Content New Publishing Tech

Q&A: How Bibblio is taking content recommendation to a new AI-driven level

Good recommendations are at the heart of leading content businesses such as Spotify and Netflix. Users now expect that same experience everywhere. The reward for publishers that get this right is more attention, which in turn leads to higher revenues. Enter Bibblio, a London based startup founded by Mads Holmen and Rich Simmonds, which helps digital publishers monetise their content more effectively using AI-powered content modules. WNIP caught up with Mads Holmen to learn more about how Bibblio’s tech is taking content recommendation to the next level.

Can you give us some background about your company?

We wanted to do something about two related problems which were bothering us: how hard it is to find the best content on the internet, and how hard it is for publishers to make sustainable revenue.

Since 2015 Bibblio has developed a proprietary SaaS product which enables digital publishers to recirculate their best content to visitors as on-page, ad-free recommendations. So far the results have been happier audiences and increased engagement for a rapidly growing community of publishers. Later this year, our company is launching a syndication platform on top of the core recommendation engine, making content discoverable across the web.

What business problem is your company addressing?

As a digital publisher, you have a straightforward but deceptively hard challenge – extract maximum value from your real estate, generally your web pages, at the same time as growing loyalty in your audience.

These days 70-90% of traffic to publishers comes from search and social media, so the challenge is how to get more out of every single visitor that arrives on one of your articles or content pages. The premise is simple: if you can show someone something that they’re interested in, they’ll click on that after they finish what they’re on, and then you’ll have generated another page view and a longer session. The result is additional ad revenue or increased sales and subscriptions.

The problem is, until now content recommendation hasn’t been done all that well.

Many sites still choose which articles to recommend manually, or they rely on basic tag matching or content popularity. They’re not using data properly to find out what people actually want – they’re guessing. And, when they add new content that’s more likely to tempt a reader, the recommendations don’t update automatically. The data is static, not dynamic and individual to each user and page.

Other sites choose to work with a vendor like Outbrain or Taboola, but their recommendations are often full of inappropriate ads that can potentially damage a brand and send visitors off-site. The actual internal recommendations are usually poor as well, and as a black-box system publishers have notoriously little control over data and presentation.

Being reliant on these approaches is causing problems for publishers. Big platforms like Facebook, Youtube, Instagram or Netflix have increasingly raised the bar for user experience by leveraging AI to match the right content to users dynamically via on-site recommendations and notifications.

What is your core solution addressing this problem?

Bibblio has a simple SaaS platform that enables publishers to drop modules on their pages that recirculate their most relevant content to their audience as they browse. It’s designed to solve the twin challenges of finding the best content to recommend and presenting it in a way that actually increases user engagement and loyalty. It combines sophisticated machine learning algorithms that clients can influence with stylish design and flexible placement.

The challenge we solve is providing easy access to a deep technology that allow publishers to start optimizing how their audience meets and explores their content across every page. The immediate outcome is more engagement from the same audience and content pool. The longer term benefit is additional new traffic as Google start to recognize the improved platform metrics such as bounce rate, dwell time and pages per session. For subscription businesses, recommendations and personalisation has also proven spectacularly effective in reducing churn rates.

How does your solution work?

Bibblio uses algorithms to read, understand and index content from a publisher’s platform, which is then provided to visitors in modules on their site. We can grab the content directly from the page and use it to create a connected map of everything on the site.

The machine learning algorithms ‘understand’ the content and the audience and find the most relevant and engaging content for each page. The software is constantly updating and learning, so recommendations are always fresh and improving. You can also build multiple modules to promote different types of content, including your own branded content or premium content. Using more modules often lifts overall page engagement.

Last but not least Bibblio offers complete flexibility of design and placement for the modules that contain the recommendations. This means that Bibblio can look completely at home on your site – using your own styling, font etc.

What are other people doing in the space and why?

The market for ‘content recommendation’ solutions was largely hijacked by Taboola, Outbrain and other ad network products. As these companies grew their ad businesses actual “recommendations” increasingly became an afterthought. Increasingly, quality publishers are realising the importance of internal recommendations as a tool for growing a loyal audience that can be monetized, rather than simply sending their users away for a few pence or cents per click.

What sets Bibblio apart is the way we look at both the user and the content when we calculate the most effective recommendations. Other entrants frequently concentrate on ‘personalization’, however we’ve discovered that a) many publishers don’t have the rich data required to really make that work, b) even when they do, over-personalizing can create a narrow and unsatisfying experience in the longer run, and c) with increasing regulatory attention on privacy and user data (GDPR!) there’s less appetite for personalization as a primary method of recommendations. It’s indicative that, across our network, “related content” modules actually outperform personal modules for search-based traffic.

Pricing?

Bibblio offers a fully transparent SaaS monthly subscription pricing in tiers corresponding to the volume of recommendations served. We’ve also recently started offering publishers a pure performance based model based on a super low CPE (Cost per engagement). On day one we aim to give publishers 4-10x direct ROI from Bibblio.

How you view the future?

Bibblio’s immediate priority is continuing to support our current rapid growth and build on recent successes with bigger publishers and platforms – luckily our investment in a significant enterprise architecture is really paying off at the moment. We’re also busy improving the product – the latest update to our algorithm means that clients can boost recent content in their recommendations, and prioritise content that leads to conversions instead of just clicks.

The longer-term goal is a simple syndication marketplace. This will allow quality publishers to purchase high-quality traffic directly from their peers via relevant in-situ recommendations, or send traffic to partners, guaranteeing the provenance of visitors and cutting out layers of middlemen all taking a cut.

Anything else you think we should know about?

You can see me (Mads) talking at a recent panel during CogX together with Grace Boswood from the BBC and Dan Gilbert from News UK discussing AI for media and publishing. It’s worth a view!

 

 

Related posts