RepKnight is a relatively new player in the reputation monitoring, crowd-sourced news and public opinion space. It has just been 12 months since John Reid, a 36-year old software manager, started the company from a garage in Drumbo but already and without any external investment they are becoming the "go-to" company for major private and public enterprises who need high volume near real-time social media tracking. Understandably John is a very in-demand guy but he popped into Belfast city centre to share a coffee with me and give me an update.
But first ... what does RepKnight actually do?
RepKnight allows you select keywords to be monitored across all the social media channels and it then captures all new conversations involving these topics within ten seconds of them occurring anywhere in the world. This information is summarised and presented graphically on a timeline on a dashboard. The dashboard does not just show the volumes of conversation but also colour codes them according to their "sentiment" - negative, positive or neutral. So you see at a glance how many people are talking about a topic at any point in the day, what the trend is and whether the balance of comments are negative or positive.
That's only the start of it however. ....
RepKnight also allows you to drill-down into these conversations. You can identify and select specific themes (e.g. "MTV", "PSNI", "Invest NI" within the topic "Belfast"). You can drill-down into specific channels (e.g. just show me Facebook conversations). You can play with the time period, for example, zooming in on just the last 3 hours. You can find out who is behind the conversations and start a campaign of response from within the RepKnight dashboard to get your "say" out there instantly when it matters with the most influential opinion-formers.
John puts it best himself:
"If you need to know what people are saying socially within ten seconds of it leaving their phone or computer then you should talk to us".
What is the "sweet spot" for RepKnight?
"RepKnight is aimed at any enterprise who need to know instantly when any one three things is lighting up the social media wires. 1: People are talking about them or their products. 2: Some news is breaking on the social web which has not yet reached the formal media channels. 3: Public opinion is being shaped in advance of a major decision (e.g. an election)."
This opens up a number of major customer sectors for RepKnight - primarily Security, Media, Political Affairs, Public Relations and Financial Sectors.
But what's actually different about RepKnight?
The space RepKnight is in is a very crowded space with a number of high-profile well-funded competitors such as Radian6 and DataMinr.
I asked John how he positions RepKnight against them:
"There are 3 things that standout for us. Firstly we are aiming at organisations who need to track very high-volume topics in near real-time - that distances us from a number of competitors. We have a direct feed from the Twitter Fire Hose - ten seconds after its tweeted it is in our database".
"Secondly rather than just give you the data and wish you good luck we make sense of the conversations for you using our proprietary sentiment analysis engine. We developed this in collaboration with the University of Ulster. It uses artificial intelligence and natural languages processing algorithms to assess whether the sentiment behind comment is positive, negative or neutral".
"Thirdly we don't just allow you to view the conversations we make it easy for you to spot who are the key contributors and to engage with them directly and instantly. If you don't have this you can only watch on as the market destroys your reputation or you miss a major opportunity."
Automated Sentiment Analysis
Some of RepKnight's competitors would argue strongly that automated sentiment analysis is just a waste of time. For example, how can a computer detect sarcasm - I put this to John:
"They are right - I totally agree with them - if you and I can't agree on the sentiment of a comment then how could we expect a computer? However this only matters when you are dealing with small volumes - if you are dealing with very large volumes, as we are, small errors average out and it is not an issue. We have proven in tests that once the volumes exceed a few hundred conversations the RepKnight automated sentiment score and the manual (human) sentiment assessment match up pretty closely."
I guess if you think about it that makes perfect sense - isn't that the whole thesis of the "Wisdom of Crowds" argument in Malcolm Gladwell's famous book of the same name? When you deal with a crowd then the small variations just don't matter!
Success Stories
RepKnight have already won major contracts with some very big players in their target sectors so I asked John to give me some specific examples of what they have been able to achieve.
"This is great stuff for dinner parties. On a serious level we were able to predict the outcome of the Irish elections and on a lighter level we correctly predicted the winners of Big Brother, X-Factor and BBC Sports personality of the Year".
"It's also very clear to me, but I don't want you to print it, who is going to win the Republican Primaries (Keyword: GOP) - you just have to look at the net sentiment of the conversations and the volume of each conversation for Romney and Gingrich.
The ability to predict the outcome of public decisions makes RepKnight very attractive not just to Political Affairs and Public Relations organisations but also major financial players who are wondering how they can manage funds around market sentiment."
The pitfalls of naïve social media analysis
John is a calm and considered kind of guy but one of things that gets him animated is his theory that things are often not as they seem when you analyse social media, that the obvious answer is often the wrong one and that it's very easy to get things out of proportion or think a topic is hot when in fact it's not.
"A good example of this was the UK Public Sector Strikes. When we analysed the conversations we found that the main public concern was not about whether they supported the strikes or not but rather how people would handle the childcare arrangements during the disruption".
"Another example close to home is that a few months back a local politician (who we won't name) generated some bad publicity around the time of the MTV awards which commentators suggested had done serious damage to our identity abroad.
However when we actually analysed the conversations on RepKnight we found that the negative impact was totally insignificant compared with the sheer volume of positive coverage generated by MTV."
Future Plans
The next twelve months look very exciting for RepKnight. They have been nominated as a Finalist in the Irish Times Silicon Valley Awards and John will be presenting at Stanford University in March. The Security sector has suddenly become very hot for RepKnight both nationally with the London Olympics (and London Riots). Also internationally things are getting interesting as no government wants to be blind-sided again by important emerging social movements such as Arab Spring and Occupy.
Conclusions
RepKnight is a smart company with a very well differentiated market proposition. They are in a hot space but have proven they have a bankable market offer by their ability to win major customers so early in their evolution. The competition is fierce but RepKnight have stolen a march through their positioning in the high-volume near-real time automated sentiment analysis space. I think they should change the name as Reputation Monitoring is only a small part of what they do and probably confuses the market. If John and his team of six can keep doing what they have been doing then there will be no shortage of potential suitors among the global media players. RepKnight are definitely a local company with a rapidly growing global Reputation.
For additional material on RepKnight including live screenshots and web links visit my blog at www.bioteams.com