[SCVS Live Chat] The Science of Social Influence - Chat 2

[PaulGi Lithium Alumni (Retired)]

[ 11/10/10 11:01 AM] Paul G - Lithium: Hi everyone. welcome to our 2nd live chat with Dr Michael Wu, principal scientist at Lithium

[ 11/10/10 11:01 AM] Paul G - Lithium: we'll take a minute to let people head over here

[ 11/10/10 11:02 AM] Michael.Wu.Ph.D: First of all, I apologize for some technical difficulties with showing all the animation on my deck.

[ 11/10/10 11:02 AM] Michael.Wu.Ph.D: We'll make sure that the archived version is good

[ 11/10/10 11:02 AM] Paul G - Lithium: welcome Michael thanks for being with us.

[ 11/10/10 11:02 AM] Michael.Wu.Ph.D: thank you

[ 11/10/10 11:03 AM] Paul G - Lithium: I will kick this off with a couple of questions, but the floor is open to everyone, so if you have something to ask Michael, please just let us know

[ 11/10/10 11:03 AM] Paul G - Lithium: You've been pretty busy over the last year developing and refining your research and findings. What were some of the big projects for the year?

[ 11/10/10 11:04 AM] Michael.Wu.Ph.D: we've developed a lot of very sophisticated algorithm in community, now we that we have acquired scoutlabs, we want to apply these same algorithm to the social web data

[ 11/10/10 11:04 AM] Michael.Wu.Ph.D: It is a much bigger data set, so there will be a lot of engineering challenges [ 11/10/10 11:05 AM] Paul G - Lithium: But you're thinking of applying some of the same thinking about health and vibrancy etc that we measure in our community platform?

[ 11/10/10 11:06 AM] Michael.Wu.Ph.D: yes, currently I am working on something like the community health index (CHI) for quantifying brand health for a company
 [ 11/10/10 11:06 AM] Michael.Wu.Ph.D: a single number that quantify the brand health, that is drill down-ale and actionable

[ 11/10/10 11:07 AM] Michael.Wu.Ph.D: it will be have a diagnostic part and a predictive part

[ 11/10/10 11:07 AM] Paul G - Lithium: so this would be across all the areas that your brand is present in such as twitter, facebook, blogs, other forums right?

[ 11/10/10 11:07 AM] Michael.Wu.Ph.D: it will be across multiple channels: twitter, facebook, etc and all about your brand [ 11/10/10 11:08 AM] Michael.Wu.Ph.D: btw, the diagnostic part of that index tells you what your brand health is now [ 11/10/10 11:08 AM] Michael.Wu.Ph.D: and the predictive part is able to anticipate and be a leading indicator of your brand health

[ 11/10/10 11:08 AM] Michael.Wu.Ph.D: so you can know which direction it will move in the near future

[ 11/10/10 11:09 AM] Sandy.Puglisi: including the company's web site?

[ 11/10/10 11:09 AM] Michael.Wu.Ph.D: it is more center around user generated content

[ 11/10/10 11:09 AM] Michael.Wu.Ph.D: so not really the coroprate website

[ 11/10/10 11:09 AM] Michael.Wu.Ph.D: unless you have a brand sponsor community

[ 11/10/10 11:09 AM] patrick.denis: Lithium Social Media Monitoring is a standalone or an add-on feature exclusive to lithium platform?

[ 11/10/10 11:10 AM] Paul G - Lithium: @patrick LithiumSMM can be used entirely standalone

[ 11/10/10 11:10 AM] Paul G - Lithium: so there is no direct relianve on community for this

[ 11/10/10 11:11 AM] Paul G - Lithium: it is used for understanding where the conversations are occuring outside of your own community

[ 11/10/10 11:11 AM] Paul G - Lithium: so Michael you have written a lot about influence, this year but what about trust? Is that something that we can quantify?

