How starting with the right data drives US digital PR success
In the United States, there are reportedly over 1,200 daily newspapers in circulation, the vast majority being smaller, regional publications. This means producing digital PR campaigns for US outreach must be large-scale enough to impress journalists at national titles as well as flexible enough to be targeted to journalists only covering their local area.
The content brief from the client was the campaign had to be firmly based inside the sporting niche. Betting.com are a sports gambling affiliate site aiming to utilise digital PR to grow their American business. Therefore building links from major sports networks was important as well as gaining local press from sports desks at regional publications.
When it comes to PR campaigns having too much data is never harmful, each data point can be used to craft a story, a new outreach angle or at worst left on the cutting room floor. The best datasets very rarely look the most PR friendly from the outset, the best stories are always those that involve piecing together a narrative from the source material and letting the data guide the story. Using this approach we can use our experience within PR to extract the most newsworthy angles.
We found a dataset originally used for university research into racial bias, research that is very different to a PR campaign. Digging deeper the researchers had compiled over 50 years’ worth of announcer transcripts, over 1,500 games worth. The raw data contained the language used about the player by the announcer, as well as cursory details about the player in question. Some checks were needed as with all data, words such as ‘blocked’ in the commentary that might be flagged as negative by sentiment analysis but in fact are in common use in the NFL.
The data spanned over 250,000 snippets of commentary. Each snippet contains a host of information about the game and the player involved, turning that raw data into something manageable requires working outside the normal data-crunching tools of Excel or Google Sheets.
When utilising large datasets the key is to use a programming approach, in this instance, Python was used. This removes any limitations on dataset size and allows for deeper and more rigorous analysis when compared to common data-crunching tools like Google Sheets or Excel. Python was used to perform a faster, more rigorous analysis of the dataset. The result was a much more ”human-friendly” layout of the data and something that could be used for PR purposes. Simple cross-referencing can tell us who those players were playing for at the time and now we have a list of teams and announcer opinions. So how do we take 250,000 lines of commentary across every NFL team since the 1960s and make a PR campaign out of it? The answer is once again found in Python. More specifically, in natural language processing. Using machine learning techniques Python is able to classify groups of text by their sentiment, giving a score between 1 (positive sentiment) and -1 (negative sentiment).
Now the regional angle is clear, we have a list of teams and the sentiment of the language announcers are using about them. With 32 current NFL teams spread over 22 different states, there is a large scope for personalising the data to each team’s region. “New York Jets are officially the most hated NFL franchise” is likely to have a much higher success rate with journalists working in New York due to its more personal appeal. In this instance it also plays into the subjective opinions that sports fans always believe that the world is against them, this was seen in the coverage of the piece, with local Jets fansite JetsXfactor titling their piece “NY Jets fans, it’s true: There is announcer bias against your team”. The campaign was adding objective truth to subjective reality.
The campaign gained both local and national coverage and earned over 50 links from USAToday, Deadspin as well as local fan sites for the New York Jets and New Orleans Saints. Results like these are only possible through having a campaign concept that stands up to both national and local outreach, the key to scoring links with both is a great, personalised narrative. The key to the personalised narrative? Data. Having great data will make creating engaging and linkable PR campaigns a whole lot easier.