The Fake SaaS and Why It Is Bad for the Customer
Ask yourself this question: Has any meaningful SaaS company used a commercial or open-source product as the basis for their service? Salesforce.com? Concur? 37 Signals? ExactTarget? RightNow? Responsys? Demandware? Workday? Success Factors? Taleo? ADP? Crownpeak? Of course not! If an existing product could be used to deliver a SaaS, there would be big competition out there based on that model. The traditional products simply can’t meet the architectural requirements for a multi-tenant shared service. Especially for the SaaS companies operating for large enterprises, where high-availability at global scale is table stakes.
Many installed product companies suddenly have a “SaaS” offering. The problem is, they are not SaaS.
A more accurate way to describe such an offering might be a “hosted offering” or an “outsourced offering”. But it’s not a SaaS.
About 12 years ago, there was a model called, “ASP”. ASP stood for “Application Service Provider”. The idea was to host 3rd party applications for many customers and have some level of added efficiency. In other words, try to do it better than the IT department could do internally. It was an utter failure. All the ASP’s died in just a few years because they were essentially trying to deliver on the promise of a SaaS without having a SaaS product to deliver that efficiency. See my post on the Crownpeak architecture to understand why a purpose-built product is totally different and totally necessary to deliver a service to a global organization.
So what’s different today that will allow an ASP or hosted model to succeed? The cloud?.
In truth, the cloud isn’t a cost saver unless there are special circumstances requiring a lot of server capacity at certain times. Mainly, it’s a complex shared hardware and network environment that at best will save a few bucks, but typically costs more to operate because the service provider has to build and manage redundancy around possible failure points. Believe me, I know the economics of cloud computing
Why does the end customer care about whether something is a true SaaS or is a hosted model or ASP? Because the benefits of SaaS are clear: cost, flexibility, time-to-market, innovation, and interoperability. And with the ASP or hosted model, those benefits can’t ever arrive.
To get an installed product to run in a redundant architecture in a scalable fashion for a global project is a very complex installation. More importantly, every organization and in some cases, different projects for the same organization require a separate, highly complex installation. Any difference between those installations in any way makes the task of managing them more complex, expensive and time-consuming. The application either has to be completely cookie-cutter (which doesn’t fly with organizations who require custom functions and integrations), or the hosted provider is facing exponential growth in complexity (and thus cost) from one-off, customized projects. And that’s the death of the model. These companies simply can’t deliver SaaS reliably and inexpensively because they can’t prevent variations or get away from complexity. This complexity also directly impacts the customer because over time, the hosted provider will require a highly-controlled software development and release cycle in order to move any change into production. Therefore they can’t enable an agile environment for the customer.
Because these last-generation products haven't been built to run hundreds or thousands of different clients on a single software platform, they need to install individual, highly redundant instances for every customer. A cloud setup can’t change the inherent product architecture. And the customer really cares about this because every on-premises product run in the cloud has a linear cost model and so they must either pass costs on to the customer or run an unprofitable business over time when additional capacity and staffing is required.
A few quick questions for customers to ask hosted-model product companies:
- What if traffic goes up by 100x for a week of time? Can the system handle that? How quickly can you add capacity? Will there be bill to us?
- Are major upgrades free? Will we ever see a cost associated with an upgrade? What if our implementation has many integration points and some customizations?
- If we decide to add a new stage to our workflow or instance of the product that requires a separate delivery environment for a legal requirement, how can we accomplish that? Will it require an additional cost? How much?
- Are hardware upgrades free? Will they require downtime? How about increases to capacity required as we add new content, new sites, and new end-users of the system?
- If we want to have 2 or 3 different 3rd party development teams implementing projects simultaneously, is that possible? Is it advisable? What type of software development lifecycle do you recommend to keep the application stable and error-free?
- Can you guarantee how fast content will be delivered in our important markets like (Brazil, South Africa, China)? If the content isn’t fast, will we have to pay to upgrade in some way?
- If security is compromised by a 3rd party developer’s work, will you mitigate the problem and how much will you charge? How quickly can you roll-back in the event of a security issue? How quickly will you know security has been compromised?
And so on. In the end, the other ASP/Hosted Providers are in a tough spot. They can get away with non-standard installations and a lack of cost-scalability when the projects are simple and the customer doesn't expand beyond a single box, require customizations or have global delivery requirements. But as the customers scale up, they have big costs and long timelines when it’s time for an upgrade or a new stage needs to be added. When there are usage spikes, they aren’t running in a shared environment and can’t simply absorb it. They have to add capacity for the individual customer and that means cost. And solving the tricky problem of global performance isn’t any easier for them in the cloud than it is for the IT team running the platform internally. It costs a lot of dough to take a single-tenant system and expand it across the globe for performance or regulatory reasons
Crownpeak has spent tens of millions of dollars and over ten years of software and infrastructure development to be able to say a few simple things: Any element of our system can have a catastrophic failure and your WCM and web sites will still be live. Further, no matter how many upgrades, capacity increases or new functions we deliver, you will never see a bill for any of that. No fake SaaS will ever be able to say those things.
Not even Adobe who has bought a few pretty good SaaS companies. It would require a total top-to-bottom rewrite and I don’t see them signing up for that anytime soon