Over the past two years, the Norwegian Schibsted brand Bergens Tidende has received attention for the collaboration with Swedish tech company United Robots on automated journalism. Our Real Estate Robot was has been so successful that it also has been scaled to other Schibsted media houses in Norway.

This text robot produces an article for each house or apartment sold in our area. The article consists of key information about the actual house, like price and price history, and various kinds of automated pictures of the properties. The robot picks suitable headlines from a bank of nearly 100 hand-crafted alternatives.

The enrichment machine

But there’s more to it than that: The robot enriches the article with all ingredients expected of an online text. It analyses the local, regional, and national market. It equips the articles with detailed visualisations of price development over the past 10 years. It creates various lists of the most expensive housing down to very small geographical areas. By that it sheds new light on micro-trends in the market and also potentially reveals blind spots.

Best of all, it is 100% automated. A year after launch, the robot has produced 10,000 articles, generated 1 million clicks, and sold almost 1,000 subscriptions. That is a considerable result for a regional media house with nearly 50,000 digital subscribers in total.

Best of both worlds

I believe there are two main success criteria. Firstly, by combining hyperlocal information with macro-level trends, subscribers get new and highly relevant insights. Keep in mind that for many readers, nothing has a bigger impact on their personal finance than investments made in the real estate market.

Secondly, by imitating all of the steps in the editorial enrichment process, the articles become a lot more appealing. This again, as we know, leads to engagement. It’s super easy these days to produce automated content. The key is also to keep it relevant — and readable.

Our next shot

Now we recently released another service based on the same principles: The Annual Report Robot. In short, this robot writes a story about each company in our area, based on its accounting. It’s as hyperlocal as you can get, as every subscriber can read about every corner store. But it’s also a macro service that looks into trends and patterns on a business and industry level. As with the Real Estate Robot, we believe this to be crucial to success. The potential for scaling is massive, I think.

As a result of this new feature, Bergens Tidende is now able to cover local businesses much more comprehensively than before. The robot produces hundreds of stories each week, and the actual journalists get more time to do investigative work. A win-win situation.     

It’s too early to say whether this will be as big of a hit as the real estate robot, but we nevertheless feel that we’ve created a feature that brings value to our readers.

Division of labour

Let me end by emphasising the obvious: Our text robots do not replace real journalism. They do not see all the connections and connect all the dots, although they can probably be trained to dive far deeper below the surface than they do today.

They do not ask the tough questions or the follow-up questions. They do not give articles fragrance and colour — at least not to the extent human journalists can.

What robots do is produce valuable content at a volume real journalists cannot manage. And in that way, it also frees up journalists to do more of all the stuff they do best.

It’s actually a pretty brilliant division of labour.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .