Every week I’ll provide updates on the latest trends in cloud software companies. Follow along to stay up to date! Is Seat Based Pricing Dead? If AI delivers on its promise, it may spell the end of the SaaS business model as we know it. Historically, cloud software businesses charged a recurring fee based on the number of users of their software - the SaaS model. Then infrastructure / dev tools software companies took a different approach - more of a consumption based pricing model (I’m generalizing, not all infra is consumption based). Look at Twilio - they charge per message sent. Or Snowflake who charges based on compute used or data stored. At the end of the day, software vendors want to align value delivered with price charged. Back to the Twilio example - let’s say I, as a solo developer, built an application on Twilio that sent messages. And that app took off and was used by millions of end users sending messages every day. If Twilio charged per seat, I only would have paid Twilio for one seat, irrespective of how popular my app became. Twilio would have massively undercharged me! On top of this, every message end users received in my app through Twilio would push costs to Twilio (these are the Twilio COGS). So not only would Twilio be undercharging relative to the value I’m receiving, but Twilio would also be incurring incremental marginal costs that they’re not recognizing revenue for.
Jamin - excellent insight! I served as SVP Product and GM @ Zendesk from 2016 through 2021 in a period of rapid growth and evolution of the product and the business model, including introduction of Answer Bot - an ML-based offering on top of the Zendesk Guide knowledge base designed for automatic resolution of customer requests. We ignited fiery discussions across the team about usage (consumption) pricing and what it meant for customers and our business.
We had already acknowledged that we walked a fine line every day with seat-based pricing because:
1) Every feature we introduced to make agents more efficient reduced demand for more seats
2) Seat based pricing was subject to abuse when customers shared licenses across multiple agents
When we introduced Answer Bot, we decided to realign the value metric (resolved tickets) with the pricing model ($ per resolved ticket). The challenge was the difficulty customers (and sales) faced in predicting how many successful resolutions AB would drive. It was hard to budget for in advance and felt risky. Additionally, the resolved tickets were generally simpler requests so the marginal value (and pricing) for each was low. Ultimately, we bundled "resolved tickets" in tiers discounted by volume and gave it a go. We learned a lot, but it was too early. AI advancements have brought all of this back to the front burner and there's clearly now a mandate to find a path forward to deliver value to customers with sustainable high-margin packaging and pricing.
In the 3 years since leaving Zendesk, I've been on a journey co-founding startup building an AI assistant for meetings and general productivity. I have thoughts about your view on agents + databases and how the value in the industry may shift in this new era. (Hint: it may follow a similar but slightly different path as the relational database era!)
Thanks for sharing, David. What I have seen in the market today is charging based on “successful tickets” closed by an AI CS bot, where success is defined by the end customer when they clicks n a thumbs up / down on overall satisfaction post ticket closure. When tickets are closed with a thumbs down, a human agent personally reaches out to the customer to resolve issues and the AI company does not earn revenue off that.
The critical challenge to resolve is to ensure that pricing doesn’t drive strange behaviour in how the product is used. You want to base it on something that scales to usage without trying to second guess value created.
Interesting piece. Esp interesting point that the most valuable SW will become “databases” rather than UIs. In the medium term, a lot of organizations use legacy SW with terrible API access that makes it very hard to extract the important data in these systems (DoD in particular has a huge problem with this), but these systems are critical to their workflows. Will be interesting to see how “RPA style” AI agents will be used to interact with homegrown and legacy apps with poor data integration support to enable these apps to take advantage of productivity that comes from these AI agents. Interesting research article on these RPA agents: https://arxiv.org/html/2405.03710v1
This is SUPER interesting. One question is how will this change the way you think about ARR? Won't it become a little harder to predict, especially when you make the transition? When you make a sale, you'll have to make an assumption on how much a customer will be using and I'd imaingine that would be a little bit of a guess and you would have to do it on a rolling 12m forward basis. How would you think about that from the investor side?
