Potion.so, a one-person wrapper that turned Notion pages into hosted websites, sold for $300,000 in April 2023 after shipping roughly 4,000 customer sites on Vercel. Vercel’s case study tells the story as a platform win. The part it leaves out: managed hosting has collapsed the cost of building a product like this down to distribution and operations, which is precisely the part an acquirer will pay for.
What was Potion, actually?
Potion was a thin Notion-to-website wrapper built by solo founder Noah Bragg on Next.js and Vercel, not a deep-tech product. The model was simple: connect a Notion workspace, and Potion published it as a real website on a custom domain. The technical core was largely off-the-shelf. Bragg built on Vercel’s Platforms Starter Kit with wildcard subdomains, server-side rendering, and incremental static regeneration (ISR), the same scaffolding Vercel hands anyone building a multi-tenant site. ISR is what let one database serve a large number of sites without melting, by regenerating pages on a schedule instead of on every request. At its peak the product served 4,000 custom domains and more than 100,000 monthly pageviews, run by one person with no employees.
The engineering was not trivial. Wiring custom domains, subdomains, and ISR at that volume is real work, and Notion’s API has its share of sharp edges. But the hard problem had migrated off the application and onto the host. A “Notion-to-website” tool is a crowded category, and several have come and gone. What made Potion sellable was not the code; it was everything around it.
How much did Potion sell for?
Potion closed at $300,000 in April 2023 on roughly $6,300 in monthly recurring revenue, a 4.0x multiple of its $75,600 annualized run rate. FounderSold records the deal at a 4.0x revenue and profit multiple, above the 2.8x median across the 51 sub-$1 million exits it documents, after a build of roughly 2.4 years listed on Acquire.com. At sale the product had about 20,000 monthly visits, 500-plus paying customers, and zero employees, according to They Got Acquired. Bragg had automated support down to about two hours a week.
One wrinkle on the date: Vercel’s case study gives the acquisition month as June 2023, while the exit databases and Bragg’s own June 2, 2023 announcement tweet put the close in April 2023. The deal closed in April and was announced in June; the Vercel figure is the announcement, not the close.
What did Vercel actually absorb?
The host ate the parts of running a web product that used to be a full-time job: infrastructure, autoscaling, and the custom-domain plumbing. Vercel’s case study credits its Platforms Starter Kit, wildcard subdomains, SSR, and ISR with letting one person serve 4,000 custom domains and 100,000-plus monthly pageviews with no operations team. ISR is framed as the specific lever that cut database load, which is the practical difference between a wrapper that scales for free and one that falls over on its first traffic spike.
This is the half of the story Vercel has every reason to tell, and it is accurate as far as it goes. What the vendor framing skips is the corollary: the same managed-hosting layer that made Potion cheap to run also shrank its defensible surface. When infrastructure, operations, and scaling are someone else’s product, the only things left to sell are the customer list and how well the business runs without its founder. A host cannot supply either of those, which is exactly why they are worth money.
What did the buyer actually pay for?
Bruno Morency, a Montreal developer and former managing director of Techstars Montréal AI, bought Potion for rising recurring revenue and a bet on Notion’s ecosystem, not for its technology. It was the first acquisition in a planned two-businesses-per-year portfolio, and Morency cited steadily rising MRR and the Notion bet as the draw. Bragg told They Got Acquired that transferring the Stripe account was the trickiest part of the handoff, a financial-plumbing detail rather than an engineering one.
The buyer was paying for low-maintenance recurring revenue. By sale time Bragg had automated support to about two hours a week and made the product self-serve with docs and guides, which is the exact shape of asset an operator-investor wants: a customer base that, in the language of the 2026 valuation frameworks, does not know who built it.
That asset is built almost entirely by distribution. Bragg constructed Potion in public on Twitter, where his first “building a SaaS in public” tweet brought 1,000 followers in 24 hours. The product landed its first 75 paying customers within two months, “pretty much all” of them from Twitter, before a Product Hunt launch hit number one Product of the Day and Product of the Week and earned a Maker Grant. The acquisition price was set by that distribution and by the operations automation behind it. Vercel does not appear among the buyer’s stated reasons.
What would a wrapper like this fetch in 2026?
Two 2026 valuation frameworks put a flat-growth SaaS wrapper at roughly 2 to 4x ARR, with churn and sole-founder dependency as the levers that move the number. ExitLoop Club’s March 2026 bands run 2, 4x ARR for flat or slow-growth SaaS (under 5% monthly growth), 4, 6x for growing (5, 15%), and 6x-plus for high-growth (15% and up). Acquiry’s 2026 data is more conservative, placing traditional SaaS under 15% growth at 1.5, 3x ARR.