[ 11/10/10 11:11 AM] Michael.Wu.Ph.D: you guys are very quiet compare to the morning group

[ 11/10/10 11:11 AM] Michael.Wu.Ph.D: yes,

[ 11/10/10 11:12 AM] Michael.Wu.Ph.D: i have work on how to quantify how trustworthy a company is to customers [ 11/10/10 11:12 AM] Michael.Wu.Ph.D: basically there are 3 factors involved:

[ 11/10/10 11:12 AM] Michael.Wu.Ph.D: 1. responsiveness, 2. transparency, 3 co-creation of value

[ 11/10/10 11:13 AM] Michael.Wu.Ph.D: and there are 2 types of transparency

[ 11/10/10 11:13 AM] Michael.Wu.Ph.D: 1. transparency between customer and company, and 2 between customers [ 11/10/10 11:13 AM] Scott.Rogers: how do you quantify co-creation of value?

[ 11/10/10 11:13 AM] Michael.Wu.Ph.D: and there are also 2 levels of value co-creation

[ 11/10/10 11:13 AM] Paul G - Lithium: remember folks, if you have a question about any aspect of measuring a social interaction or engagement, just jump in and ask

[ 11/10/10 11:13 AM] Michael.Wu.Ph.D: co-creation can have 2 level too

[ 11/10/10 11:14 AM] Michael.Wu.Ph.D: 1. when company passively listen to conversation and make their product better [ 11/10/10 11:14 AM] Michael.Wu.Ph.D: 2. they can actively crowdsource the customer for ideas and innovations

[ 11/10/10 11:14 AM] Michael.Wu.Ph.D: the amount of effort spend in each is essentially the value co-creation

[ 11/10/10 11:15 AM] Michael.Wu.Ph.D: make sense?

[ 11/10/10 11:15 AM] Michael.Wu.Ph.D: or should i clarify further

[ 11/10/10 11:15 AM] Paul G - Lithium: @Scott, does that help?

[ 11/10/10 11:16 AM] Michael.Wu.Ph.D: any more questions?

[ 11/10/10 11:16 AM] Scott.Rogers: yes, but the feedback loop of acting upon info from customers is outside the group. yes?

[ 11/10/10 11:17 AM] Paul G - Lithium: @Scott can you clarify that. Are you saying the team who measures and the team who responds are different in your org?

[ 11/10/10 11:17 AM] Michael.Wu.Ph.D: if that is the case, they should work together

[ 11/10/10 11:17 AM] Scott.Rogers: about your CHI...what are the components in the algorthim?

[ 11/10/10 11:18 AM] Michael.Wu.Ph.D: The component of CHI has 6 factors

[ 11/10/10 11:18 AM] Paul G - Lithium: So Michael, if you take Scott's CHI question then Angie's

[ 11/10/10 11:19 AM] Michael.Wu.Ph.D: 3 of them diagnostic: members, content, traffic

[ 11/10/10 11:19 AM] Scott.Rogers: Sorry Michael - yes, for instance the teams were different, and the areas impacted by co-creation of value were often different from those two

[ 11/10/10 11:19 AM] Michael.Wu.Ph.D: 3 of them predictive: responsiveness, interaction, and liveliness

[ 11/10/10 11:19 AM] Michael.Wu.Ph.D: yes, that is often the case with social media

[ 11/10/10 11:20 AM] Michael.Wu.Ph.D: hence the question of who owns social: Marketing/PR or support/customer service etc [ 11/10/10 11:20 AM] Michael.Wu.Ph.D: let me get back to Angie

[ 11/10/10 11:20 AM] Michael.Wu.Ph.D: yes, actually we found more different types of than Forrester

[ 11/10/10 11:20 AM] Michael.Wu.Ph.D: There are people who are product experts with high credibility

[ 11/10/10 11:20 AM] Michael.Wu.Ph.D: they can drive greater conversion

[ 11/10/10 11:21 AM] Michael.Wu.Ph.D: there are definitely connectors who connect different groups of population [ 11/10/10 11:22 AM] Michael.Wu.Ph.D: For example some people may not be very well connected, say they only connect to 3 people [ 11/10/10 11:22 AM] Michael.Wu.Ph.D: but the 3 people are 3 presidents of some countries

[ 11/10/10 11:22 AM] Michael.Wu.Ph.D: you will almost surely miss this type of network bridges if you try to look for them with traditional methods

[ 11/10/10 11:23 AM] Michael.Wu.Ph.D: such as number of connections they have

[ 11/10/10 11:23 AM] Paul G - Lithium: So actually I have a question for the audience

[ 11/10/10 11:23 AM] Michael.Wu.Ph.D: u are welcome...