Great post. This with Benedict Evans comments on the Turpentine pod make me think the Tabular acquisition has much larger ramifications than I previously imagined. Agents and models will be like databases - there will be many of them purpose built for applications and DB will abstract away all of the decision making - value will float to the top.
Jamin, your thoughts mirror what I heard last week from a group of friends whom I met …all developers from industries like financial services, healthcare, insurance and healthcare. They all confirmed that AI copilots are being used more and more for software development and they forecasted lower developer hiring in the future. Fewer programmers, fewer testers and fewer IT operations people. Aka seat based licensing could likely decline in favor of consumption as you discussed above. Cheers!
Jamin - excellent insight! I served as SVP Product and GM @ Zendesk from 2016 through 2021 in a period of rapid growth and evolution of the product and the business model, including introduction of Answer Bot - an ML-based offering on top of the Zendesk Guide knowledge base designed for automatic resolution of customer requests. We ignited fiery discussions across the team about usage (consumption) pricing and what it meant for customers and our business.
We had already acknowledged that we walked a fine line every day with seat-based pricing because:
1) Every feature we introduced to make agents more efficient reduced demand for more seats
2) Seat based pricing was subject to abuse when customers shared licenses across multiple agents
When we introduced Answer Bot, we decided to realign the value metric (resolved tickets) with the pricing model ($ per resolved ticket). The challenge was the difficulty customers (and sales) faced in predicting how many successful resolutions AB would drive. It was hard to budget for in advance and felt risky. Additionally, the resolved tickets were generally simpler requests so the marginal value (and pricing) for each was low. Ultimately, we bundled "resolved tickets" in tiers discounted by volume and gave it a go. We learned a lot, but it was too early. AI advancements have brought all of this back to the front burner and there's clearly now a mandate to find a path forward to deliver value to customers with sustainable high-margin packaging and pricing.
In the 3 years since leaving Zendesk, I've been on a journey co-founding startup building an AI assistant for meetings and general productivity. I have thoughts about your view on agents + databases and how the value in the industry may shift in this new era. (Hint: it may follow a similar but slightly different path as the relational database era!)
Thanks for your insightful comment!
Thanks for sharing, David. What I have seen in the market today is charging based on “successful tickets” closed by an AI CS bot, where success is defined by the end customer when they clicks n a thumbs up / down on overall satisfaction post ticket closure. When tickets are closed with a thumbs down, a human agent personally reaches out to the customer to resolve issues and the AI company does not earn revenue off that.
The critical challenge to resolve is to ensure that pricing doesn’t drive strange behaviour in how the product is used. You want to base it on something that scales to usage without trying to second guess value created.
Interesting piece. Esp interesting point that the most valuable SW will become “databases” rather than UIs. In the medium term, a lot of organizations use legacy SW with terrible API access that makes it very hard to extract the important data in these systems (DoD in particular has a huge problem with this), but these systems are critical to their workflows. Will be interesting to see how “RPA style” AI agents will be used to interact with homegrown and legacy apps with poor data integration support to enable these apps to take advantage of productivity that comes from these AI agents. Interesting research article on these RPA agents: https://arxiv.org/html/2405.03710v1
Thanks again for sharing!!
This is SUPER interesting. One question is how will this change the way you think about ARR? Won't it become a little harder to predict, especially when you make the transition? When you make a sale, you'll have to make an assumption on how much a customer will be using and I'd imaingine that would be a little bit of a guess and you would have to do it on a rolling 12m forward basis. How would you think about that from the investor side?
Great post. This with Benedict Evans comments on the Turpentine pod make me think the Tabular acquisition has much larger ramifications than I previously imagined. Agents and models will be like databases - there will be many of them purpose built for applications and DB will abstract away all of the decision making - value will float to the top.
Jamin, your thoughts mirror what I heard last week from a group of friends whom I met …all developers from industries like financial services, healthcare, insurance and healthcare. They all confirmed that AI copilots are being used more and more for software development and they forecasted lower developer hiring in the future. Fewer programmers, fewer testers and fewer IT operations people. Aka seat based licensing could likely decline in favor of consumption as you discussed above. Cheers!
Awesome post.