Bursting the Outcome-Based Pricing Bubble

Posted by

When generative AI arrived via ChatGPT two years and six days ago, it quickly upended everything from coding to data center spending (see exhibit 5) to the power grid.

Everything except pricing, that is. The first ChatGPT subscription plan, introduced scarcely two months after ChatGPT itself, sold at per user per month rates like pretty much everything else delivered across the cloud. Anthropic, Google, and Perplexity subsequently adopted that same familiar model.

AI pricing has grown more interesting over time, however. Intercom, to name a much-cited example, charges $0.99 every time its customer service agents resolve an issue, and precisely nothing when they don’t. “Automated resolutions” from Zendesk, similarly, cost $1.50 to $2.00 apiece once you’ve used up a monthly allotment included in your subscription. Salesforce, which expects to have over a billion agents in use by this time next year, bills its customers a flat $2 for every conversation one of those agents performs.

And really, if you think about it, makers of AI-powered software often have no choice but to embrace “outcome-based pricing” schemes like that. It’s hard to grow sales of a solution priced per user, after all, if the whole point of buying the solution is to reduce the need for users.

“The better your software is, the fewer users there will be and the less money you’ll earn,” observes Steven Forth (pictured), managing partner and CEO of SaaS pricing consultancy Ibbaka.

The question I (along with a whole lot of other people) keep asking is will outcome-based pricing in AI inspire end users to demand outcome-based pricing everywhere. In theory, the answer is yes. Who, after all, will want to pay for mere access to software by purchasing seats once they’ve seen it’s possible to pay only for results instead? In reality, all we can say for sure about SaaS pricing in particular going forward is that it’s going to be a lot less uniform than it is today.

“Right now, about seventy to eighty percent of SaaS companies have some form of user-based pricing,” Forth says. “In three years, I suspect that will be significantly under 50%.”

Thanks for reading Channelholic! Subscribe for free to receive new posts and support my work.

Access versus usage versus results

Fundamentally, there are three ways to pay for software.

“You can pay for just access to it, which is basically what seat-based pricing is, you can pay for how much you use it, which is what consumption-based pricing is, or you can pay for the results you get out of it, which is what outcome-based pricing is,” says Sam Shinner (pictured), global partner and managing director of strategic advisory firm L.E.K. Consulting. “I can’t really think of a fourth way.”

Ultimately, all three models strive to align the money buyers spend to the value they receive in return. The trick, of course, is measuring that value.

“The easiest proxy measure for most people to wrap their heads around is the number of users,” Shinner says, which is why user-based pricing is both extremely popular and unlikely to disappear any time soon. “It’s just so simple and intuitive and easy.”

It’s also, however, poorly correlated to value in a lot cases. Indeed, something like 40% of SaaS seats deliver no value because they never get used.

Enter consumption-based pricing, which buyers of cloud-based storage capacity and compute cycles in particular tend to like. “They only pay when they use it, which is hopefully when they’re getting value,” Shinner observes.

The problem, of course, is that sometimes they don’t get value. A company that uses a marketing automation product a lot but generates no leads will probably end up frustrated.

Which brings us to outcome-based pricing, arguably the only viable model for some forms of AI software and a compelling model in other categories as well—for everyone involved. Vendors like it because it usually results in greater revenue. People buy software assuming it will provide value, according to Forth, but can’t be sure it will, which limits how much they’re willing to pay for it.

“I refer to that as the risk discount,” he says, adding that outcome-based pricing eliminates it, and that end users usually don’t mind. “I’m happy to pay more if it’s a result that’s valuable to me and I can quantify that value.”

Structuring an outcome-based pricing model, however, is harder than it sounds. “You have to satisfy three criteria,” Forth says, beginning with a clearly defined and readily measured outcome, like Zendesk’s automated resolutions.

“It sounds obvious, but for a lot of software out there, the outcomes aren’t really crisp,” Forth says.

Clear that bar and buyers and sellers alike still need some basis for both forecasting how often a given outcome will happen and attributing results. Bill salespeople every time they close a deal, for example, and they may claim their persuasiveness rather than your software deserves the credit.

All of that complexity, Forth predicts, will prevent more than perhaps 20% of vendors from eventually adopting outcome-based pricing. Shinner agrees.

“Outcome-based pricing will remain niche,” he says. “Usage-based pricing is going to grow a lot.”

Both models, however, are a significant departure from seat-based pricing, and a serious threat to vendors that fail to recognize what customers increasingly want. Continue charging for the possibility of value when others in your market are charging for its realization instead and…well, Forth sums it up nicely.

“You’ll be screwed,” he says.

