Will AI Accelerate Vertical SaaS Adoption?
One of the lessons I, along with many other founders and investors, have learned in my roughly 15-year journey in SaaS investing is that most businesses adopt new technologies at a painfully slow pace. This might sound like an odd statement, considering how rapidly products like YouTube, Instagram, and, more recently, ChatGPT have garnered 100M+ users. However, there’s a vast discrepancy between the speed of mainstream adoption for consumer apps and the rate at which most companies transition to a new technology and delivery mechanism like SaaS. The first software category that crossed the famous chasm, CRM, took over 15 years to reach its tipping point⁽¹⁾, and currently, approximately 58% of enterprise software spending still goes to on-premise solutions.
When I invested in FreeAgent in early 2009, I found the value of an online accounting solution SO compelling that I naively believed every small business in the UK would just have to adopt online accounting. If you look at the benefits of a product like FreeAgent compared to an antiquated desktop accounting software or pen-and-paper bookkeeping, the choice is obvious, isn’t it? Well … more than a decade later, about three-quarters of SMEs in the UK still use desktop accounting software or a mix of Excel spreadsheets and manual records.⁽²⁾
Turning to vertical software, the State of the Cloud 2022 report (always an insightful read) from Bessemer highlighted the low Cloud penetration across various industries:
“Leading vertical cloud players act as a proxy for their industry’s adoption — for example, as of 2021, Toast, the restaurant management SaaS, is only in 6% of U.S. restaurant locations. Optibus, the cloud-native AI platform that makes public transportation smarter, better, and more efficient, has reached less than 1% of the total public transit software market. ServiceTitan, a leading vertical SaaS for home services, penetrated 1% of its core TAM of 900,000 home services businesses. Across other verticals such as in auto repair with Shopmonkey and laundry services with Cents we see 2% penetration. And when we look beyond North America, global cloud maturity rates vary widely, too.”
The experience of many vertical SaaS companies is that there seems to be a natural, stubborn limit to how fast you can grow. Each year, only a certain percentage of target customers are “in the market”. One reason is that most companies don’t swap out their “operating system” often. If you’ve purchased software to run your business within the last few years, you are unlikely to replace it within the next few years. (Of course, the fact that vertical software is so hard to replace is also one of the reasons why vertical SaaS companies have so much staying power and why we love them so much!)
Trying to grow significantly faster tends to be prohibitively expensive because once you’ve captured all the (relatively) low-hanging fruits (i.e., warm leads), wooing colder leads often becomes disproportionately costly. This has profound implications for vertical software companies. One is that if there’s a limit to how fast you can grow by adding new customers, it’s all the more important to continuously increase ACV by layering on new features and products and to create additional revenue streams, e.g. from payments, etc. (known as the layer-cake strategy).
Who will leapfrog the pre-AI Cloud?
Given the amazing progress of AI in the last few years, I’m wondering: Will AI accelerate the adoption of SaaS, particularly vertical SaaS? Could some slow-moving industries bypass the “pre-AI SaaS” phase and leap straight into AI-driven solutions?
There are a few reasons why I believe this might be the case:
1. AI will be everywhere.
Every B2B SaaS startup established in 2023 or beyond will be designed with AI in mind, and nearly all existing SaaS companies will seek to enhance their software by integrating AI features. For SMBs still running on pen & paper or Excel & Outlook in 2023, the first SaaS product they adopt will most likely be AI-enhanced, so many of these companies will skip the “pre-AI SaaS” phase.
2. AI can create magical user experiences.
Take e.g. Jasper and MidJourney. The value you get from using these tools is so enormous that it allowed these companies to achieve growth rates previously unheard of, even for the best and fastest-growing pre-AI SaaS companies like Slack or Dropbox.
Before AI, software mainly helped you streamline workflows so you could spend more time doing your job, or it provided the tools to do the job more efficiently. With AI, software will increasingly do the actual work itself. In many cases, the ideas aren’t new (e.g. ticket deflection in customer support, AI-based contract drafting in legal, or automated creation of medical notes in health). What’s changed is that LLMs are now much better than they were a few years ago, and a variety of APIs and a fast-growing tooling ecosystem make it much easier to add AI to existing software applications.
3. AI will reduce “time to wow”
There’s no question that millions of SMBs in all these under-penetrated verticals would benefit from using more software — but apparently, so far, they didn’t think that the change they’d have to make is justified by the value they’d get. If you can use AI to demonstrate the value of your software within the short amount of time and attention that you get from the user, this could be a big unlock.
An example is our portfolio company Mokker.ai. Their initial product, Colorful, was a 3D “photo studio” to build a scene and export professional product photos. Compared to traditional 3D rendering software, Colorful is very easy to use. But you still need a 3D model of your product and have to think about things like light and camera position. The new product, Mokker.ai, uses generative AI to create professional photos from a single product image. You just upload a product photo, choose some templates, and boooom, you get a variety of professional photos, showing your product in various environments and scenes.
Another example from the P9 Family is Qwilr, a platform that lets you create beautiful sales proposals, presentations, and other collateral. Their new AI page builder makes it much easier to get started. You just answer a few questions, and the Qwilr AI will spin up a draft for your page, which you can then edit and finalize. Here’s a Loom of Mark Tanner, Qwilr’s co-founder & CEO, explaining how it works:
Super Mario in Generative AI Land
Back in 2012, I wrote that SaaS companies with a low-touch sales model should aspire to build an awesome product and a frictionless user experience, which lets the user gradually discover the product, while getting a lot of gratification along the way. The analogy that I used was games, which have perfected the art of onboarding:
If we stick to this analogy, my bet is that AI will help lift users across adoption barriers and provide more value:
It’s still very early days for LLM adoption. Most of the LLM-powered features launched by SaaS companies so far are relatively low-hanging fruits. Many SaaS companies are still trying to figure out which AI features add the most value, how to get enough confidence in the LLM’s output, how to reduce latency, how to lower API costs, how to price those features, etc. Larger enterprises will likely hold off on widespread LLM utilization until issues like hallucinations and data privacy are addressed.
The biggest factor is human inertia, though. As Benedict Evans said, “changing a big company’s workflow, operating model and depreciation structure doesn’t happen quickly”. So when I read reports saying things like “30% of all work will be done by AI within a few years”, I think it will take longer. That said, I am convinced that AI will accelerate adoption in SaaS. If you’re building the hot air balloon for Mario, please ping me! :-)
⁽¹⁾According to Bessemer’s State of the Cloud report from 2015, CRM SaaS spend was about to reach 50% in 2016, more than 15 years after Salesforce’s launch.
⁽²⁾According to this research report published by Codat, ca. 25% of SMEs in the UK used Cloud-based accounting software in 2021. Roughly 25% utilized desktop accounting software, and about 50% relied on a mix of Excel spreadsheets and manual records.
