My friends at KeenKong have been busy perfecting their conversational mining platform since my last visit to Montreal in December and have come up with something new, Conversational Metrics including a Net Promoter Score for their conversation meaning platform that includes a semantic analysis overlay.  I first wrote about KeenKong’s initial offering at KeenKong – Conversational Analytics last November.

KeenKong asked me to come Montreal to give my opinion about their platform much as Compete.com has (but they didn’t fly me to Boston for that) yet I’ve found my opinion being asked for, more and more – and my friends at KeenKong.com have taken some of my advice – just as my friends at Compete.com have, and you will see some of it (for Compete) on May 15th – as a new feature of Compete that I directly influenced.  The same goes for these metrics – though the team at KeenKong went even broader than what I asked for.

First, the Net Promoter Score for Twitter Conversations around a brand – it’s still being tested and in the process of certification but I can say KeenKong is the first to come up with such a metric for conversations on Twitter to my knowledge.  I won’t go into the Net Promoter Metric for KeenKong as I was asked to wait on that – except to announce the capability exists – and KeenKong is doing it – contact Frederic Guarino at KeenKong.com if you want to know more about the new KeenKong metrics module – again, this was an idea I floated to them last December – but I’m sure I’m not the only one who asked for it and I did  not specify the Net-Promoter score – another of their beta testers did.

For some reason Frederic Guarino compared me with @conversationage – Valeria Maltoni – who is a friend of mine, for this analysis – I’m sharing what Frederic prepared for me. I had no input in this decision but am sharing it as it was given to me.

The first metrics KeenKong supplies me with is total mentions and total reach of those mentions – as you can see, Valeria has a larger following and therefore, more mentions and reach than I do (looks like, for the selected time period, a tweet from her could be seen by 9 times more people than me).

Taking a direct suggestion from me, KeenKong not only publishes it’s metrics but shows the formula it uses to calculate that metric.  This is very important to me and something I stressed, I see way too many industry metrics that don’t publish the essential information on derivation – I saw that often with metrics derived from Comscore – and I never liked it.   If you can not explain how you derive an number you should not publish it.

Having said that, the Intimacy rate is a metric that you’d need KeenKong’s semantic analysis engine to come up with as you’d have to first define “intimate” conversations (what one shares about oneself) and in this sense KeenKong is showing Valeria and I are about the same in how we talk about ourselves.

Here are some more Conversational Metrics from KeenKong.com (I left the Net-Promoter Score out as it’s still a work in progress).

Well, according to this, I answer more of my tweets but also talk more about my personal life.

I guess the last metric above shows that Valeria needs to answer more of her tweets – on the other hand, if you have a lot more followers it might be harder to keep up with all the requests.  Jeff Pulver (@jeffpulver)  has over 350,000 followers, more than 100 times what I have – I wonder what his conversational metrics would be here?

Anyway, we are only looking at two days in this metrics analysis, April 7th/8th – I’m more interested in seeing the metrics over a period of 3 months – I’m sure that KeenKong has a lot of directions they could go with this new offering.

If you want to know about KeenKong as a platform and their semantic conversational metrics capabilities contact Frederic Guarino at KeenKong.com

KeenKong – is what I call a “conversation analytics” platform that is just about to launch (English and French) in Beta; I’ve known of KeenKong since early this year, when a friend, Frederic Guarino showed me what he was working on.   Occasionally, I’d hear references to KeenKong, and I saw a short demo of it this summer.

Yesterday, I was able to see a working version and get questions  answered  by Fredric Guarino and Oliver Berger.  KeenKong is based in Montreal, but Fredric is a frequent visitor to New York, so I have seen him often here, and I may go up to Montreal at some point in the near future.

KeenKong is designed to handle and add structure to a large number of personalized conversations with Brands and/or Individuals, where it is also desirable to reply personally.

KeenKong captures, parses and groups conversations into several buckets that are defined, partly by the conversation, and partly by the beliefs of it’s creators (I believe, based on Neurolinguistics) that conversations will always fall into 24 (or so) brackets or buckets, if you will – and this is a built in feature of human communications.

KeenKong is built to first take whatever conversation your having online, and structure it into those buckets – so you can better sort on the conversations and respond, in kind.  For example – if you have several thousand people trying to converse with you (say your a famous celebrity), or a with a well known Brand, through a blog or website (KeenKong will soon be available for Facebook, MySpace, Identi.ca and will have iPhone and Blackberry applications available in the near future), KeenKong will greatly simplify how you can respond and build trust by providing personalized answers, and back and forth communications.   As more and more people converse with each other – it becomes neccessary to invent platforms like this one – to facilitate much more optimal conversations.

You start by viewing your conversations on a “TalkBoard” (see above) that groups the conversations into the buckets or categories, on the left, then finds common objects in the conversations (ie: shirt, belt, shoes, etc) – Demographic information is provided, when it is available.

when you select one of the power words or objects, you get all the conversations they’re part of and you have the option of responding to those conversations in the same way by tagging individuals as a group you want to identify.

Once you have the conversations, you can look at the effect of your work using KeenKong over time for Sentiment and Effectiveness.

When I saw the demo for the first time this summer, @whitneyhess was testing it with twitter conversations she was taking part in and it was amazing what she was able to accomplish with KeenKong.

After looking at the Demo yesterday I was tempted to think KeenKong would work out really well for Customer Service, but Fredrick and Oliver were quick to point out that KeenKong was designed for a much wider variety of uses.   We also talked about what the definition of a conversation is, and of “active conversations” (back and forth between one individual and one respondent) and “interactions”, which are more like what most Social Media tools count today (all Tweets, Blog Posts, comments/posts on Facebook and MySpace, when they can be read, photos and videos) as conversations when they are often really one way “interactions”.

Since there is no “standards” that are in place for each vendor of a Social Media package to define what a conversation is, each vendor ends up deciding this for themselves – often making information between platforms totally incompatible – a much worse situation than what Web Analytics had/has with different platforms reading raw data logs in their own way – but at least, most Web Analytics platforms agree with the definitions of what a unique visitor, visit, pageview, bounce rate, etc, are – whereas most Social Media platforms, have no such agreement (see my presentation on the Future of Social Media Monitoring, above for a fuller rundown on this issue).

By the way, I’m excited to give the presentation to a large audience in London next week – and what I’m sensing is KeenKong is a platform that falls in the category of “applied social media monitoring” – it’s more of a next generation application that’s used to make life much easier for individuals that need to understand the conversations and respond to them, in kind.  KeenKong is not a monitoring platform that tries to find out what people are saying about a topic or Brand – it’s purpose is to enable and supercharge conversations between “You” and your customers.

What do I think of KeenKong?

I don’t know about the subdivisions/bucketing of conversations – but as long as it works, whatever way KeenKong wants to classify information is OK.  Furthermore, by classifying conversations at the GetGo, KeenKong is “segmenting” information much in the way some websites are able to segment visitors by offering a simple homepage, but pointing visitors to options (do you want to “learn”, “shop”, “buy”, “get support”, etc) – much as sites like IBM.com, have done – and customers self select/segment themselves by what links they click on and what pages they land on.

This kind of “self selecting” / “navigational” segmentation works very well for Web Analytics – but hasn’t really been applied to “conversations” till now.

What KeenKong has done is look at conversations as if it’s a website, and given us navigation – while adding unique tags to the conversations so you can respond to them in a highly personalized way – but in volume.

There’s a lot of promise in KeenKong – since it’s just launching now – I’m sure there will be an open beta soon, and I suggest signing up for it, ASAP – I know I will.