The power of a subscription revenue model
Synopsys announced Q4 results today and, apart from being very good, they illustrate the fundamental power of taking revenue ratably. Primer on the difference at the end of this post…
Now, going to ratable (subscription) revenue is not a new idea, and both Cadence and Synopsys (the two leaders of the EDA industry) had been flirting with this idea for more than 15 years. Both had done a partial shift, reported what percentage of new business was coming in as subscription vs. term, postured at each other as to which was better, but both were fundamentally still playing with structuring the revenue on big deals to make the quarter. The problem with that approach is eventually it catches up with you and you a) miss the quarter or b) can’t hide the growing percentage of your business that is revenued up front. But for a CEO it takes courage to restructure for the long term, sustainable health of the business and put himself at risk as he does it.
In July 2004 Synopsys missed a quarter and made the brave step to move to almost 100% subscription revenue. It was tough. Investors punished the stock and management had to cut costs to deal with the reduced revenue. But today the courage paid off and Synopsys not only beat expectations but (I speculate) probably has fantastic visibility going forward because of the smoothness of the model.
Consider Cadence in contrast. These two vendors run neck and neck in market share, they both have excellent products and loyal customers. But Cadence never took the bold step to restructure its business model to the degree Synopsys did and as a result they don’t have the same visibility. In October 2007, even though Cadence beat street estimates, they reported that their up front license revenue exceeded 50% and so they were downgraded by Raj Seth of Cowen because of reduced visibility and reduced backlog. As a result the stock dropped 15% and has continued to decline in the rough market.
To put FirstRain in context, when I took over the company it had a blend: some perpetual and some subscription business. I’ve done a lot of modeling on this issue over the years and so I made the decision that FirstRain would be 100% subscription – and SaaS (more on what on that later). Subscription is much tougher on cash consumption, but if you are playing to win over the long term it’s worth the initial investment.
My crystal ball prediction: the visibility and stability of the Synopsys business model will cause SNPS to pass CDNS in market cap in 2008 – although they will continue neck and neck in true – seat count based – market share.
Primer (skip if you know this stuff)
In software there are two basic business models:
License revenue – this is typically a perpetual license to use a piece of software; the revenue for the product license is taken up front and then the customer pays maintenance over time. So, for a $100,000 software product the customer will pay $100,000 up front and then maintenance thereafter. The revenue treatment in this case is:
Quarter 1: $100,000
Quarter 2-n: $0
Sometimes you see a slight innovation from this called a Term license where the license lasts 3 or 5 years and then has to be repurchased but then revenue treatment is the same. Term is typical in EDA.
Subscription revenue – this is typically a 1 to 3 year license to use the software, paid annually or quarterly; the revenue is taken smoothly over the period of the license. For a $100,000 3 year license the revenue would be
Quarters 1 – 12: $8,333 per quarter
Over time, as the business grows, the subscription model provides smooth, predictable revenue growth – plus you tend not to build the P&L too far ahead of revenue because of the cash impact. In contrast, with license revenue, you are always vulnerable to one deal slip blowing the quarter – and so companies discount, play with terms and do all kinds of unnatural acts to get a deal in, instead of allowing the business to build naturally.