliff and way down from todays levels – will be it’s death. Instead, much greater focus on customer satisfaction at all times, which is really what it should have been all the way along.
Having been doing my AWS Accreditations today, I have plenty of ideas to simplify things out to fire up adoption in Enterprise clients. Big potential there.