Our ethos, as a fund, is to be as open and transparent as possible with everyone in the industry, especially the startups we are lucky enough to meet and work with.  Not only do we feel this is a good way to do business but it’s just respect, plain and simple. With that in mind, we’ve decided to begin making public our reasons for backing our portfolio companies. Kicking off this series we go back to our first investment, Bibblio.  Bibblio has built the world’s most advanced knowledge discovery engine. In simple terms, it’s a system designed to help you discover the next piece of content you want.
 
“The future comes highly recommended” – Mads Holmen, CEO of Bibblio.
 
Current recommendation systems are built to produce “personalised” feeds of information that drive our consumption in a predetermined direction; from timelines on Facebook & Instagram to yield focused links for articles powered by the likes of Taboola or Outbrain. Why does this matter? Well, for example when we go out and buy a physical newspaper you and I are both presented with the same editorially curated content. Online, on the same day, the algos are surfacing different content to each of us. For the most part, when Netflix suggests a show to watch or Facebook curates the most exciting update amongst our friends to read, that’s fine; but given most of us consume all knowledge digitally, this gets more serious. In the newspaper world, the algos have become the “editors” (curators more accurately) of what we see and read.
 
Recommendation systems are nothing new
 
The move to digital has been a massive challenge for the media and knowledge industry, and many attempted to generate digital revenues by making content free and using advertising to generate revenues. As a result, the key metric became “yield per page” and page views, and gave rise to the big two players who help achieve this, Outbrain and Taboola. They have been excellent at this, but their approach often comes with the price of some “click bait” type of recommendations that we have all seen (e.g. “8 reasons the crowd cheered”).  This approach to content recommendation has begun to lead to some unforeseen and unwanted consequences, including linking to unsafe websites and posting offensive advertising, and high-end publishers are reconsidering the approach they take.
 
This business model hasn’t worked for all companies; Specifically, the organisations that have historically invested heavily in producing high quality content. Learning from the music industry, who never gave up on the fundamental value of the content, many publishers who fit this description have moved towards a subscription or hybrid subscription/native advertising business model. Examples of this are Financial Times, The Times, Wall Street Journal, The Information etc. For these companies, the dynamics have now changed and they are no longer about pure pageviews and yield per page. They obsess about customer retention. Metrics such as number of articles read per session by the user and ease of discovering and consuming their content are now extremely important. For these companies, the existing recommendation systems aren’t ideal - they don’t optimise for this.
 
This is the market Bibblio serves. As the media and publishing world begin to fork, with ad-yield focused free content on the one hand and engagement focused high-value content behind a pay-wall on the other, we think it’s a great time for a player to serve this latter market. This is what makes Bibblio exciting. Bibblio has just launched their platform commercially, and their first product is the content recommendation engine that suggests articles you may want to read next from that publisher. Soon, their customers will also be able to use Bibblio to recommend articles or content in their newsletters, and even offer personalised front pages.
 
“Cool, they do recommendation. What else?”
 
I’ve used the terms “content recommendation” and “personalisation” in this post. Which bring me onto another key reason why we backed Bibblio. Recommendations are not homogenous and there is not just one “way” to make a recommendation. At the highest level, you could begin your recommendations from a content piece; e.g. related content, similar content and most popular content read after this article. Alternatively, you can start with the user; e.g. content people like you read, related content from what you most recently read, content you typically read etc.
 
Take a practical example of why this matters; let’s say you are reading on climate change. From a content perspective, there is a “horizontal” journey which goes from the chemical compounds causing climate change to the social issues caused by changing climate. Within each point on this horizontal, there is the “vertical”, reading further to understand a specific issue caused by climate change in detail.
 
When it comes to personalisation what we read and our persona varies considerably with time of day, mood and our context. Personally, on a Monday morning, the content I want relates to my job to keep up to date on industry trends, key news and researching a topic I’m currently focussed on. On a Friday night, well, in truth the content I’m looking for relates to what film I should watch, or what new restaurant or bar I should try! Overlay that onto how different people use the same publication, take a Wall Street Journal or Financial Times, some of us use it principally for commentary and industry news, while others use it to be updated on specifics such as reporting base rate decisions or to check on earnings reports.  
 
Bibblio is not building a single homogenous “recommendation”, but rather a dynamic recommendation that can be customised by the publishers and perhaps even the user soon. In a world of fake news, filter bubbles, ad driven content, and politically polarised content; we’ve bought into their vision. As individuals we have limited time and attention and this helps us maximise the value we get from the content we consume.
 
First and foremost we’re a deep tech investor, and we’ve not talked about how great their technology actually is. That’s because technology is at its best, when we don’t have to talk about it. It just works, and the complexity of the tech stack required to deliver all of this shouldn’t be the user’s problem.  To this end, they have built a beautiful UI and really focused on driving seamless value to the customer. They have an incredible team who we’re proud and lucky to work with, and rather than gushing about how awesome they are, we’d rather let their product speak for itself. Check them out at www.bibblio.org; We think you’ll love them as much as their customers do.