This Ugly Website Does $106M/Year (866,000 Pay For It)

Managed about 4.8 billion keywords with about 35 million pages, covering forms and documents SMBs search for. Worked with keyword clustering, a semantic core and discovery of new keywords; most keywords are unproductive, so the work is finding which clusters perform and covering as many of them as possible. The same customer-acquisition technology is shared across pdfFiller, signNow and DocHub.

Because AI and search engines need well-structured information and the existing content is messy, the team is reducing what it published before and making the rest more productive. Guest's view: quality over quantity, and real sources like Wikipedia and Reddit will matter more than the number of landing pages.

US Legal Forms, about 85,000 forms hand-built by lawyers, earns ~$6M itself but mainly sends traffic and value to pdfFiller. Its thousands of templated form pages pull in search traffic, and the form data is used to build new products such as the AI agreement drafter.

Built bizdraft.com, an AI tool for SMBs that creates legal agreements, with an interface that combines editing and AI, using the lawyer-built forms as data. Not publicly announced; most likely will not be monetized and instead used as a traffic generation engine.

With the first outside money (2015), pdfFiller bought signNow when it was doing about $3M a year, rebuilt the entire product and applied ('weaponized') its marketing machine to it. pdfFiller's cash funded it.

pdfFiller grew to about $18.5M in revenue before its first outside capital (2015). It was run as a cash machine whose profits were invested in other products. Guest's advice: raise as late as possible, don't pay yourself much, don't spend, preserve cash, and think about exit from day one.

Acquired brands keep their names. A small set of common services (marketing, devops, payments) and the same customer-acquisition technology are shared; everything else is siloed, with each product having its own GM, development team, roadmap and tech stack. An AI Labs group builds common AI pieces, but products are not required to adopt in-house tech. Reasoning: the brand and top of funnel differentiate, not tech, and economy of scale no longer works with AI.

Used high-level math for PPC bidding while competitors were unsophisticated. Guest says the edge is shrinking because everyone now has AI tools like Claude.

Acquisition targets must show product-market fit and domain expertise, and have more than ~$5M, ideally $10M, revenue so they make a significant contribution; $1M products are not interesting.

Plan to go from $200M to $400M: 20-30% growth from existing products, the majority from new AI products. Stay horizontal while verticalizing: the product morphs to the user segment (small medical office, pool contractor) and handles their documents, compliance and permits. No rush; wait for the feedback loop before scaling.

Guest says he has tried the tactics gurus sell for influencing ChatGPT results and they do not work; models are getting smarter and unpredictable, so the answer is adaptability, product and reputation.