Richard Cusick


[ 00:00:19 ] I say data is like religion you know it’s an abstract concept that you need real world examples to to describe to people.

[ 00:00:27 ] I mean long story short if you can’t find content you know it has no value to you has no value to the content creator. So the way to think about it is it’s the secret sauce that makes content discoverable Yeah it’s great.

[ 00:00:45 ] It’s a great question. Yeah. There’s a practical problem with your conta creator. Look your business is not making your content discoverable right. You’re creating a piece of content. You’re not thinking hey how do I optimize my description for example of my show or the genres that I put around it to to make my content discoverable even if you do do a good job at that. Is a you know obviously tens of thousands of creators who are trying to do the same thing but all doing it differently. Right. And so what we do and grace note is we take entertainment meditatively from thousands of different sources we normalize it not only for format.

[ 00:01:25 ] And so for example a standard word length standard genres but we normalize it in terms of distribution format so that it makes it really easy for companies creating discovery interfaces search a guide on a cable box a DVR.

[ 00:01:41 ] Apple you know iTunes or Apple Music making a common format that the industry can use to make all that meta data that was really dissimilar discoverable and searchable in a really nice looking UI because the industry is so fragmented we have to be pretty flexible in terms of how we accept content.

[ 00:02:09 ] So we can be in the old days fax carrier pigeon. You know email today obviously more ex-MIL some type automated formats but we’re pretty flexible in terms of how we accept content. Our secret sauce is a combination of technology and editors actually we have twelve hundred editors around the world but we also have 500 engineers that apply technology and real human expertise to normalize standardize and maintain the quality that metadata. And then we distribute it. Obviously we have our own standard format and how we distribute them edit it. But again our clients are varied and not only would produce custom formats like photo types for them but we often distribute in custom format so you know I think if you take Experion or any other companies that you see in the data space. The real fundamental value they provide is taking messy space. They clean it up and provide kind of an independent you know service for an industry. I think that what we’re producing you if you think about music it’s album cover or the tracks. And we have unique idea. It’s a higher profile structure. So for all the nerds out there think of just like a Dewey Decimal System for for content meaning you take Michael Jackson there was Michael Jackson Thriller. Michael Jackson has a unique ID which by the way when he collaborates with somebody else another album will treat those unique ideas that are searchable thriller probably has you know not only the album but the track probably has you know thousands of different releases every country in the world. How do we display the iconic release.

[ 00:03:59 ] Right. And then the track itself how do we make sure that that track is associated with the iconic album so that when you pull up Spotify for example you’re not seeing cover art for Johnny’s wedding mixtape and you’re seeing the cover art that you expect. So it’s a kind of a basic example of what we do. It gets a little more complicated is we also have really complex descriptors that power search recommendations. So for example on TV or DVR or make it really easy to DVR game of thrones whether that’s on HBO or amazon prime or something. You know that’s an easy example. But let’s say you wanted to search for you know show me kind of the micro Zoners that you see on the carousel. Your favorite UI whether it be Gas-X Warner or Apple. Apple iTunes. You know the new Apple TV you know romantic Westerns set in Italy in the 1950s spaghetti westerns right. That will be example of genres that we would that we would create for music. It’s even more complicated than that. There are thousands that are for music taste types.

[ 00:05:08 ] And so what Apple and Spotify and other people in the music space use great for is that the deep descriptors are around different types of music so will differentiate for example between let’s say Taylor Swift her early career where she was very much a country artist and is now obviously in her later career become much more of a pop artist. Our classification system for tracks would differentiate between those two types of cat tracks and the different portions of her career.

[ 00:05:43 ] So we so far we work directly with all country creators. We have relationships with obviously all the major music labels with all major studios and networks and we work with them to capture information since it’s created by their marketing departments and we all make when just that we also have a business that actually tracks things and pre-production. So we actually do start tracking assets the minute a script is announced or movie ideas being put together we will create an entity for so we were tracking as an example you know Star Wars Roug won probably 10 years ago when it was just an idea and we’ll follow that all the way through. You know obviously when when it goes into theaters and then through its lifecycle. So yes we and I think the other thing is we are working closely with consecrates is getting early access certainly helps us get ahead of the game. The other thing is we use technology as I mentioned with twelve hundred editors. Yes clean up the content but we use technology to automate as much of the process as we can.

[ 00:06:52 ] So as an example for music we use machine learning to analyze individual music tracks. If you think about the scope theres about 250 million music tracks in the world today.

[ 00:07:05 ] We use machine learning to actually listen to a track and determine along the characteristics and tracking for characteristics like mood era genre and thats obviously with editors who train the machine but you cant do that at scale without some of machine learning you know the basis for any machine learning right is it looks for patterns. So you train it that fits these patterns. And in that beauty of machine learning obviously too is it’s thousands of patterns so you know we train the machines to look for certain patterns. It applies the patterns we them do. Obviously we continue to train it and it becomes smarter as it goes on. That’s kind of a high level how we approach it.

[ 00:08:09 ] We saw about three or four years ago that there was a transformation happening in content. Obviously the world was moving from traditional schedule based TV to streaming the music world was moving from downloads to streaming. And we saw that that discovery for consumers was going to be a real problem. And we saw that actually the traditional metadata providers the companies that power the TV Guide whether in your local newspaper or on your cable box we’re not really keeping pace with next gen you guys in next gen search so if you think about next gen you guys whether it be Spotify playlists and search right or Netflix it’s carousels and imagery and technology meaning recommendations being a key component so discover weekly for example on Spotify or you know or Netflix recommendations that metadata really need to change to power this next gen you guys and things like descriptors and photography and so we really set about number one. Transforming the metadata industry by investing in technology to really drive the connection you guys.

[ 00:09:19 ] We also saw though that the world was going international and we saw companies like Roku or Apple were going worldwide. And so we said you know what. What we do at scale in the United States we do it scale around the world we brought huge value. And if we can produce the highest quality at the lowest cost we win.

[ 00:09:37 ] Kind of like an auto manufacturer right if you’re GM and you can produce the highest quality car at the lowest lowest cost for a while you win and then the third component was really we saw these genres breaking down and we said hey you know Comcast you’re your competitor is not is not necessarily Verizon your competitors Apple and you have to be competing not just for the you know the video ball but you have to be competing for the music ball or the ear I guess. And so we said hey we do in video we still do in music and sports as well and so you know kind of we read along those lines and the company has grown from about 60 million three years ago to over 200 million today in terms of revenue and we’ve grown from probably foreign employees to 2000 boys today and used to operate only in the United States. Now we operate in 20 countries around the world and offer our product in 80 countries around the world. So we count as our customers virtually every large manufacturer every network every cable operator around the world. So you know whether it be Apple TV or whether it be Amazon Prime or whether Samsung TVs or Comcast X-1 or or DirecTV we power spotify with we provide data that powers all these interfaces.

Thought Gallery Channel:
Backstage Conversations
Backstage Conversation Season: 2017