How Bergens Tidende Created Rich and Unique Automated Texts and Gave Readers Brand New Insights In the Real Estate Market
Overview of this campaign
When BT decided to test content automation, our goal was always to offer readers more than just a variety of available information in plain texts. We also wanted the stories to provide value not found elsewhere as well as offer the readers unique relevance and analysis. Our goal was automated journalism, not just automated content. To do so, we joined forces with United Robots, the Swedish automated content supplier specialised in robot journalism.
In July 2020 we launched the brand new real estate section, populated exclusively with texts about every single property sale in the region of Vestland.
The text has three layers. As you zoom out in the text, the more unique it becomes. Zoom level 1 provides information about the actual real estate. On this level, the robot collects official info from Norwegian Mapping authorities and does some basic calculations. It also adds spectacular aerial photos from a huge Norwegian database, and 3D panorama photos or/and street view from Google. The automated 3D-solution is a sophisticated new feature and gives readers a great overview of the area.
Zoom level 2 diggs further into the neighbourhood, and the robot uses data from level one to create a new layer of information. The robot looks at numbers of sales, average prices and general development. We’ve also introduced another brand new feature in the local marked: top list for very specific areas. This provides complete new insights for both readers and people doing real estate business. We now have hundreds of these lists. This also adds an enjoyable touch of gamification, which we’ve further explored by giving all houses or apartments in the top 50 percent a rank by the price.
Zoom level 3 includes all the official monthly numbers for the region, and a very detailed graph of the large cities in Norway with a massive ten years lookback. And finally, the robot lists the 10 biggest sales in the respective municipality. This also generates completely new information.
Results for this campaign
With the automated texts, Bergens Tidende tells the story of the whole property marked in both recent and historic times by simply highlighting each and every house in the same market. It’s hyper local journalism and regional journalism combined, which makes it something brand new. It shows how the market develops in the street you're living in, but also how your region is doing.
Since launching mid July 2020, we have sold 500 subscriptions from nearly 6000 automated articles. We anticipate roughly 1000 new subscriptions pr. year. That’s five percent of all the articles conversions in BT, overnight establishing the robot as the most popular service to our readers. This provides Bergens Tidende with a new revenue stream without at all inflicting the capacity of the newsroom.
Also, with 3000-5000 views daily, we’re passing the one million mark during the year. This is mostly down to another innovative feature, the headline generator. Titles are crucial to reader engagement. We’ve created a sophisticated “if this then that”-logic, which scans text for best headline. For instance, “This is the most expensive property in the area so far this year” or “This villa paves its way to the top ten list”. We almost have 100 different headlines in the bank for now, and still counting. As a result we can run a selection of stories on the front page, without any help from editors.
Although the real estate robot since the start alone has produced one fifth of the total numbers of articles in a newsroom with 100 employees, we don’t look to replace real reporters with automation. Quite the contrary, in fact. We’ve also strengthened our traditional coverage of the real estate market, since the automated texts provide journalists with great new insights and statistics. This also helps to shed light on blindzone areas in the market, and by that serves a journalistic purpose.