[ 11/10/10 11:23 AM] Paul G - Lithium: How many of you are measuring your social interactions with metrics that are more than just fan counts or subscribers?

[ 11/10/10 11:24 AM] Michael.Wu.Ph.D: but we can find these type of influencers reliably too

[ 11/10/10 11:24 AM] Paul G - Lithium: Michael and I are sort of interested in where people are

[ 11/10/10 11:24 AM] Paul G - Lithium: show of hands - any one have a sophisticated measurement program running for social?

[ 11/10/10 11:25 AM] Scott.Rogers: yes

[ 11/10/10 11:25 AM] Michael.Wu.Ph.D: cool there are some YES

[ 11/10/10 11:25 AM] Michael.Wu.Ph.D: good start there @angie

[ 11/10/10 11:25 AM] Scott.Rogers: using one for quantity and another for unstructured text analysis

[ 11/10/10 11:26 AM] Michael.Wu.Ph.D: great

[ 11/10/10 11:26 AM] Michael.Wu.Ph.D: any deep and insightful measurement will have to involve numericla metrics + unstructure text

[ 11/10/10 11:26 AM] Scott.Rogers: ran closed community years ago, but hard to keep up with small staff

[ 11/10/10 11:26 AM] Michael.Wu.Ph.D: such as sentiment analysis, topic discovery, etc

[ 11/10/10 11:27 AM] Michael.Wu.Ph.D: so how many staff do you have?

[ 11/10/10 11:27 AM] Darius.Dunlap: In the old days we called it "reading the forums." :-)

[ 11/10/10 11:27 AM] Scott.Rogers: beyond word cloud topics

[ 11/10/10 11:27 AM] Scott.Rogers: actively moderating forums - lot more work

[ 11/10/10 11:28 AM] Michael.Wu.Ph.D: yes, nowadays, there are more interaction in the community than just forums [ 11/10/10 11:28 AM] Darius.Dunlap: Indeed.

[ 11/10/10 11:28 AM] Michael.Wu.Ph.D: yeah word clouds are nice, but not very actionable
 [ 11/10/10 11:28 AM] Darius.Dunlap: Indeed - to Dr. Wu's point, as well.

[ 11/10/10 11:28 AM] Scott.Rogers: 4 - 5 years ago....yes, can do so much more now

[ 11/10/10 11:29 AM] Scott.Rogers: word clouds are rarely actionable - agree

[ 11/10/10 11:29 AM] Michael.Wu.Ph.D: that is why I am working on a topic discovery algorithm that summarizes the conversation

[ 11/10/10 11:29 AM] Scott.Rogers: how so?

[ 11/10/10 11:29 AM] Darius.Dunlap: That's difficult. Many have tried

[ 11/10/10 11:30 AM] Michael.Wu.Ph.D: you can summarize the topics, so you can identify emerging problems

[ 11/10/10 11:30 AM] Michael.Wu.Ph.D: and identify them early

[ 11/10/10 11:30 AM] Darius.Dunlap: Just "finding the heat

[ 11/10/10 11:30 AM] Michael.Wu.Ph.D: yes, sort of

[ 11/10/10 11:31 AM] Michael.Wu.Ph.D: finding the heat is what the hot topics algorithm do

[ 11/10/10 11:31 AM] Scott.Rogers: is the quantity of nuetrality a function of the channel?

