MDC Partners on Tuesday launched PRophet, which it calls the first AI-driven platform to help PR professionals predict media interest, sentiment and coverage before a story idea is pitched. The platform’s technology samples millions of stories across thousands of media outlets via natural language processing and machine learning.
The platform was developed by MDC Ventures and is the brainchild of KWT Global Aaron Kwittken, who is serving as founder and CEO of PRophet, with MDC chairman and CEO Mark Penn. They chat with PRWeek Dashboard’s Aleda Stam about the technology.
What does PRophet do in layman's terms?
Kwittken: I submitted my idea through the MDC Ventures Inside program, which is kind of like “Shark Tank.” I've always had this idea that if movie studios and book publishers can run their scripts through intelligence such as machine learning and natural language processing to predetermine the future commercial viability of a book or a movie based on past genres, why can't I do that for PR professionals? Wouldn't it create better efficiency in the marketplace if I was using technology to predetermine a reporter's interest in a story before I waste their time? That's the idea behind PRophet.
Using that database that creates a "persona" for reporters based on their past work, I can predetermine a reporter's interest in a potential story idea, what the sentiment could be based on what they've covered in the past and then based on how other media either syndicate or pick up their stories, what the viral or contagion effect might be. No one has ever applied AI to the pitch process before. At first, people asked if this was a better Cision or Muck Rack, and the answer is no. PRophet makes the information you get from Cision or Muck Rack more actionable.
Mark, what was it about PRophet that caught your eye during the Ventures Inside program?
Penn: I used to be CEO of Burson-Marsteller, so I have a deep knowledge of the PR marketplace. I saw PRophet as a fantastic aid to communications teams figuring out comms strategy in a way that was additive to the expert knowledge of people working in a world that is so much more complex. We solicited ideas from within the company, and Aaron's rose to the top because A, I thought it was inventive, and there's nothing exactly like it. B, it was within the frame of technology that could be developed, and C, it had a favorable marketplace because of the enhanced techniques it utilized.
What kind of measurements do you provide?
Kwittken: Let's say we want to pitch something about Chipotle bringing back carne asada. You write up a press release and enter keywords to maximize your results. Then the algorithm runs the pitch against our database of about 16,000, or your own media list if you want. You will get a dashboard that gives you percentages of how likely a reporter is to cover the press release. Green means they'd probably look at it positively, red is negative and yellow is neutral. We also have a contagion factor, which goes up to five, that tells you how often a reporter's stories get picked up by other media. A brand can look through individual reporters or group by publication.
Who are your clients?
Kwittken: Currently, we're only selling to brands. Eventually, we want to sell to agencies because I know agencies will find a lot of value in this. It will be a great tool to help their clients, but they will also use it as a business tool when pitching to clients. We're also licensing by brand, not by user.
How has the pandemic affected what you do or how PRophet works?
Penn: I think PRophet created new situations. Regardless of expertise, I would not have been able to understand how thousands of reporters and others were writing about the virus. Instantly having PRophet condense that information, analyze it and rate it gives me knowledge set in changing circumstances that no matter how expert I was, I couldn't have predicted. Now that we have six months of a database full of how people write about COVID-19, we can see who's more skeptical, who's more believing and talking about a vaccine. PRophet gives us a real roadmap, for example, for pharmaceutical clients for examples on how you really get your message out.
What do you see in the future for AI in comms tech?
Kwittken: Products like PRophet one day will notify you when there’s potential interest in your story or pitch. I envision a world where PR people upload a few variations of a pitch with the timing, weightings and key words like PRophet does only PRophet will continue to run those pitches and variants in the background and then notify or alert the PR person when the simulations produce favorable results. So the PR person can actually perform other tasks while the technology is testing the pitch against various genres of media. I believe that AI will one day actually edit pitches for us to optimize placement and maximize sentiment.
Penn: For a long time when I ran Burson-Marsteller, clients wanted more data and now we're in the position to supply more data. But I don't want to happen to communications what's happened to a lot of newspapers where it becomes not about larger strategy but momentary clicks. The holy grail was always was gaging the impact of what gets written on behalf of clients, and the important thing is not getting sidetracked by a false holy grail. Data plus an intelligent read against a strategy you planned in which you find the balance.