Fieldbook

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

~ the obit ~

The most widely used database in the world is a spreadsheet, and a spreadsheet is a bad database. It has no idea that the word in one cell is the same thing as the word in another, so you copy the customer name into forty rows and then spend a morning finding the four where you spelled it differently. Fieldbook's whole pitch was to fix that without making you learn anything. Rows could link to other rows. Columns had types. Everything you built was readable over a REST API the moment you built it.

Jason Crawford started on the idea in 2013, settled on what the first version should be that October, and raised a seed round early in 2014. Ben Bernard came in as CTO once the prototype existed. The public version went up on Show HN on 4 November 2015, and the post that actually landed came six weeks later, on 17 December, when the pitch was the database and the API together. That one got 335 points and the comment thread you want: people describing the exact spreadsheet in their own company that had turned into a liability.

I never had a Fieldbook account. I came to it through the postmortem instead, years late, by the same route most of these names reach me. Fieldbook had no Turkish moment and not much of an American one either. What survived it is the exit document, and the figure sitting in the middle of that document is the reason the name is worth keeping.

~ the rap sheet ~

~ the cracks ~

The problem was not the product. It was that the people whose spreadsheets had become a liability did not know that what they wanted was a relational database, and the people who did know were already in Postgres and not shopping. Fieldbook had to teach the concept before it could sell the tool, and teaching is slow and expensive at twenty dollars a month.

Airtable was building the same category at the same time with more money and a louder pitch about templates rather than schemas. By early 2018 Fieldbook was under sixty thousand dollars in annual recurring revenue, with the larger seed round from 2016 running down. Crawford put it plainly in the postmortem: he could not see how to bend the growth rate fast enough to justify the company continuing to exist.

~ the last friday ~

Crawford posted the notice on 11 May 2018 and named the final day as Friday 15 June. The way users were let out of the building was built properly, and that is not how these endings usually go. One click packaged every book you had into a single ZIP: an Excel file next to a JSON snapshot that carried the full structure, down to the column types and the formulas, with the attachments alongside it. You could also take books out one at a time or drain the whole account through the API. CSVshare, the side project that shared sheets as files, closed on the same day.

After 15 June the web app and the API both stopped, and the content stopped being reachable by any route. Flexport bought the team in August 2018 for one and a half million dollars. Somebody had already built a converter that turned Fieldbook books into SQLite files, posted to Hacker News eleven days before the shutdown notice. That timing only looks clairvoyant from here.

~ the inheritance ~

The category Fieldbook was early to is now worth billions and belongs to Airtable and Notion. Fieldbook's own contribution to it is a document rather than a product: a founder writing down a revenue figure that most people in his position describe as traction, and then saying out loud that it was not enough. Under sixty thousand dollars a year, after five years and two seed rounds, with a working product and happy users.

@}-->--

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

« previous  ·  random  ·  next »