Automated journalism customer cases | United Robots

Automated real estate texts drive subscription sales in Norway

Written by Cecilia Campbell | Nov 10, 2020 3:29:00 PM

Problem solved: Automation has enabled creation of a Homes Sales section on the Bergens Tidende site, driving subs sales through geotagging.

“There are 15–50 sales of houses or apartments in Bergen every day, and for buyers, sellers, neighbours, or people moving into the neighbourhood these texts are highly relevant. So we assumed they would generate subscription sales – and we’ve been proven right. The high quality of this real estate content, combined with the fact that it’s automatically generated, provides a net value for us – and our readers. But if we’d had to employ 5 journalists to do this job, the gain would have been cancelled out by the cost in newsroom hours.”

Jan Stian Vold, Project Lead, Bergens Tidende

 

Update on key figures April 2021:
– The automated real estate texts are now responsible for 5% of all article conversions on the entire BT site.
– ≈ 1,000 subscriptions sold / year @ €24/month from this content
– ≈ 12,000 automated articles published / year, generating 3,000–4,000 pageviews / day.

By early 2020, a small, monthly newsroom innovation project at Bergens Tidende (part of the Schibsted group) had turned into the development of an entire new Boligsalg (Homes Sales) section on the BT site. The only journalist involved is Project Lead Jan Stian Vold – all the content production is done by a robot. Without automation the Boligsalg section would not have been viable – the newsroom resources required would outstrip the value of the content. As it is, the value produced is high enough to drive subscription sales.

The robot writes 3–4 paragraphs per house sale, articles which include not just information about the actual “news” – the particulars of the sale – but also details about the property, including size. The robot then calculates price per square meter, and compares the sale to others in the neighbourhood, city section and Bergens as a whole. Each article includes a Google Streetview photo of the property, a satellite map and an automatically captured Google drone footage video frame with the address marked out.

In order to put the most relevant articles in front of site visitors, BT have developed a geotagging system that is more granular than the one used by the estate agent industry. In addition to using the tags to segment the content at the point of distribution, the geotagging is used to generate lists. “In each article, we list the five most expensive sales in the past 12 months at that most granular neighbourhood level, with links to the tag, down to street level, which of course increases the value of our set-up,” says Vold.

The top KPI for the Boligbot is the number of subscription sales. Since the launch in summer of 2020, this number has stabilised at 60–70 conversions a month, a number the team is happy with. “These are not extremely high numbers, but they play an important part for our overall subscription business, and we’re pleased with the results.”

On working with United Robots:

“A key success factor for the project has been how responsive the United Robots team have been to ideas I’ve had – the route from idea to impact and increased value is short. It’s been fantastic to work with developers who are just as keen to drive innovation as I am. With projects like these, there’s always a risk that you launch and leave, but the United Robots team continue to walk that extra mile to keep improving the content."

Jan Stian Vold

 

More resources:

Full BT case: Download Real Estate Playbook

Webinar with Jan Stian Vold: How one newsroom robot drives significant revenue

Download Real Estate Guide with best practices

Hyperlocal real estate articles drive reader engagement at Mittmedia