Leiki Newsletter for Digital Media Professionals
Issue 2007/1
Solving Online Media's Biggest Challenge
Welcome to the Leiki Newsletter for Digital Media Professionals! In our first issue "Solving Online Media's Biggest Challenge", we look into the world of online media and content linking.
E-mail us to subscribe to the Leiki Newsletter
In this issue:
Online media is here to stay. In the last year, daily hours spent on online media has already overtaken time spent with print media in Europe - in the US the odds were stacked a staggering 5 to 1 in favour of online. We asked Dr. Petrus Pennanen who's been on the internet daily since 1986 how he sees the online media market today?
 Online media has come far in terms of impressive multimedia content such as videos and flash ads, and the content selection has of course increased all the time. A big trend has been the introduction of user-generated content, mostly as discussion forums and blogging platforms. The basic approach has remained pretty much the same - publishers and media companies have separate websites for each of their media brands with much of the same content as in the traditional channels (newspaper, tv show etc), organised into multiple sections. There are now a billion times more webpages than 15 years ago, but the normal methods of content discovery are still the same: keyword search engine, manual link lists and word of mouth.
Most people are quite happy with the current content discovery methods. Why would it be an issue?
I think it's a big issue. By transferring the traditional media properties online, typically a large number of separate websites are created - with much of the same content as in the traditional channels. This is then compounded by the introduction of new web-only services on new
 sites. Individual media websites and portals keep adding new sections and services, leading to long top-level menus with even more submenu items. Discussion forums and blogs also appear as either separate services accessed from their own menu, or comment chains tied to individual articles.
 A major challenge with all of this is decreasing usability and increasing fragmentation. In this situation the old content discovery methods are not sufficient; the jungle of site addresses and navigation paths is not very user-friendly, and is not increasing the page views as one would hope. Most users only go to a couple of websites of a particular media company regularly, and the one or two sections on those sites they are familiar with. They don't discover the wealth of content on offer that would be of interest to them if they knew it was there - and don't view the ads that come with it.
What do you think the media company should do about it?
To attract the users to multiple sections and websites, media companies can link the different sections and content types - editorial and user-generated content, classifieds and banner ads - with each other across different sites.
A typical large publisher we work with has a multitude of sites. For example, let's consider a person who's looking for information on a sporty German car - a typical publisher has plenty of such content:
1 - a couple of newspapers that (on occasion) print reviews and news about German premium cars;
2 - several magazines that write reviews and news about quality European vehicles
3 - discussion forums with high-volume exchange on performance sedans;
4 - classifieds site(s) that sell Bavarian high-end automobiles; and
5 - ads from a German manufacturer of environmentally friendly family-oriented performance cars.
Now a typical user only reaches one or two of these content sources. For example they read one magazine and have been to a discussion forum. However, if this person is considering (or dreaming of) buying the car he's probably interested in all of the 10 or so content sources relating to this type of car.
 The challenge for the media company is that most of the user's time is spent on competing sites. There's that other company whose newspaper the user reads daily; there's a third company whose magazine site has a lively discussion board; and there's that fourth company whose classifieds site the user visits regularly.
A media company can drastically improve the discoverability of content on all of its sites by adding a 'similar items' space in each article-level view, which links content from groups 1-4 above and also include a targeted ad space which shows ads from group 5 above. I call this 'unification of internet properties'.
People are likely to click on links that are directly relevant to their current interest, e.g. classifieds for BMW when reading a review of a BMW in a newspaper.
And it doesn't stop at just finding the same brand; the linking system can automatically find which cars are most similar, and recommend content about other German manufacturers with cars that have a similar chassis and engine.
 The beauty of all this is that it works automatically for tens of thousands of other categories on the same level of detail as 'BMW', connecting each of the detailed topics to first very specific and then more general context. This means the system works well across a wide breadth of content areas such as music, sports, cooking, travel and financial news.
That all sounds like a lot of work for the media company?
Fortunately it isn't much work at all since the content matching is automatic, and it's done based on very detailed automatic content analysis. Normally an editorial person of each media brand or website needs to spend an hour thinking of all the sites within the media company they'd like to link with. A technical person at the media company usually takes a week to add a "similar items" display to all the sites running on one platform.
On the Leiki side standard interfaces such as the RSS content format and a flexible API for receiving content recommendations have reduced the amount of work to 5-10 days even for use across all sites of a major publisher. (Leiki's 7 years of product development creating a state of the art, automatic classification and matching is not included here :)
OK, content linking can improve usability but I think media companies are mainly interested in very practical ROI.
When standard banner advertising is targeted, the increased relevancy and click through rates give a value add to the advertiser. This translates to higher fees per impression - that's the theory and the increased relevancy has been well proven. But what about in practice? Advertisers might not be too happy about an increase in CPM banner rates even with a promise that the impressions will be more relevant to users.
 One very practical way to increase revenues is by signing new kind of campaigns which are only enabled by targeting. This can be either several different ads for a generic product that can be coupled to different contexts, or advertising many products in one targeted campaign.
For example, an eCommerce site can give their whole catalogue to the media company. Each product will be linked with the most relevant context; LCD-TVs advertised next to an article about home theatres, Rolling Stones CDs next to news on Mick Jagger, running shoes next to latest marathon results.
Another example is a yellow pages site. Instead of "click here to go to the yellow pages", how about listing local renovation companies next to an article about redoing your kitchen, and likewise for the rest of the yellow pages database.
As the media sales department gains an understanding of these possibilities, setting up a targeted ad network is quick and easy and the costs are a small fraction of the income derived from the new kind of campaigns. That's why I find media sales to be the most eager customers for Leiki Focus.
More about Targeted Marketing in the next issue! | Subscribe to the Leiki Newsletter
In the start of 2007 United Magazines commissioned Leiki to enhance one of the Finland's largest portals, Plaza.fi with Leiki Focus dynamic content linking. With Leiki's personalisation engine editorial content on the Plaza.fi site was linked in real time with the most similar user-generated intra-portal content, external news sources, blogs and Wikipedia. In a similar manner, user-generated content was linked with the latest editorial content and other sources.
Leiki Focus content linking improved the portal's use immediately: instead of having to navigate menus to find relevant content, all information around one topic were presented in the same view, thus serving the whole portal to the user instead of just the section the reader was used to navigating in. External sources such as Wikipedia can also present reference material around the subject.
In October United Magazines and Leiki extended the use of Leiki Focus to ad targeting and released the first intelligent advertising system in the market where web ads are targeted using natural language understanding. With the new
 system editorial articles, user conversation and advertisements that deal with the same topic are matched automatically, even if they do not have the same keyword. The user's earlier selections that depict the user's interest can also be used to target content.
The first text advertisements targeted based on topical analysis of content appeared on Plaza.fi. Later on the system will be applied to other United Magazine
s' web services, such as the respective magazines' websites and NettiX classifieds sites including Nettiauto.com. Picture ads can also be used in the future.
The advantage of the system for the advertiser compared to other current systems is a higher response rate to the campaigns and the targeting method fits to all types of advertisers. Companies that particularly benefit from the targeting system include those with a large selection of products as well as those that can advertise their products naturally in relation to different types of content.
How does it work?
 Leiki Focus is a unique product with patented technology that enables easy content discovery on any digital channel. The technology is based on the usage of very detailed ontologies, hierarchies of concepts, which cover all fields of the portal content. The Leiki Focus engine, developed by Leiki Ltd, uses the ontologies to profile the content articles and match them intelligently with similar content items. The engine analyses the content and user click-stream and uses this data to create real-time content profiles and user interest profiles. These profiles are then used to recommend content such as news, blogs, videos, products, advertisements and also other users.
Click here to download the case study [pdf]
We're pleased to announce the release of Leiki Focus 5.7.0. This release fixes known bugs and provides significant performance & feature improvements.
Key improvements include:
- Significant performance improvements including caching of results in the Rest API responses and queued processing of user feedbacks using background threads
- Better scaling to very long articles and more sophisticated disambiguation mechanism in content profiling module
- Improved content and feed management features in the admin pages
- Improved documentation including step-by-step installation & integration guide
Latest Leiki News
Click here for the latest Leiki news.
Next issue (Ad Targeting - the Future of Online Advertising) >
For more information or to subscribe to the Leiki Newsletter please e-mail us.
Leiki Newsletter for Digital Media Professionals is a newsletter about Digital Media and Learning Personalisation published by Leiki Ltd. Leiki provides software technology for digital media to automatically personalise content and community services. Established in 2000, Leiki is the winner of the 2007 European ICT Prize.
|
