The Bloating SaaS Bills
In today’s digital landscape, Software-as-a-Service (SaaS) has emerged as a go-to solution, whether a CRM solution for smaller startups or human resources management for large multinational corporations. However, there’s a twist in the tale. Gradually, these SaaS subscriptions start to inflate, transforming into significant expenses that weigh heavily on a company’s financial statements. This begs the question: what exactly led to this bloating in the SaaS industry, and how can we navigate this challenge? This blog aims to unravel my perspectives on this pressing issue.
Evolution of SaaS Companies
The concept of the SaaS business model started gaining traction in the late 2000s, particularly between 2005 and 2010. Salesforce emerged as a trailblazer, with numerous companies, especially in Silicon Valley, quickly embracing and driving this trend. Globally, companies began adopting SaaS at an increasing rate. Notable names in the SaaS world include Salesforce, Google, Atlassian, Zoho, Quickbooks, Zendesk, Microsoft O365, and HubSpot, to mention a few that readily come to mind. Some new-age and popular SaaS companies are Asana, Zoom, Monday, Rippling, Deel, Workday, Flexport, and more.
But how cost-effective can SaaS truly be for large companies? Drawing from my experience building a SaaS company, I understand the extensive resources required. It’s not just about the product; there’s a whole infrastructure involving operations, billing, account management, customer relations, site reliability engineers, DevOps teams, front-end and back-end developers, database engineers, data analysts, etc. Running even a moderately complex SaaS business demands a significant workforce.
Inevitably, these operational costs are passed on to the customers, encompassing not just the service but also sales and marketing efforts. This reality prompts a reflection on the model’s sustainability, especially for larger companies. What challenges do enterprises face in this context? And what does this mean for the future of SaaS in big business settings? Let’s explore these questions further.
Some Real Examples of High SaaS Bills
In my research, I’ve found some interesting examples highlighting the escalating costs of SaaS models. For instance, the CTO of a prominent tech company revealed that their annual bill for Datadog, a SaaS platform for managing AWS service costs, surprisingly exceeds their AWS expenses. It’s an ironic twist where the tool intended to control costs is more expensive than the monitoring service.
Another eye-opener came from a recent episode of Jason Calacanis’s podcast, “This Week in Startups.” David Heinemeier Hansson, co-founder of Basecamp, discussed their new venture, once.com. This platform advocates for a ‘buy-once’ approach to software, moving away from recurring subscriptions. The inspiration? A conversation with the CEO of Spotify, who disclosed their staggering Slack bill running into millions, a figure he’d rather not dwell on.
These examples are just the tip of the iceberg, and I’m sure many can relate similar stories within their organizations. The ballooning costs of SaaS subscriptions have reached a point where the invoices seem almost ludicrous.
A TCO Analysis of SaaS vs Hosted Options
Let’s dissect why escalating SaaS costs are a significant problem. Initially, SaaS appears to be a convenient solution, but over time, it can become a financial burden. Here’s my perspective: Many SaaS companies adopt a ‘per-seat’ pricing model as their default. Another common approach is usage-based pricing, charging for metrics like API calls or data storage. The latter, exemplified by OpenAI’s GPT-4, seems more reasonable.
Consider OpenAI’s ChatGPT+ and ChatGPT Enterprise models. While ChatGPT+ charges a flat $20 per user, ChatGPT Enterprise jumps to $60 per user. Let’s take Gartner, for example. Gartner is in the knowledge industry and a perfect customer for ChatGPT Enterprise. Below are Gartner’s numbers.
- Revenue: $1.41B
- Net Income: $180M
- No of Employees: 21,000
- Revenue/Employee: $67,142
If Gartner procures ChatGPT+ for all 21,000 employees, we are looking at a monthly bill of $1.26M and an annual spend of $15.12M. This is 8.33% of the net income, and even a knowledge company like Gartner, for whom a language model tool like ChatGPT is super useful, will find this number relatively high.
Now, if Gartner builds an internal ChatGPT type app and uses GPT4 API instead, assuming each employee of Gartner will do about 20 API calls a day, the annual spend will be $4500 (for GPT4Turbo), about $2500 (for AWS hosting). So this is about 2160 times more than the app version.
