The fall and rise of SaaS
Why the death of SaaS is greatly exaggerated, the costs and pain of new or replacement software and some thoughts on what will be need to for SaaS businesses to succeed with AI...
A bad day for software companies on the US exchanges yesterday - Salesforce, Adobe, ServiceNow all fell sharply. So did Gartner and IBM - not software but linked. This has encouraged many commentators to dig out their “death of SaaS” thinking.
The argument is simple: why buy pre-built software when you can create your own bespoke software for virtually nothing? Not only does this save money, it should give you better software because you would be building for your precise needs.
The pain of change
Perhaps I used the wrong word. These arguments are not simple, they are simplistic. Here are a few things they ignore:
The true cost of software includes setup, implementation, change management, training and so on. These can add up to 30-75% of the first 3 years subscription cost. 3 years because the time can drag as well.
Those extra costs also include integration with existing software. The more bespoke software you create, the uglier integration is going to become.
Integrating new software is like pulling out one piece of a jigsaw and replacing it with one a different size and shape. The hard bit is cutting the other pieces or creating new ones to make it all fit.
So implementation, integration and change is where all the pain lies, especially with modern software. Numbers matter but is a legacy of challenging or failed implementations that really scares business buyers.
None of these things go away with bespoke software. If anything they become harder. Vendor templates for training and implementation may be flawed but at least they are based on real world experience.
For larger companies, standardised software is an illusion. Implementing something like SAP or Salesforce takes time and involves a lot of configuration and customisation. Look at the numbers for Accenture, Infosys or the other big systems integrators if you don’t believe me.
The superiority of bespoke software also rests on one huge assumption. It only works if you know exactly what you need. True, many SaaS products include features your business may not need. But those features are there because customers asked for them. So there is a fair chance some of them will be things you do need but just haven’t thought of yet.
The real world of new software choices
Imagine you are responsible for planning the tech roadmap for your business - any size business. You have been running for a while. You have a bunch of stuff that works. You have a list of grumbles and perhaps more serious issues with some of the existing software.
And you have a boss who is demanding progress with AI and/ or using tech to transform business performance.
Where does replacing existing SaaS with buzzy new vibe coded bespoke software fit on your agenda? Not very high I would suggest. Nowhere if you have a huge application like SAP and it is either in the working well or minor grumbles pile.
How might SaaS succeed with AI?
So what might the future look like for existing SaaS businesses:
Resistance to change from existing customers will fade slowly. That revenue will decay but not disappear overnight.
AI almost certainly makes configuration and integration easier and quicker. This will help a bit but pretty small in the context of the pain caused by major software changes.
There is an opportunity to add customer coding “features” to enhance existing SaaS products. Think something that enables users to vibe code with controls that ensure security of data and automatic integration with your existing software.
Established businesses have a much greater surface area of knowledge and insight about what businesses want than any new entrant. The winners will leverage this to evolve into powerful AI vendors.
What does that mean for share prices? No idea. The long term future will depend on the ability of leaders in SaaS to learn and evolve. There is no formula for working out which companies have the best culture and leadership to exploit this opportunity.
Thanks for reading
I've been in the IT industry for over 10 years, and I totally agree with this post. SaaS and other software are part of a company's system, and they're usually pretty big and hard to replace.
If AI works out, it'll meet needs we don't even know we have yet. But it won't replace the solutions we already have.
This article really made me think: AI’s value comes from its non-determinism—the same thing that makes it hard to trust for critical workflows.