Services as a Software

Software as a Service, and what changed

Software as a Service needs no introduction. It was the model that birthed some of the most valuable companies on the planet. While there is aggressive talk that SaaS is dead in the age of AI, a more nuanced take is that alternatives to pure SaaS just became more lucrative. For instance, technology services are becoming a more viable model as AI scales productivity and allows small, lean teams to take on more clients and boost margins.

The goal of this article is to highlight the factors that have made SaaS models so successful over the past two decades and how services became a more viable alternative in the age of AI.

SaaS Model: Economics

The primary driver toward the SaaS model is because of the economics:

Marginal cost is near $0: How much does it cost to produce one more software seat? Basically nothing. Very different from traditional businesses that produce widgets, or services businesses where labor costs scale with growth.

Recurring revenue beats fixed revenue: It is a lot easier to keep a customer than it is to find net-new customers, so a business model that runs on recurring revenue is a lot more attractive to owners.

Very high LTV:CAC ratio: If you can keep a customer on, this ratio keeps going up and up.

So while the user experience is likely better since clients are not responsible for management and maintenance of the software, there are strong economic drivers toward the SaaS model because of the recurring revenue model and high gross margins.

But because the prize is so shiny in this business model, a lot of businesses sacrifice user experience. There is internal pressure to lock customers into long-term contracts. Having worked in several SaaS startups, the real milestone celebrated internally is contract signature. That is bad news for the client who is just starting their journey, and a lot of SaaS companies have fallen into the habit of under-investing in implementation and enablement. Churn becomes the biggest killer of SaaS companies over time, and it is a direct reflection of the user experience.

SaaS Model: User Experience

The real service in SaaS has been hosting and maintenance. But at the end of the day it is still a tool. There is still a learning curve. There is still a huge commitment from the client team. Churn is the biggest killer. Tech is not the hardest part in this business; change management is.

What clients really want to pay for is outcomes. Software has been a great way to get outcomes at scale, but it still requires a huge lift from the client to get those outcomes. This causes a few issues:

There is a big gap between the current state and the ideal future state for the client. This slows the sales process when the perceived friction and commitment required from the client is high to see success.

Even after winning the deal, change management becomes the main barrier to success. Enabling an organization around a new way of working is the biggest challenge, and if it is not done correctly, they churn.

But this has always been the case with SaaS. Why are some saying that we have entered the SaaSpocalypse?

LLMs and agents are hinting at the possibility of bringing clients closer to outcomes than SaaS has ever come before. That means less work, less change management, less learning curve for clients, and a tighter feedback loop between invoice and outcome. This theoretically speeds up deals and shortens the time to value.

So what is the fix for existing SaaS players? The answer has split into three pillars:

Slap AI onto your company name and website. Many legacy SaaS companies are doing this, dot-com-bubble style, in hopes of driving the value of their business up without having to do any real work.

Heavily adopt AI and pivot the company to operate on this model. Much easier said than done, and there are very few companies pivoting successfully.

Adopt a hybrid services plus software approach. Instead of a pure technological change, reduce the burden on the client to make the change. Palantir popularized this model with the Forward-Deployed Engineer.

But they were all of them deceived, for an existing business model was emerging as a more lucrative option enabled by LLMs and agents.

Services Model: The New Economics

While this is far from a new business model with massive legacy players like Accenture, Deloitte, McKinsey, and others, small boutique services struggled with this model because of the human-intensive nature up against the economies of scale of the big players. Time was always the main constraint, and costs scaled almost linearly with growth. And even SIs and consulting firms with huge economies of scale traditionally have not come even close to SaaS margins.

But if we simplify AI down to its root, it is an amplifier of abilities. If you do not know what you are doing, it will amplify that incompetence (AI slop). If you do know what you are doing, you are working more than you ever have been and are addicted to the level of agency you now possess, like a newly acquired super power.

This new toolset is giving small boutique services firms the economies of scale previously only possible with large SIs and advisory firms in the past. And this new category is punching way above its weight in categories that were previously dominated by pure SaaS players. Now clients that could never afford Accenture fees are getting custom-built software, integrations, and completely done-for-you solutions that were previously shelfware in their organizations.

Software vs Services

I think the answer is both will be required to win in the AI age. SaaS is not dead. Services are not completely replacing it. But the model will look different, and the companies that succeed will rely on both: software for the scalability, services for change management and adoption. Because AI has driven the cost of code to near zero, pure software will become even more of a winner-take-most industry, and the companies at the top will rely on other levers for differentiation. This trend toward concentration is happening already in talent, as top talent is being fought over now like top sports prospects.

For everyone else not in the pure software game, hybrid or pure services will get a lot more popular. We already see this direction in the Palantir model of Forward-Deployed Engineers. Customers are going to expect more hand holding to the end result instead of buying a tool and going through the messy learning curve.

Eventually, agents are going to take more and more of the lift off of humans for enablement. Right now, Apero heavily uses agents and liaises between them and our clients. My prediction is as organizations get more AI native, they will have their own agents highly trained on their company's context that they will want to handle implementation and management of any new software. Agents will increasingly become the primary user, and companies will expect less and less manual work to get the outcomes they are ultimately paying for.

Positioning Apero for Success

Apero is adopting this hybrid approach. While we structure relationships like traditional services deals, our deliverables are increasingly custom software that can be repurposed for future clients. Building our own software just becomes one of the arrows in our quiver for solving whatever problems the client has, and sometimes we find something we think could be repackaged for other clients. So by getting deep in the weeds with solving unique problems, we sometimes stumble on repeatable solutions. A kind of R&D motion for a software factory.