SaaS is Eating Enterprise Collaboration — but not for the reasons you think.
The enterprise collaboration market is undergoing a massive industry transition, pivoting from premise-based software to cloud aka SaaS based delivery models. It is my firm belief that SaaS delivery will completely dominate the market over the next few years. It hardly takes an expert to predict that. However, the reasons for this takeover matter a lot, and the main reason is not the one you might think.
Most people think SaaS will win because it’s cheaper, and outline three reasons:
- An opex-based licensing model (vs. a capex one for premise software)
- Cost savings by avoiding the need to deploy and maintain servers and server software
- Usage based licensing models
While these are all true, in aggregate it’s not always clear that enterprises definitely and positively save money in the long run. The SaaS vendors are highly incentivized to price their products high enough that it’s cheaper than the costs of running the software on premise, but not dramatically so. Indeed — the growth opportunity in these markets is all about taking the spend which would be directed to IT staff and server infrastructure, and redirect that spend towards the SaaS vendor. This is sufficiently important that it’s worth repeating — the growth in these SaaS markets is based on redirecting spend on headcount and hardware towards the SaaS vendor themselves. The SaaS vendors don’t want to eliminate that spend, they want to eat it.
The real benefit of SaaS is that it fundamentally produces better software. This is what I call the fundamental theorem of SaaS:
SaaS is a superior model for software delivery because it puts a single provider — the SaaS vendor — in complete control of the three pieces of the solution — the server software, the client software, and the operations of both. This singular ownership unlocks better user experience, better quality, and better innovation velocity.
To understand why that is, it’s helpful to look backwards on how enterprise collaboration technology was delivered. It has evolved through three stages:
The Hardware Era was introduced with the advent of digital technology and circuit switching. Enterprise collaboration was nothing more than telephony. The telco operators provided it, and they obtained their server hardware from one vendor and endpoints from another (some vendors made both). Cellular telephony was similar — a fundamentally three party ecosystem.
The move to enterprise circuit switched PBXs shifted things to a two party model for the most part — solution vendors like Nortel providing phones and switches, and IT taking on the role of operations. This accelerated as we entered the software era. In this era, enterprise collaboration began to expand beyond telephony and into meetings, email, chat and other tools. This era is firmly two-party — the software vendor to deliver a solution, and enterprise IT to operate it. In the software era, desktop applications grew to prominence and the solution vendor provided both the server software and its associated desktop app. IT was responsible for deployment and upgrade. This means that new features came at a pace dictated by IT upgrade timetables. Upgrades were hard and complicated. Consequently, upgrade cycles were measured in quarters or years. Similarly, because IT operated the service, any kind of metrics around quality, reliability or usage were known only to enterprise IT. This meant the solution vendor was blind — besides testing done in its own lab, it didn’t know whether the software was working well in its deployments with customers, and it didn’t know how it was being used.
All of that is changing as we enter the SaaS era. In the SaaS era, the software vendor takes responsibility for operations. It can upgrade the clients and the servers at will, and do it frequently. Mobile app stores have made frequent (every other week) app upgrades the norm; web-based clients and modern auto-updaters for desktop apps put those on fast upgrade cycles as well. Modern cloud software enables upgrade without downtime — a property absent in the software era — allowing the upgrade cycles to happen weekly, daily or even more frequently. SaaS vendors have built automation to enable these upgrades to happen automatically so they could happen faster and faster. Put together, the SaaS vendor can now change any aspect of the software, and do it without dependency or permission from enterprise IT. This moves upgrade cycles from once or twice a year, to once or twice a day, and improvement of two to three orders of magnitude.
Because the SaaS vendor is responsible for operations, they can instrument the software to provide metrics and analytics on every aspect of its usage, from quality (how often did a call fail? how long did it take to join a meeting) to engagement (how often did a user use the chat feature? how many messages did they send?). They can measure uptime at any one customer or across all of them. They can carefully track latencies across the entire system and understand how fast it operates. They can collect and analyze clickstreams — looking at exactly where and when users clicked within the app. This data can then be used to improve the software.
The combination of rapid upgrade and instrumentation is what unlocks better user experience, higher quality, and greater innovation velocity.
SaaS provides a better user experience.
Because a single vendor owns everything, they can iterate quickly on the user experience and make changes to the product. Because they own operations, they can collect metrics and data on how the app is being used, and run experiments to determine how to make it better. They can adjust constantly and see the results of those adjustments. They can identify where and why users fail to engage with a new feature, and make adjustments to the product to improve it. They can even run AB testing — using distinct client experiences for different groups of users, and test to see which one performs better. User experience can be (in part) powered by data science. That was not possible in the software era.
SaaS provides better quality.
When done well, SaaS products can provide better uptime than classic software products. This may, at first glance, sound counter-intuitive, but it’s not. Because a SaaS vendor owns operations and software development, they can produce complex software systems with global deployments and data center redundancy. They can design for backups; they can deploy and operate complex database technologies with survivability. In the software era, IT was responsible for this kind of complexity and it was simply unachievable for any but the largest of enterprises. In the SaaS era, this complexity is borne once by the SaaS vendor and the costs of that complexity are amortized over every customer. Similarly, the SaaS vendor can design their software for the one and only deployment environment they operate, and thus ensure if functions well there. When one looks at vendors that do this extremely well — Google comes to mind — ask yourself this — when was the last time Google search was down?
It’s not just uptime — it’s also reliability. In a SaaS model, reliability can be measured. Every join event in a meeting can be instrumented and tracked, and the SaaS vendor can know whether it worked and how long it took. Every phone call can be analyzed, to know how long the setup delay was, and where the delays came from. Every message transmission can be instrumented, and the SaaS vendor can know whether it succeeded or not and how long it took to send. In the software era, reliability was measured by customer found defects. In the cloud era, reliability also becomes a data science, and doesn’t require the customer to find the problems. The data points them out, and it can find even the most corner-case, infrequent problems.
SaaS provides better innovation velocity.
In the SaaS model, if a vendor wants to ship a new feature, it can bring the time from ideation, to when that feature is in the hands of every end user, to a number measurable in days — depending of course on the complexity of the feature. Most features require changes to the clients and servers. In the software era, those changes came out in infrequent packaged releases, and relied on IT to install them and then enable those features. This took many quarters or even many years. In SaaS, the only gating factor is the amount of time required to develop and mature the feature. Once ready, the servers can be upgraded, the clients can be upgraded, and then every user has the feature.
But it goes beyond that. Innovation can also become a data science — and that is the core concept behind machine learning and artificial intelligence. Machine learning is transforming many aspects of software, and it relies on massive amounts of data that are collected by the product and then used to modify the algorithms (aka models) used in the product. It requires constant iteration. The models must be constantly tuned and retuned. It also requires significant levels of complexity to store and manage the data, and even more complexity — in the form of compute — to modify and train the models. This complexity can be borne by the SaaS vendor and then amortized across all customers, whereas it would be impossible for each and every enterprise IT department to do it on its own.
This is why I get excited about what SaaS can do for enterprise collaboration software. The fundamental theorem of SaaS has proven itself true across many enterprise software markets, and it will be true in enterprise collaboration. Vendors that embrace it, and leverage the characteristics of SaaS which make it unique, will win. Others, especially those who view SaaS as just deploying and operating the same version of their premise product, will never be able to move as fast, be as innovative, or produce software with as much quality as their competitors, and thus ultimately lose.
I’m personally putting my money where my mouth is, and for my next job I’m only considering pure-play SaaS vendors.
