AI Risks to B2B SaaS Companies: A Framework for Estimating Risk

0


Artificial Intelligence will affect B2B SaaS very unevenly. Most companies will be affected only marginally, as AI becomes another tool in the toolbelt for technologists to deliver business value. However, there are some parts of the B2B SaaS ecosystem that will be utterly hollowed out by AI, as entire departments or functions disappear. We suggest here some mental models to use to help you estimate the risks to your SaaS business from the rise of AI.

What we mean by “AI”

AI Risk Framework for SaaSWhat’s being called today “AI” (namely: generative language or image models, and good general-purpose classifiers) is a big change in computing capability. Few advances in computing are quite as momentous: maybe the Graphical User Interface, the wired Internet, and mobile device wireless. However, like those prior new capability shifts, AI will mostly not fundamentally change what businesses do.

AI will, in most cases, be a new tool that businesses use to speed along things they already do. Consider how you used to have to go to your desktop computer – or even visit a teller! – for the online banking tasks you now do from your phone. It is more convenient to be able to pay the water bill while sitting in a park or a cafe; but fundamentally, you’re still making debits and credits much like great-grandpa once would have done with paper and pen.

“OK! So, my SaaS business will be safe and sound, then, right?” Sorry, dear reader, but the truth is somewhat darker: AI is going to perforate the B2B SaaS ecosystem and leave it looking like Swiss cheese. And it’s going to happen in a way that’s subtly harder to think about than prior advances.

AI will vaporize some SaaS businesses while others are untouched.

If your SaaS product does something that still needs to happen, and if AI helps that happen somewhat better within the same product or process, it survives. (Maybe it even delivers better value and charges more!)

But if AI means that the thing the SaaS company’s software is doing simply no longer needs to happen – at a vertical market or horizontal functional level – that’s when there’s serious business risk.

And if AI can deliver the SaaS company’s business value outside of an existing SaaS product – such as if a once-hard computational problem is now easy for AI to solve – that’s when there’s existential technology risk.

AI Risk Framework

Here’s our high-level systematic framework for thinking about the business and technology risks to SaaS companies from the rise of AI:

1. Biz risks

Consider carefully and with intellectual honesty what might occur to your vertical market, to the functional area you sell into, and to the actual end-users who interact with your product.

  • Risk to your target vertical (do you sell software to buggy whip manufacturers). High risk; pivot or shrink.
    • Mitigation: Accept niche status and dominate share-of-market, or shift verticals.
    • Non-SaaS Example: Don’t sell chips that control gasoline cars’ fuel injection; pivot to chips that control electric vehicle battery charge and usage.
  • Risk to your horizontal/departmental function (do you sell into the typing pool or mailroom). High risk; pivot or die.
    • Mitigation: “Eat” the entire department. This function may not exist in a few years. Become the replacement for even having such a department.
    • Non-SaaS Example: Every company used to have a “typing pool,” and now none do; word processing and email made every employee a typist.
  • Risk to your end-user (does your daily user lose his job to AI). Low risk; remediable.
    • Mitigation: Move up the value chain, and ensure the departmental/functional owner gets promoted for using your product.
    • Non-SaaS Example: The elevator still goes up and down, but there’s no elevator operator anymore.

2. Tech Risks

Consider how much of your “secret sauce” / sustainable competitive advantage is technology-based.

  • Advantage based on laws of nature or man. Physical/mathematical principles; legal/regulatory compliance. Low risk.
    • Mitigation: Probably safe.
    • Non-SaaS Example: Even after calculators, Excel, and TurboTax, complying with the IRS means job safety for knowledgeable tax accountants.
  • Advantage based on pace of development and know-how: beware. One of the most important effects of AI will be in making programming many multiples faster. Medium risk.
    • Mitigation: A “1x dev” and a “10x dev” are going to converge as they both use Copilot and similar AI; you need to build other moats.
    • Non-SaaS Example: It doesn’t matter how heavy a box you can lift if everyone in the warehouse has a forklift.
  • Advantage based on domain-specific algorithms (not constrained by math/physics/law), such as natural-language processing and tuned machine-learning models: danger!!
    • Mitigation: Immediately build orthogonal areas of sustainable competitive advantage, and hoard access to training data. The ability to create and train algorithms/models is no longer differentiated.
    • Non-SaaS Example: Books used to be hand-crafted by monks, so having one was very special and expensive. Machine learning models and algorithms used to be hand-crafted by geeks, but they’re about to be mass-produced by AI systems.

Safety strategies

What will remain absolutely rock-solid? First, Systems of Record. Because a SoR’s chief value is in being an online and archival single point of truth about its inputs, not in cleverly transforming those inputs, AI will have relatively little impact. (However, the underlying business process being logged into the SoR may be at risk!)

Network effect/marketplace plays which facilitate commercial transactions between buyers and sellers should remain healthy, provided the underlying parties being facilitated are healthy – a big assumption.

The closer the underlying business process is to revenue generation, the lower the risk. Beware of cost center buyers, particularly in times of uncertainty and industry shake-ups.

The risk is lower the higher up the end-user is in his or her organization. Make sure someone with P&L authority looks like a genius for buying – or ideally, personally using – your SaaS product.

Danger zones

Knowledge management, digital asset management, task management, digital workflows. These are super-fraught tooling areas.

AI will excel at tasks like, “find all the video clips that have 20-60 seconds of quirky pets being funny but also have reasonably inexpensive licensing terms for use on YouTube.” If your SaaS helped people do that last year, look out: there won’t be any people doing that next year.

Your CTO may have agonized over specific tweaks to the search interface, or the milliseconds of UI response time. No matter. The AI will directly operate on the underlying data and one or more layers of “User Experience” have just been completely obviated.

Financial signs and signals

If AI is threatening your business at a horizontal level, you will see pressure on Gross and Net Revenue Retention. Make sure contract lengths or usage minimums aren’t obscuring this: you may need to peek through to underlying usage metrics as well.

If AI is threatening your business at a vertical level, expect lengthening sales cycles and close rates, slowing overall growth, and eventually, a drop in retention rates.

If AI stands to improve your SaaS company prospects, such as if it is a capability that increases the speed/value you deliver to customers, you’ll see it in increased Gross Margin and Net Revenue Retention.

Conclusion

We hope this framework gives you concrete ideas as to how to think about risks and mitigation, as well as specific financial and operational metrics to watch. We welcome your feedback, and we remain keen to work with B2B SaaS companies with a bright future – which today means those companies that are vigilant about AI risks and proactive about mitigating them.

 

Randall Lucas

Managing Director, SaaS Capital

SaaS Capital® pioneered alternative lending to SaaS. Since 2007 we have spoken to thousands of companies, reviewed hundreds of financials, and funded 100+ companies. We can make quick decisions. The typical time from first “hello” to funding is just 5 weeks. Learn more about our philosophy.



Source link

You might also like