Potion’s 4x was 2023 data and sat above the indie median at the time. The 2026 bands suggest a comparable flat-growth product would not automatically fetch more today; the floor for slow-growth SaaS has if anything softened to 1.5x at the low end. What has hardened is the diligence. The lifters and killers from ExitLoop map almost one-to-one onto the Potion case. The multiple-lifters most relevant to a wrapper are recurring revenue under 3% monthly churn, a clean documented codebase, operations that run with no manual work, a customer base that does not know who built it, and a six-month growth trend. The killers are their mirror image: no handoff docs, no ops transition plan, customer concentration, and sole-founder dependency. Potion cleared the first list by the end and skated close to the second on founder dependency, the standard risk for any one-person business.
So is the default outcome an acqui-exit now?
The realistic read is that a managed-host wrapper’s default outcome is an acqui-exit rather than a durable independent business, because the moat has shrunk to the two things a host cannot provide: distribution and operations discipline. When infrastructure, scaling, and custom-domain plumbing belong to Vercel or its competitors, the sellable asset is a customer list plus a system that runs without its founder. That is precisely what a small buyer will pay 2, 4x ARR for, and it is thin enough that few of these businesses grow into something the founder wants to hold long-term.
The economics this implies are uncomfortable for the independent-company framing. A wrapper that one person can stand up on managed hosting in a few months, and that realistically exits at a low single-digit multiple of a modest ARR, is not a company in the venture sense. It is an annuity with an exit option, and the exit option is the most likely way the founder realizes its value. Planning for that outcome from the start, rather than discovering it after two years of growth, is what changes the build.
This does not make the exit easy; it relocates the difficulty. Distribution is the part Vercel cannot ship you, and the Potion case shows how much of it is just sustained public work: roughly two years of building in the open before the first 1,000 followers compounded into a paying customer base. Automation is the other half, and it is a discipline, not a feature: document the codebase, kill manual support, write the handoff plan. The 2026 diligence trends point the same direction. Acquiry flags gross-margin and inference-cost scrutiny as the new focus for AI-native deals, which is the updated version of the question Potion answered in 2023. When the host absorbs the infrastructure, the buyer’s diligence migrates to the only things left: who uses it, how sticky they are, and whether the product still runs the morning after the founder walks away.
Frequently Asked Questions
Potion sold at 4x ARR. Why above the flat-growth band?
Vercel’s case study records Potion’s revenue growing 6% month-over-month, which places it in ExitLoop’s ‘growing’ tier of 5-15% monthly growth (the 4-6x ARR band), not the under-5% flat tier at 2-4x. The multiple tracked growth velocity rather than the stack, so a stationary product at the same $6.3K MRR would today land at the lower end.
Does the playbook work on a non-Vercel host like Cloudflare or Netlify?
The structural argument holds for any managed host that absorbs infra, autoscaling, and custom-domain plumbing. What changes is the scaffolding: Vercel’s Platforms Starter Kit handed Potion wildcard subdomains and ISR out of the box, and a host without equivalent multi-tenant primitives shifts that wiring back onto the founder. That work is precisely the cost the acqui-exit thesis assumes the host has already eaten.
Why was the Stripe handoff harder than the code transfer?
Stripe accounts bind to a legal entity and its risk profile, so a buyer cannot simply re-point payments; the acquirer must either assume the existing account through Stripe’s transfer process or migrate 500-plus subscribers onto a new account and pass fresh KYC. Every recurring charge is in transit during that window, which is why a subscription handoff turns on financial plumbing, not engineering.
What’s the specific failure mode for a Notion-to-website wrapper?
Platform dependency. The product bets that Notion leaves the publishing layer to third parties, so a single Notion API change or a first-party publishing feature can gut the value overnight, and several Notion-to-website tools have already folded. The sellable pieces (customers, MRR, automation) survive, but the acquirer is underwriting Notion’s continued restraint, which the wrapper cannot control.
Would an AI-native version of this face harder diligence?
Yes. Acquiry’s 2026 framework adds gross-margin and inference-cost scrutiny to AI-native deals that Potion’s 2023 sale never faced, because a wrapper reselling LLM calls carries per-query costs that compress margin as usage climbs. A buyer would demand unit economics, cost per query, and gross margin after inference before pricing, on top of the NRR and churn checks that already apply to any subscription business.
