Software has undoubtedly permeated every facet of the modern world, reshaping industries, and business operations at an unprecedented pace. On average, an enterprise now deploys a whopping 250+ software applications , with larger corporations boasting over 2000 applications. The proliferation is palpable, driven by the enticing promise of heightened productivity through automation.

However, let's scrutinize the tangible gains in productivity over the past 15 years. 

Take the S&P 500 for example. The number of employees needed to generate $1M in revenue has decreased from approximately 3 in 2005 to approximately 2 in 2023. In other words, enterprises have increased revenue from ~$333,000 per employee in 2005 to ~$500,000 per employee in 2023 or ~50% gain spread over the past 18 years. This means a 2.2%  gain per year in productivity over the past 18 years. To put this productivity gain in context, inflation alone during the same period has been ~2.5% per year on average.

Let’s look at what enterprises are paying for these efficiency gains.  Software spending as a share of total company cost is close to 14.1% according to the SaaS Inflation report published by Vertice. In addition, over 73% of all software vendors increased prices in the last year with an average increase of 12%.  Is the price increase warranted? Are you receiving 12% more value than you did last year from the same software?

All this begs the question : Is the software spend commensurate with the efficiency gains in the enterprise?

Recently, I came across a research report by Nexthink on Software Waste which talked about software usage within the enterprise. The data that stood out the most for me was this one:

“Of the 6 million anonymous customer devices from 9 industries, and 12 regions, roughly half (49.96%) of all installed software and licensed SaaS applications went unused by employees!"

Does this sound right? Considering the rapid software adoption and rising expenditures on software in the past two decades, I would have anticipated that enterprises have taken steps to minimize soft-waste. This is particularly relevant given the revenue model for the most part is on a per-user per-month basis.

The report further went on to highlight that “in a poll of 200 IT leaders, only 5% of them claimed ‘complete visibility’ into the total number of software licenses being used”. That explains why despite non-usage, the spending has continued to increase. 

Are there prevailing norms for measuring actual usage and ROI in an enterprise once a new software application is deployed. And if not, should there be?

While software spending constitutes a substantial portion of total company costs, accounting for 14.1%, the steady increase in software licensing expenses at an average rate of 12% annually exacerbates what's been termed "SaaS-flation". Paradoxically, despite these soaring costs, a staggering 50% of users fail to utilize these pricey licenses effectively.

All this begs the question - Are enterprises really getting the benefit from all the spending on software in the past two decades?

The software market commands a staggering size of $650 billion as of 2023, with projections indicating an impressive growth rate of 11.7% over the next decade. These figures underscore the undeniable demand for software and automation today. However, it's worth noting that despite the burgeoning demand, an estimated 50% of users within an enterprise do not leverage the licenses that have already been paid for. Furthermore, the overall productivity of enterprises has shown minimal improvement over the past two decades. This presents a compelling paradox deserving of exploration: 

Why does such robust demand for software coexist with stagnant productivity levels and underutilization?

One perspective posits that generic, one-size-fits-all software fails to sufficiently engage end-users. Could it be that these applications, laden with features attempting to cater to a broad spectrum of users, ultimately lack the ability to cater specifically to the needs of individual enterprises ?

Is the development of custom software applications tailored precisely to the needs of individual enterprises the right approach to increase usage? Yet, what about the cost and risk entailed in the development, deployment, and upkeep of bespoke software solutions? It's evident that enterprises are choosing to economize by opting for one-size-fits-all solutions, given the relatively modest $35 billion size of the custom software market compared to the overall software market. The question arises: why must enterprises find themselves torn between cost-effectiveness and optimal utilization?

Can we build a solution that addresses both – low TCO and high customization?

Customized to solve unique challenges of an enterprise - Existing market solutions often adopt an 80-20 approach to software development, and it's a strategy that undeniably holds merit. Crafting software tailored to individual enterprises entails significant time, expense, and risk.

The reality is that a substantial portion of feature requirements are likely universal across enterprises. Emphasizing these common features is often sufficient to persuade enterprises that off-the-shelf software addresses their needs. Furthermore, the allure of simplified change management and the promise of spreading development costs across numerous enterprises and users contribute to the appeal.

However, this convenience comes with a hefty price tag - a per-user, per-month cost that can quickly escalate as organizations grow. As a result, typical enterprises find themselves allocating a considerable portion of their total expenditure to software providers - ~14% to be precise.

The aim, therefore, should be to offer greater customization, simplified maintenance, and straightforward change management to enterprises without incurring exorbitant costs and encountering low user adoption rates. In essence, the objective is to develop customized software that genuinely serves the end user needs while also having the following characteristics:

• Low Total Cost of Ownership - This one is a no brainer. Every enterprise wants to save costs as long as they do not have to compromise on functionality, quality and user experience.
• Fast time to Value – Customization, in addition to being expensive, has long been synonymous with two unwelcome words: Risk and Time. Mention customization, and the knee-jerk reaction is often that it'll be riskier and more time-consuming than off-the-shelf solutions. Hence, any customized solution must not only offer cost-effectiveness but also mitigate risks and address time-to-value considerations.
• Security, Auditability and Compliance - A customized cost-effective solution that is delivered quickly is not enough if the solution is not secure and compliant with the needs of specific enterprises. Hence, security, auditability and compliance become table stakes when developing and deploying any software application to the enterprise.

We founded myl AI to be a platform that addresses these four key themes effectively and efficiently. The goal is to be entirely focused on making enterprises successful while building a profitable business that can have a lasting impact on how software applications are developed and consumed.

Timing is everything! I have always wondered why SaaS companies charge on a per user per month basis and gets away with it. They do have some cloud infrastructure expenses that need to be covered but the initial development cost is amortized across thousands of users and hence the total cost of ownership for enterprises should be far less than the traditional old-school licensing models. 

During the licensing era, companies charged several million dollars for multi-year software application licenses. I recently saw an article from the early 2000s that talked about Siebel systems, the fastest growing company in 2000 that then sold software through traditional licensing models. The article mentioned that the average cost of a three-year deployment for Siebel was $6.59 million, including software, consulting services and support personnel. That cost worked out to about $18,000 per user. This got me thinking!

In addition to offering all the flexibility that comes with SaaS, cloud software companies also had to be more cost effective. How cost effective were they? Here is some back of the envelop math for consideration:

A deal size of $6.59M and $18,000 per user, implies $6,000 per year per user or $500 per user per month. When an enterprise today spends $50 - $100 per user per month for a similar SaaS application, they are in effect 5-10X more cost effective than traditional licensing models. 

Going by conventional wisdom, 10X is what is needed to make a market shift towards a technology. SaaS companies were able to effectively deliver this to customers and achieve massive success over the past two decades. This is on paper though as half the users do not use the solution in a given month.

If we need to move the market away from existing players, we need to deliver the promised hyper customization at 5-10X more cost effectively or provide software that eventually boils down to $5 - $10 per user per month from a customer perspective.

Is it possible to do all this – hyper customization, lower risk and lower TCO in a Gen AI world?  

This is the question we started with when founding myl AI and we are well on our way to answering this!