Crocodoc

» see crocodoc.com as it lived, on the wayback machine

~ the goodbye ~

Crocodoc exists because four people gave up on their first company over a weekend. The first company was WebNotes, started out of an MIT dorm room by Ryan Damico with Bennet Rogers, Matt Long and Peter Lai, and it let you highlight and annotate text on any web page. Damico's own description of how it went is a spectacular failure, on the grounds that nobody would pay for it. The piece of it that worked was the document rendering underneath, so they threw away the product and kept the engine. By Monday they were a different company.

What the engine did sounds small and was not. Hand it a Word file or a PowerPoint deck or a PDF and it would give you back a web page: text converted to HTML and images converted to SVG, with the layout held together in CSS. Everyone else in 2010 solved this by rendering each page as a picture, which gave you something you could look at and not something you could select or read on a phone. Crocodoc gave you a real document in a browser tab with no plugin under it, at the exact moment the industry was walking away from Flash.

That made it infrastructure rather than a destination, and the customer list is the proof. LinkedIn used it, and so did Yammer and Blackboard, along with Edmodo and SAP, which means an enormous number of people opened a document through Crocodoc in those years without the name ever appearing on their screen. The company put its lifetime numbers at more than 200 million documents converted and 14 million annotations made. It was a Y Combinator company, and the cap table ran through SV Angel and a row of individual names including Paul Buchheit and Joshua Schachter.

The distance I have from this one is about what a working day produces. Crocodoc was built for a world where the artifact under discussion is a document, where a draft goes around by mail and comes back with marks on it, and where the tool that matters is the one that lets six people write in the margin. That is real work and it is a real category and it is not the side of the desk I have been on. The material I have spent my career arguing over lives in a repository, and the margin note is a comment on a line of code with a hash next to it. Crocodoc solved somebody else's review problem very well.

~ the chronicle ~

~ a document is not a picture ~

The technical bet was that a document should arrive as text. It sounds obvious now, and in 2010 the entire industry was doing the opposite: Scribd was on Flash, Google's viewer produced page images, and every enterprise system that showed you a PDF was really showing you a stack of PNGs. That works until you want to copy a sentence out of it, or open it on a phone where a full-size page image is unreadable.

Getting from a binary document format to clean HTML is much harder than it sounds, because the formats it had to eat are decades of accumulated exceptions, and a PowerPoint deck is not a description of a page so much as a set of instructions for drawing one. Crocodoc's conversion pipeline was the whole company, and it is also why the acquisition made sense: Box did not want a document viewer, it wanted the years of work that made the viewer come out right.

~ what box actually bought ~

Box announced the acquisition on 9 May 2013 for $13.2 million in cash and stock, and the point of it was preview. A cloud storage product lives or dies on whether clicking a file gives you the file, and Box was competing with Dropbox and with Google at a moment when everybody's preview was mediocre. Damico became a director of platform at Box and ended up running content services there, which is an unusually good outcome for a founder in a purchase of this size.

The technology went where it was bought to go and became Box Preview, sitting under every document view in the product. What it stopped being was a company anyone could buy from. Box kept selling the engine for a while as the Box View API, so outside developers could still put a viewer in their own products, and that was the last public trace of Crocodoc as a thing with its own name.

~ two endings, three years apart ~

The first ending came on 1 November 2015, when Box switched off Crocodoc Personal along with WebNotes, the annotation tool the founders had started with in the first place. That is the consumer-facing half gone, and it went quietly, since by then almost nobody outside the API customers thought of Crocodoc as something you visited.

The second came on 15 January 2018, when Box put both Crocodoc and the legacy Box View API to end of life in favor of a rebuilt New Box View. Developers who had wired the old API into their products found the replacement did not match it feature for feature, and the support forums from that year are full of people asking where a specific capability went. crocodoc.com does not resolve in 2026. The pipeline is still running inside Box, still converting documents nobody credits it for, which is the closest thing to survival a component company gets.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »