One press release for each municipality in Scandinavia on how the bankruptcy state was in that very spot – in a publish-friendly format – automatically. Well distributed to adequate local media, it led to 120 different articles. Sounds good? With AI and NLG techniques, it was possible. The project was managed by content automation company Lingmill, in close collaboration with Consilio and Bisnode.
✔ 400 unique press releases
✔ 800 e-mail media recipients
✔ 3 languages
✔ 120 articles in local media
✅ Fully automated
Bisnode wants to preserve their position as thought leaders in business development in Scandinavia. Customizing Bisnode’s important, adequate data on a hyperlocal level and sending it in a journalistic-angled text to exactly the right local recipient constitute a massive impact. Of course, it’s a challenge to do so, but with the help of AI, it is possible.
With Lingmill’s techniques, bankruptcy data was selected and analyzed for every geographic location. Lingmill sorted out what’s most interesting for each municipality and wrote unique press releases for each one. “Huge increase of bankruptcies in Helsingør,” for example.
– This was perfect for this kind of statistics. To really take advantage and achieve a hyperlocal outreach of our data like this is of course extremely positive, says Tomas Hedenius, Head of Marketing and Communications, Bisnode Scandinavia.
400 unique press releases were then allocated to the right local media recipient and distributed to them. With the right text, and with the right subject line. For Denmark, about Denmark in Danish; for Sweden, about Sweden in Swedish; and for Norway, about Norway in Norwegian.
The result was 120 different articles in local media.
The framework and thresholds are prepared in advance. If data is missing or odd in any way, the algorithm gives a signal and doesn’t write that particular text. Tone of voice and overall style are also set in advance – strategically put together by Consilio together with Lingmill. The whole process of getting this project done, from ordering to the press releases sent to all recipients, took only 7 days.
In this case, the texts were about the number of bankruptcies in 2019 compared with 2020. The settings can be arranged to compare monthly or quarterly and so on.