Owner says every login is a failure. That breaks the metric your board underwrites.

Adam Guild took the SaaStr AI stage and said that at Owner, a restaurant software company he frames as Shopify for restaurants, a customer logging in is a failure signal: if someone is in the dashboard manually fixing what the software did, the software did not do its job ^{[1] [2]}. He extended that to the metrics themselves, arguing DAU, WAU and MAU should go down as the product starts working ^[3]. SaaStr's own summary concedes the consequence: this is the conclusion that will make your board deck look worse ^[4].

The claim gets a hearing because the revenue line did not cooperate with the usual story. Owner says more than 83% of new customers now start inside its free AI product, up from 0% two years earlier ^[5], a shift of more than 83 percentage points in the acquisition mix ^[6]. The company reports faster growth YTD 2026 and in 2025 than in 2024 ^[7], and describes itself as accelerating after $100M ARR ^[8].

Set those two things next to each other and the standard diligence panel stops working in both directions. Engagement falling is supposed to be an early warning; here it is presented as evidence of function ^[3]. Growth holding up is supposed to be reassurance, and Guild's fourth takeaway is that growth metrics are a lagging indicator of a platform shift, with Owner's numbers excellent the entire time the ground was moving under them ^[9]. That is the awkward pair for anyone underwriting a vertical SaaS business on a session count and a net revenue retention curve: low engagement can mean the product works, and strong growth can mean nothing yet.

The SaaStr account does not name the replacement denominator, but the moat argument points at one. Guild's position is that agents only pay off on top of an extremely opinionated product, where users get the outcome without configuring anything ^[10], and that the opinionation is what insulates Owner from Claude and ChatGPT: a foundation model can build a decent restaurant website, but it cannot know which components on that specific site correlate with sales growth, and you only generate that data by enforcing one system across the base rather than letting every customer configure their own ^[11]. That is a customer-outcome metric, measured on the restaurant's revenue, not on the vendor's logins. A board that accepts the login argument has to fund the measurement of the outcome instead, which is a harder and slower instrument than a product analytics dashboard.

Worth noting who was against this. According to the account, investors called it CEO thrash and wanted the new direction layered in gradually rather than funded with millions and reassigned people ^[12]; Owner's own PMs and engineers pointed to 100 customers asking for things already on the roadmap ^[13]; and restaurant industry veterans said owners were terrified of AI and a pivot would alienate the base ^[14]. Guild's read was that the pressure from AI-native startups doing generated websites and AI phone ordering ^[15], plus public incumbents putting hundreds of engineers on clones and distributing them into installed bases of hundreds of thousands of similar customers ^[16], would have Owner bleed out slowly with no business in a few years ^[17]. None of it was visible in the numbers at the time ^[9].

What to watch: this is one founder's account of his own company at a conference, with self-reported figures and no cohort or retention disclosure. The testable part is whether Owner publishes a customer sales-growth measure to sit where engagement used to sit, and whether the 83% free-AI entry point converts to paid at a rate that survives contact with a diligence process. Guild's related line, that "same plan, smaller team" is a choice about ambition rather than an inevitability ^[18], is the one most likely to be quoted by people who mean the opposite.