AI: Reversion to the old days - Software as a product (SaaP) [POST-1]
How the AI revolution will force the creation of a new breed of software companies selling products not services
In February of last year (2020) A16z published a stimulating article around the differences they perceive between SaaS and AI, or what is now being referred to as AISaaS [LINK].
As an AI founder, I have reverted back to this article a year later and it prompted me to reconcile my observations and their hypotheses.
In the following 2 part series I will dive below the surface to actually provide the history and business of SaaS and why venture capitalists flock to the model. In the second post we will revisit the article and I will share my thoughts and observations around running an AI business from the trenches.
History of SaaS
Most people forget that SaaS is not a new concept. Time sharing of computer systems has been around since the early days of large mainframe computers of the 1960s. These mainframe time shares were siloed in organizations such as universities, banks, and governments etc.
As the cost computing decreased and personal computers became more prevalent through the 80s, the time sharing industry had to adapt. Local networks (LANs) began becoming more popular and compute was expanding in a non siloed way accentuating the critical limitations of small small/personal computers. Specifically, the limit of hard drive space (HDD) became the saving grace for the distributed computing industry.
By the 1990s, software began to contain a bunch of additional programs included with the original program - commonly referred to bloat or bloatware. Bloat compounded the problem of limited HDD space and led to large costs forcing companies to further adopt distributed hosting of software on centralized clusters.
This trend also paralleled the Dot com boom of the 1990s. The mass adoption of the internet opened the door for distributed compute to go global and it gave rise to the Application Service Provider (ASP) industry. ASP was the first true precursor to SaaS.
Yet, ASP was riddled with technical problems that needed to be solved for wide scale adoption to occur. These issues were primarily technical, such as scalability, speed, distribution, and cost. For example, a company called Mi8 tried to provide e-mail and groupware services based on Lotus Notes. It ran “as slow as running a marathon on crutches” [LINK]. How far we have come.
As an industry, ASP failed, primarily because it was ahead of its time. But, ASP provided companies a blueprint to develop SaaS. In 1998, the company that would morph into the first SaaS company went public: it was called Concur and it sold CD-ROMs for travel software. When the bubble burst, their market cap imploded. They had to cut costs to survive — they stopped selling CDs and start providing their software online… In 2014 SAP bought Concur for $8.3 billion 😊 [LINK]
What about today?
Today the SaaS pendulum is swinging too much the other way. While SaaS was used to make businesses more efficient and reduce bloat, today businesses are suffering from being ‘over SaaSed’. They subscribe to too many SaaS services , which they don’t actually use effectively playing into the strategies that make SaaS sticky and cause vendor lock in.
A few of my observations for why SaaS works so well in enterprise revolve around 2 primary tricks used by the industry. They are small price points and fragmentation. By signing up 100 users at a company on a per seat license, the SaaS vendor can provide its services to different teams all having some budget to spend. Furthermore, the low price points allow SaaS services to fly under the radar and get lost in financial reviews at large organizations. Take for example PipeDrive (a Salesforce competitor) and an incredibly successful company.
At $12.50-$50 / user per month the price point is low. It is well positioned to fall significantly below discretionary spending allotted to many sales teams.
This fragmentation and siloing in enterprises is obscured from their internal spending policies. A large company may be spending millions on a single SaaS vendor and never notice it until an audit is done, at which point they will be upset about spending costs but it’s too late because they are locked in. Internal knowledge can’t shift at the speed that executives wish to wean off a vendor (once they notice the spend) and so the only option is to reach out to the vendor and renegotiate a rate (again all part of the strategy - see the enterprise deal pricing above…😊).
In my opinion, this seems like an inversion of what SaaS was initially built to solve - i.e. bloat. Companies today are faced with similar bloat, powered by unnecessary or unused products by various disparate teams paying for services. However unlike before, where the limiting factor was linked to HDD space, it is software bloat that limits mental resources. Software which is either not used or disparately used minimizes the possible financial efficiency across an entire organization.
VCs love SaaS, it is their gold standard of business. The reasons are pretty straight forward and align incredibly well with the venture capital model of investments. Here are a set of common things that make SaaS attractive to VCs:
capital efficiency (paying $<1 to acquire $1 of ARR)
good margins (usually >60%)
predictable and forecastable scaling and growth.
Effectively, SaaS is built for infusing large amounts of money into a business today in order to predictably return a multiple on those dollars in the future. And by any standard a successful SaaS business is incredibly lucrative for investors. SaaS performance on the public markets has also historically mirrored this success. To many SaaS is king.
This post continues here: [LINK]