I went through a week of podcasts talking about SaaS. Seat pricing had a rough week
Hey,
One of my slightly nerdy hobbies is pulling podcast transcripts on a topic and reading what people actually said on air. Founders and investors are a lot more candid on podcasts than they are in blog posts, and almost none of it gets written down anywhere.
This week I’ve been looking at SaaS. From 19 to 24 September I tracked every mention of SaaS, churn, ARR, MRR, seat-based pricing and vertical SaaS, plus a handful of SaaS shows followed in full. That came to 114 mentions across 92 different shows. I analysed all of them, and pulled out the bits I think are actually useful if you're running a SaaS company.
TL;DR:
- Seat-based pricing had a terrible week. At least seven shows talked about moving to usage or outcome pricing, including one founder who had to abandon seats because his heaviest users used 10,000x more than his lightest.
- AI is squeezing the classic 80% gross margin, and at least one show said vendors are quietly eating the compute bill to stop customers churning.
- The fundraising bar moved. Miro sold for $1.35bn after a $17.5bn valuation, and "triple, triple, double, double" no longer gets you a meeting.
- Plenty of companies are still growing fast, and a few of them are tiny.
- Retention was the number everyone came back to, especially anyone thinking about selling.
Methodology (pinch of salt)
- 19–24 September, only shows that get transcribed. It counts mentions, not minutes, so a passing "we're a SaaS company" counts the same as a full segment.
- The "churn" tracker also picked up sports roster churn and a school board. I ignored those; they're not in anything below.
- Quotes are from automatic transcripts, lightly cleaned. Speaker names come from the transcript too, so I've only named people where I could confirm who was talking.
- Six days is a small sample. Treat this as what came up this week. It's too short to call a trend.
1) Seat-based pricing had a terrible week
This was the loudest thread by a distance, and the best example came from someone who actually tried seats and backed out. Keith Peiris runs Lightfield, an AI-native CRM. They launched on pure seat pricing because that's what buyers expect from a CRM, and it went down well with customers. Then they looked at usage:
"We started with pure seat pricing because it just matched the sort of Salesforce HubSpot world. It was I think it was received really well by our customers, but then the head was using 10,000x more than the tail. Using in terms of consumption. Yeah, in terms of consumption, and we weren't going to be in business for a long time with per seat pricing."
Keith Peiris (CEO, Lightfield) on The a16z Show
They moved to a hybrid model: a platform fee for access, with the AI work (forecasting, automation) priced separately. On a different show (Topline) the same week, he claimed their agentic CRM now lands deals from 200-person companies that look like Salesforce deals from 5,000-person ones.
Wade Foster (Zapier) made the general version of the argument:
"I think it may still make some sense in some small areas, but by and large, when intelligence is such a core part of these product experiences, I don't see how a fixed seat-based pricing mechanism really makes much sense for that at all."
Wade Foster on The Cognitive Revolution
SaaStr put a number on it: two in three companies on the Forbes AI 50 have some form of usage-based pricing. SaaS Therapy, The HighLevel Experience and Venture with Grace all covered the move to outcome-based billing (charging per resolved ticket or completed task, for example).
The one counterpoint came from the incumbents. On SysAdmin Weekly, Microsoft was described as raising prices on core products while adding pricier AI tiers, and keeping it complicated on purpose:
"Microsoft has no intention of making this simpler overall because this complex licensing landscapes makes it possible for them to do complex enterprise agreements where they control they have a lot of knobs when you're negotiating the next three years of your licensing agreement."
Paul Schnackenberg on SysAdmin Weekly
My take: if your heaviest customer uses orders of magnitude more than your lightest, seat pricing already means your small customers are subsidising your big ones, AI or not. The Lightfield number is the one I'd go and check against my own usage data.
2) The margin squeeze nobody puts on the pricing page
The classic SaaS model assumes around 80% gross margins. The SaaS CFO and Eli the Computer Guy both covered AI-native and agentic products breaking that assumption, because every AI action costs real compute.
The most specific claim came from MRKT Matrix:
"Vendors are absorbing high API compute costs from OpenAI and Anthropic to prevent enterprise client churn, creating margin compression across the SaaS sector."
