Corona-Ampel: Automated local Covid-19 data newsletter
Organisation: Quo Vadis Veritas / Addendum.org, Markus Hametner (private, after QVV folded)
Organisation size: Small
Publication date: 11 Sep 2020
Credit: Markus Hametner, Gerald Gartner
Like many news outlets, we saw a huge interest and a large amount of repeat users on our COVID data analyses, with lots of interest on their local areas. We decided to support the local-first use case by offering a daily, automated newsletter that sends local-first information directly to users’ inboxes, so they don’t have to visit the dashboard every day. The newsletter provides local data and also puts warnings about data inconsistencies front and center – Austrian Covid data was never very reliable, so there are automated warnings when test positivity rates are abnormally low or high or data
More than 5000 users subscribed to the newsletter, which was more than a quarter of our daily audience at the time. We had a lot of positive feedback and as far as we know, this is the only automated, personalized newsletter of its kind.
A large-enough number of users donate to keep the newsletter available for free. Those donations do cover infrastructure costs, with small amounts left to cover maintenance cost.
Web scraping and data validation using R, newsletter generation using R markdown, Mailgun as infrastructure. PHP for signups and user management.
Most of the infrastructure was already in place for our automated dashboards, but we invested in making it rock-solid, since we can’t just take back and correct an already-sent newsletter.
What was the hardest part of this project?
Sending the newsletter only if all data was valid and scraping worked. There are comprehensive scripts for data validation that prevent the newsletter from being sent if anything goes wrong.
Generating more-or-less interesting subjects depending on milestones being reached, especially outside of waves of infections.
Most newsletters do not include pictures and graphics, because mail clients generally do not load images from external sources – or display warnings before showing images. We managed to automatically create graphs and embed them in a way that works in all clients that have any image support. Some Gmail clients force a dark mode that do weird things with colors though.
What can others learn from this project?
Putting users first may mean creating new formats and distribution mechanisms. The story is not always the best way to serve readers.
www.corona-ampel.org/401.html – Newsletter Example