GPT4Turbo Pricing
This analysis highlights a critical flaw in the SaaS model, particularly for larger corporations. The per-seat pricing, while straightforward, can rapidly become unsustainable, pushing companies to seek more flexible, usage-based solutions.
Even if we consider other AI tool providers, many of which are effectively ‘GPT wrappers,’ the pricing dilemma persists. These companies, such as Jasper.ai, Jenni.ai, Writesonic, and many others, largely adopt a per-seat pricing model. Imagine a CIO at a 1000-employee company looking to deploy AI tools to just 20% of the workforce. The cost of acquiring licenses for multiple products for 200 employees can still be exorbitant.
- General Usage: ChatGPT = 200*$60 = $12,000 per month
- CopyWriting: Copy.ai = 200*$10 = $2,000 per month
- Analytics: DataGPT.ai = 200*$20 = $4,000 per month
- Knowledge Base: Klu.so = 200*$15 = $3,000 per month
- Coding Copilot: Replit = 200*$20 = $4,000 per month
- Marketing: AdCreative.ai = 200*$20 = $4,000 per month
- Sales: Intently.ai = 200*$20 = $4,000 per month
And there are many AI copilots for many other functions. The above list takes the monthly burn to $33,000, which is already unsustainable.Â
In my opinion, the AI tools pricing model becomes unsustainable as companies grow beyond 50 employees. Most offerings are essentially wrappers on GPT – thin, medium, or thick – but wrappers nonetheless. I must admit that some of our earlier products that we started building in the early part of 2023 were also GPT wrappers. So, with GPT API prices being 2000X cheaper than the apps, is the SaaS pricing model the right way forward for AI companies building on GPT tech? And should businesses pay for them at prices equivalent to ChatGPT?
Exploring alternatives to the traditional SaaS model, I’ve identified five viable options:
- Lifetime Deal Platforms: Services like Appsumo and once.com offer software with a one-time payment called Lifetime deals. You own the software for life, eliminating recurring costs. The drawback, however, is the need for ongoing maintenance and the uncertainty of future updates or support.
- In-House Development: Building custom software tailored to your needs sounds ideal. This option allows for greater customization but comes with its challenges, such as managing a development team, documentation, and the inherent complexities of software creation.
- Outsourcing: While outsourcing can alleviate the burden of development, it often becomes time-consuming, messy, and potentially more expensive than a SaaS platform.
- Leveraging Open Source: Utilizing open-source platforms as a foundation can be highly effective, especially with robust community support. It’s a cost-effective method that allows for building on proven, existing frameworks. Probably the #1 option I would recommend for companies who know how to build on open-source frameworks.
- Lyzr Pricing Model: Here’s where we introduce a novel approach. Lyzr offers SDKs for the rapid development of Generative AI applications featuring low-code, highly abstracted functionality complete with front-end components. This model supports unlimited users, unlimited data, unlimited API calls, unlimited queries, and logs, all running on your infrastructure. We provide ongoing updates and improvements to the SDKs, adapting to changes in backend technologies like OpenAI or vector databases like Weaviate. It’s a middle-ground solution, bridging the gap between traditional SaaS and custom development.
Each of these options presents unique advantages and challenges. The Lyzr model, in particular, strives to offer the flexibility and scalability of SaaS while empowering businesses to maintain control and minimize costs.
The New AI Era Needs New Pricing Models
The SaaS industry, once a revolutionary concept, now seems to be approaching a crossroads. The golden era of SaaS may be waning, making way for innovative, disruptive business models. It’s intriguing to think of SaaS as ‘traditional,’ considering its relatively recent emergence. But in a fast-paced technological world, even two decades can render a model archaic. With advancements like GPT and no-code platforms, creating software has become significantly easier and more accessible. The high costs demanded by SaaS providers are increasingly challenging to justify in this new environment. We’re entering an age where the ease and efficiency of modern development tools outweigh the convenience offered by SaaS.
The future seems poised for models that break away from the conventional SaaS structure, embracing more flexible, cost-effective, and user-empowering approaches. This shift is not just about technology but a broader evolution in how we think about software and service delivery.
As we navigate this new era, I’m curious to hear your thoughts. What alternative business models might emerge to redefine the landscape? Share your insights and ideas as we collectively explore the future of software services in a post-SaaS world.
Book A Demo: Click Here
Join our Slack: Click Here
Link to our GitHub: Click Here