MRKT Matrix
And TBPN talked through Harvey, the legal AI company, growing to $400M ARR with negative gross margins, while token consumption went up 20x this year as they moved to agentic workflows.
I don't know how common the "eat the compute cost to keep the logo" move really is. It was one show's claim, so pinch of salt. But if you've added AI features to a flat-price plan, it's the first thing I'd check in my own numbers.
3) The fundraising bar moved
20VC spent a chunk of an episode on Miro selling for $1.35bn after a $17.5bn valuation, and on multiples shifting from revenue to EBITDA:
"Death comes for us all in SaaSland, right? And it was exactly right. It was just an inevitable cleanup operation because it was just so far wrong in terms of pricing."
Rory O'Driscoll on 20VC
On This Week in Startups, Jeff Clavier (Uncork Capital) was blunt about what no longer works for a Series A:
"If you show up with a triple, triple, double, double, double, which guaranteed you financing three, four years ago, right now, nobody cares."
Jeff Clavier on This Week in Startups
The point of that episode was that growth alone doesn't raise money any more, and investors want to see sustainable margins behind it.
The small-company end said the same thing differently. On the Buying Online Businesses podcast, Dev Shah talked about multiples dropping for smaller SaaS acquisitions (good news if you're buying, bad if you're selling), and MicroConf On Air had PE buyers prioritising stable revenue over fast growth.
4) Meanwhile, some people are growing just fine
It wasn't all doom. The growth stories that came up:
- Clay: CEO Kareem Amin confirmed on The Information's TITV that they're targeting $240M ARR this year and plan to double it next year.
- Vanta: $300M ARR, and apparently won 600 customers without a website (Brydge Club).
- Juicebox: $1.5M ARR with a team of three (Knuckle Up with Nakul).
- ZeroRank: mid-six figures ARR within four months (Niche Pursuits).
- SaaStr (the company itself): doubled sponsorship revenue in 12 months by running 21 AI agents on top of Salesforce. Their inbound agent handled 17,000 conversations and booked 600 meetings, and new business and renewals were both up 60%.
If you're indie or bootstrapped, the Juicebox and ZeroRank ones are the interesting pair. Small teams are still getting to real revenue fast.
5) Retention is the number everyone came back to, especially if you want to sell
"NRR is the crucible. It is the single most unforgiving metric in recurring revenue models."
The CEO Growth Podcast
The practical version came from Run the Numbers. Rick Smith, a multi-time CFO, warned that churn during a sale process is one of the main reasons buyers cut the price after signing (purchase-price adjustments). Alex Hormozi made the point that high churn in small-business segments is often structural, and needs handling differently from churn caused by something you're doing wrong.
And the acquirer's view, from Bending Spoons:
"We have businesses with lower retention, businesses with higher retention, but I can't recall a single instance where retention got worse after Bending Spoons took over."
Luca Ferrari (CEO, Bending Spoons) on Sourcery
6) The contrarians
Not everyone thinks SaaS is dying.
"I think it's going to be like an inverse SaaSpocalypse. I think SaaS is going to be supercharged by this. They're the ones who are like most in the know of what things are valuable to automate."
Latent Space
The argument there: SaaS survives as the reliable, database-like layer that AI agents run on top of.
"SaaS has a moat when you have a point of view. ... If you're just building commodity SaaS, meaning like you're just providing features that accomplish tasks to people, that's fundamentally a commodity."
The Startup Ideas Podcast
And the one I've been thinking about most since:
"If you actually build software that's working and the people want and it solves a problem, guess how quick it is for somebody to copy you and spin up their own version of that. Right. And guess who wins? The person with the most distribution."
Katelyn Bourgoin on Attributed
Happy to pull the full list of all 114 with shows and episodes if it's useful, or dig properly into one of these threads (the pricing one has plenty more in it). Just say which.
Full disclosure: I run Pod Engine, a podcast data API, and built scoop.fm on top of it, which is what tracked all of this. I built it mostly because I kept doing this kind of digging by hand. Every quote above links back to its episode on the source page if you want the full context: https://scoop.fm/s/saas-watch/archive (you can also run your own analysis for free too which gets you a weekly email debriefing you on whatever you care about, SaaS or not, as said in podcasts)
How they grewPodcast transcript digging and a weekly email debrief
