The newsrooms want more visual, compelling articles that make good use of data. Time pressure in newsrooms, lack of appropriate tools and information caught inside article silos are all roadblocks they face when tackling this problem. That’s where we come in: Frames brings simple charts to 50+% of articles published and while we were at it, we also created a way to generate additional revenue. We are a small company based in Lisbon and we create a solution, based on a Machine Learning technology, that allow publishers to have charts, created by our team, automatically at the moment they publish the articles.
And how do we do it?
Time pressure is solved by outsourcing the data visualisation team. We continuously observe the news stream and proactively create charts that give context to current world events. And when your journalists write an article, we automatically suggest and insert related charts.
We fixed the lack of appropriate tools with an easy to use chart builder that integrates right into the CMS of the publishers. And every chart created is automatically fully on brand and available for reuse in other articles. That’s not unique, but what is, is that most journalists will never need to use this tool in the first place.
Article silos are no more because our solution, focused on the wider context of articles, are reused across publishers and even within publishers across dozens of articles about the same topics. The effort of chart creation is externalised and has finally an amazing economy of scale, allowing for massive coverage.
A new revenue stream in the form of text-based sponsorship ads under relevant charts have proven to be a desirable format for brands keen on using user-friendly and ads that integrate right into the content. In our case studies, we quickly amassed very relevant revenue for the publisher with big brands. We charge publishers a fee for this service which they can recover many-fold by exploring this new revenue stream.
What makes this project innovative?
Frames is a unique combination of features. What differentiates us most is: 1) Automatic matching is a game changer that allows for massive article coverage; We use a Machine Learning solution to facilitate the matching of articles and our charts. We can point Observador, one of the newsrooms using Frames, as an example: out of 16,412 articles published from April to August 2017, 0,64% of hard news articles use charts produced by the journalists using their visual team or external tools, such as Infogr.am. In contrast, 40,15% of hard news articles published on observador.pt in March 2018 show a chart produced using our solution. The coverage rate ramp up as we perfect our matching system and improve the topics covered, looking at data, user feedback and the feedback and suggestions received. 2) Outsourced data gathering and chart designing team avoids training and enables economies of scale; The economies of scale require… scale. All too often publishers shy away from sharing content or tech with competitors. Which makes sense when it comes to what makes them unique amongst each other. But with Frames the equation is quite different. Producing Data Journalism at the scale is only financially viable if multiple publishers share both the cost and benefit. Here in Portugal, where Frames is based, we are very glad to see the fear of sharing disappear when it is weighed against the immense benefit of having Frames in 30, 40 or even 50% of the articles published of each publisher. The typical Publisher only has charts on about 1% of their articles. There are now four newspapers using Frames in Portugal: Dinheiro Vivo, Jornal Económico, ECO and Observador. None of them would have had a financial case to produce and show this much data on their own. 3) A new revenue stream makes this into a no-brainer for publishers; Advertisers have proven to be keen on new contextual premium ad format in a sponsorship model. From the onset, we designed Frames to allow for a new, non-intrusive, advertising format, giving publishers a new revenue stream that recoups their investment in Frames manifold. We understand advertisers particularly like the association with data and visuals, appearing as experts in their area of business. And we’ve proven it in a grand way: Deloitte Portugal launched two campaigns on one of our clients sponsoring all charts related to the Government Budget and Christmas Consumption, strongly pushing for the message of Deloitte as the company with experts in the respective fields.Vodafone Portugal joined the ranks of innovative companies keen to have their logo and message placed next to our charts related to Smartphones and Apps - topics that are at the core of their business areas. SPORT TV also jumped on board sponsoring charts related not only to football, but all kinds of sports. Frames explores this revenue stream directly with the commercial team of the publishers. We can assist with pre-designed material to help convince advertisers of this new sponsorship format and our backoffice allows the publisher or the advertiser to build, configure and track campaigns. Our team customizes the style (color, type, framing and spacing) so that all charts that appear on the publisher’s website to match their brand. Readers can see no mention to Frames, so it looks seamlessly integrated into your front-end. This is a new, powerful in-line format that is particularly well suited to created awareness for the brand, placing it as an expert and go-to place on the topics it chooses. What we charge a publisher to use our Frames product is not only easily recouped, but massively surpassed by getting advertisers on board this new revenue stream.
What was the impact of your project? How did you measure it?
Our product is not a prototype. The product is fully implemented, it is online and generates very good KPIs on our launch customers. A/B testing consistently proves a higher number of time on site, shares, and returning sessions for users exposed to charts produced using our solution. Also proven is a significant revenue generation, allowing our customer to retrieve their investment manifold. We ran A/B testing on observador.pt for three months, during which half of the audience saw charts and the other half didn’t. For those who did, overall total monthly Pageviews were up by 6%, a massive increase with minimum effort that proves readers love better, more informative and visual articles. This was tested when our coverage was at only 36%, which proves the success of our solution. However, our main parameter of success is coverage. As we mentioned before, our main goal is to place contextual charts in the articles about current world events using our Machine Learning solution. And the results are impressive across publishers. On March 2019, our charts were present in 57% of all hard news articles published by Observador. That means almost 6 out of 10 articles have a chart produced using our solution. We can see the results in other clients as well, as the coverage rate on Jornal Económico, Eco and Dinheiro Vivo on March 2019 were 55%, 50% and 40%, respectively.
Source and methodology
Our methodology has two main dimensions. The first one is curation, which is the process of keeping a close eye on current events. We have a data journalist committed full time to observing current events and then researching and curating relevant data about them. After gathering the data, we design the charts using our technology. At this moment, we have over 700 continuously updated charts. We produce, on average, 2 new charts per day, on top of the always up-to-date database of Frames. Our sources are very diversified: from Eurostat, World Bank, United Nations, National Institutes of Statistics, governmental agencies, FIFA, FIA, Apple, Google, OCDE, and many others. The topics we cover are Politics, Business, Economy, Sports, Society, Health, Technology and so on. Furthermore, the journalists can duplicate, edit and create as many charts as they want from within the publisher’s CMS, in the front-end, without having to go out of the page. The second dimension is the automation. When we create a chart, we associate it with key-words that define its main subject matter. We “teach” our system to “read” the articles and get from the headline, standfirst, tags and first paragraphs the main concept. Our algorithm analyses every article on the publisher and then decide if there’s a relevant chart from our database to insert, based on the key-words, and adds it to the text.
Leo Xavier: Founder; Nuno Rodrigues: Senior Software Engineer; David Morais: Frontend Developer, D3 specialist.