The pitch was a new digital music store built around quality. Lossless files, 320-kilobit MP3s, and access to the component stems inside a track. I was interested.
My first practical question was: how many tracks are we talking about?
I had left Beatport in August and moved to New York for Peek, a small company making a dedicated mobile email device. The jobs were different, but my head was still full of music catalogs. At Beatport, the storefront was only the visible part. Behind it were ingestion, processing, search, metadata, storage, bandwidth, and a constant argument with growth. A catalog is a product, but it is also a pile of files that has to fit somewhere and move on time.
The new idea was called Soundfiber. The plan was to sell high-quality major-label audio and eventually stems, not only to DJs but to producers, remixers, audio professionals, and people using tools like GarageBand, Ableton Live, Pro Tools, and Logic. There were possibilities for consumer retail, recommendations, editorial, multiple interfaces, and business customers. There were also a lot of very large audio files.
So I wrote a long list of questions.
What experience did the customer actually need? Was this a simple store with exceptional search, or a tool for viewing stems, tagging them, mixing them online, and sharing the result? Would full tracks and stems live in the same product? Could the same catalog support several storefronts for different kinds of customers? Was there a business selling content or infrastructure to games, advertising, television, and film?
Then I got to the part that looked familiar.
Network and servers were straightforward enough to plan. Storage and content processing were going to decide whether the budget worked. Would the company ingest the entire catalog on day one? What formats would arrive? Would the system need to manipulate and schedule the audio the way Beatport had? How much material was actually coming from the first label?
The estimate in my head was more than 30 terabytes and less than 100, with the important caveat that I might have been remembering the MP3 number. Lossless audio could change the calculation. I asked for the track count so I could start working through storage costs, "if nothing else for my own entertainment."
I wanted enough information to build a real estimate.
I had spent three years at Beatport watching an online music business turn into an international infrastructure problem, in the best and worst senses of the word. When I joined in 2006, I came in as a senior systems administrator. I was promoted in early 2007 and took on the technology function as the company grew. By the time I left, I had moved well beyond the nine web servers and five database servers I had described in a 2007 job pitch. The resolution of the questions changed with the size of the system.
The 2009 exchange also covered customer size, conversion, licensing, recommendation, community, retail economics, and why a major label would participate. The size and shape of the catalog would determine what customers could do, how quickly they could do it, and what every transaction cost.
I was also asking another direct question: when did they want me involved?
I had just moved to New York and taken a demanding startup job, and I was already considering another idea. The discussion moved from audio quality into a product outline, a storage model, possible interfaces, and a possible role for me.
I do not know whether Soundfiber will become a company or what my role will be. For now, I want answers to the basic product, catalog, licensing, and infrastructure questions before the team commits to receiving the full catalog.
The next step is to get the track count and file formats from the label. With those numbers, I can estimate storage, processing, and bandwidth costs and compare them with the proposed customer experience. I also need to know when the group expects me to become involved beyond this early discussion.