Learning from Senteon

It’s been about three months since I last wrote about ConnectWise’s PitchIT contest and about four weeks since I watched security startup Senteon claim this year’s $70,000 first prize on day one of the vendor’s IT Nation Connect event. I had to dash out of the room immediately afterwards to watch ConnectWise CEO Manny Rivelo make his first conference presentation, so I didn’t speak with co-founder and Chief Customer Officer Zach Kromkowski (pictured left, celebrating) about his triumph then or later at the show.

This week, thankfully, I finally got a chance to compare notes with him and Sean Lardo, who runs PitchIT for ConnectWise, about the lessons other startups can learn from Senteon’s success. I count at least five:

1. It’s good to be in security. Especially if you make a solution tightly targeted at a specific use case, which according to Lardo many startups in security and beyond don’t.

“They’re very convoluted in what they actually provide,” he says. “They almost become a Swiss Army Knife of everything.”

Senteon, by contrast, provides one service a lot of MSPs need: continuous scanning of managed endpoints for security issues and automated resolution. “There are so many tools and platforms on the market today that tell you what’s wrong,” Kromkowski says. “We actually fix the things that other tools only tell you are broken.”

2. It’s even better to be in security if you can help people with compliance. Senteon bases its assessments and remediations on CIS standards, which align closely with regulatory requirements generally and cyber insurance requirements specifically.

“A lot of the insurance firms, at least the more innovative and forward-thinking ones on security, are aligning to CIS,” Kromkowski says. Indeed, DataStream (an insurer you’ve read about here before) is partnering with Senteon based on its embrace of CIS best practices.

“They’re providing verified mappings of the settings they need in place to get better approval rates and lower premiums,” Kromkowski says.

Significantly, and probably wisely, Senteon also partners with Compliance Scorecard (another company covered here before) rather than take on the huge and complex burdens associated with providing full-blown CIS governance, risk, and compliance services.

“We’re only covering the technical policies of that framework,” Kromkowski says. “We’re not a GRC tool and we’re not covering all of PCI or all of HIPAA or all of CMMC.”

3. Evangelizing your solution well is just as important as raising capital, writing code, and finding product-market fit. And maybe even more so, Lardo says. Having a great value proposition got Senteon into the PitchIT competition. Articulating it with clarity and especially passion won it all that prize money.

“Storytelling 101 is extremely important for everybody,” Lardo says. “You can have the greatest product in the world, but if you’re not excited about it nobody else will be excited about it either.”

4. Relationships matter. One of the problems with product-led growth, I’d argue, is it can leave customers feeling like they have a relationship with a website rather than a person. Senteon, according to Lardo, understands how forging real relationships can turn partners into advocates. “They’re a very human company,” he says.

5. It pays to have a little faith in yourself. Kromkowski and his colleagues were initially skeptical they’d even make it into the PitchIT competition, let alone win it. Lardo sees that kind of thinking a lot. “That imposter syndrome runs wild in entrepreneurship,” he says.

Fighting through it, however, got Senteon more than just a hunk of money to invest. “In 16 weeks, which isn’t that long, we’ve nearly doubled in size, which is absolutely awesome,” Kromkowski says.

Share

Three things from AWS re:Invent

I wasn’t there, I should note, and couldn’t muster the time, energy, and courage required to swim through the tsunami of news from Amazon’s giant cloud conference. Three items managed to catch my attention anyway:

1. AWS has a Buy Now button. Actually, it’s a “Buy with AWS” button, but it sounds a lot like the Buy Now button that Pax8 has coming. The idea is to let the hyperscaler’s over 5,000 sellers tap into some of the $157.7 billion worth of cloud commits businesses had on account with AWS as of Q1, according to Canalys, by connecting customers straight from their website to the AWS Marketplace. One more variable for the channel to consider when placing marketplace bets.

2. Amazon (finally) has LLMs. Speaking of marketplaces, AWS launched one for AI foundation models this week, including a brand new family of its own models called Nova. Based on at least one observer’s first impressions, Nova’s three flavors are in the same ballpark capability-wise as alternatives from OpenAI, Anthropic, and Google but way cheaper.

3. AWS has discovered leading with the channel! Prepare to do a lot of scrolling to find the details, but buried deep in a roundup of partner news was word of what AWS calls “the first Partner-led channel sales motion” in its history, named the Small Business Acceleration Initiative.

“Moving forward, AWS Partner Territory Managers will provide dedicated support to Partners serving these customers and we will drive customer awareness through AWS demand generation channels- passing leads to SMB Competency Partners and distributors.”

The aforementioned competency first arrived in January, and looks to target a small-ish set of large-ish companies at present. Could it and this new sales initiative be a beachhead for something broader too?

Also worth noting