[ 11/10/10 11:31 AM] Paul G - Lithium: We have just about 10 mins left - so time for a few more questions

[ 11/10/10 11:31 AM] Darius.Dunlap: Can this extend to outer channels? facebook, twitter, blogs, ...?

[ 11/10/10 11:31 AM] Michael.Wu.Ph.D: if you get their conversation data, you can

[ 11/10/10 11:31 AM] Michael.Wu.Ph.D: the algorithm is general

[ 11/10/10 11:31 AM] Paul G - Lithium: @Scott, do you mean does the medium or the channel have an influence on how vocal people are?

[ 11/10/10 11:32 AM] Michael.Wu.Ph.D: @scott, i guess I am not sure what you meant by neutrality

[ 11/10/10 11:32 AM] Scott.Rogers: channel - meaning community vs facebook, twitter, etc

[ 11/10/10 11:32 AM] Paul G - Lithium: Because certainly if you're measuring a channel that your audience isn't really participating in, your measurements will be skewed

[ 11/10/10 11:33 AM] Michael.Wu.Ph.D: @angie, that is great, the more proper way to do it is to measure each channel then quantify each channel

[ 11/10/10 11:33 AM] Michael.Wu.Ph.D: @scott, yes, i understand the channel, but what do you mean by neutrality? [ 11/10/10 11:33 AM] Scott.Rogers: find lots of neutrality in social networking sites and more usable info in sponsored communities or communities on specific topic

[ 11/10/10 11:34 AM] Michael.Wu.Ph.D: @angie, yes sentiment analysis is very difficult now

[ 11/10/10 11:34 AM] Scott.Rogers: neutral in terms of sentiment, but not necessarily in content

[ 11/10/10 11:34 AM] Michael.Wu.Ph.D: so sometimes it cannot pick up all the linguistic discourses.

\[ 11/10/10 11:34 AM] Darius.Dunlap: More narrow focus means more meaningful content?

[ 11/10/10 11:35 AM] Darius.Dunlap: Depends on the topic, right? Technical topics much more so?

[ 11/10/10 11:35 AM] Michael.Wu.Ph.D: @Darius, yes more focus the more meaningful content, but you also get few people participating, so you need to balance them

[ 11/10/10 11:35 AM] Scott.Rogers: also differences between customers knowing you are listening versus not

[ 11/10/10 11:35 AM] Darius.Dunlap: Yes, that's what I was getting at. Thanks.

[ 11/10/10 11:35 AM] Paul G - Lithium: We have a couple of mins left, so probably time for 2 more

[ 11/10/10 11:35 AM] Darius.Dunlap: Schroedinger's cat. ;-)

[ 11/10/10 11:36 AM] Michael.Wu.Ph.D: @scott, most companies are listening regardless...

[ 11/10/10 11:36 AM] Drew.Jordan: one issue that I have trouble with is, the measuring the change in user sentiment...for example if you've got a major promoter that sings your praises in many mediums, but then he hates your new release and becomes a detractor. You have many positive data points from this person, but in reality he is now negative. how do you address that?

[ 11/10/10 11:36 AM] Scott.Rogers: LOL  

[ 11/10/10 11:36 AM] Shawn.McPike: Great discussion guys - have been enjoying silently.

[ 11/10/10 11:36 AM] Michael.Wu.Ph.D: yes,

[ 11/10/10 11:36 AM] Michael.Wu.Ph.D: but that can be tackle with smaller time window

[ 11/10/10 11:37 AM] Michael.Wu.Ph.D: so you just compute sentiment over shorter time periods

[ 11/10/10 11:37 AM] Darius.Dunlap: Thes conversations have always been going on, but now they are more often where companies can hear them, instaed of at the kids little league game...

[ 11/10/10 11:37 AM] Michael.Wu.Ph.D: so you can track the sentiment over time

[ 11/10/10 11:37 AM] Michael.Wu.Ph.D: @shawn, you are welcome

[ 11/10/10 11:37 AM] Darius.Dunlap: Multiple curves are useful, I assume? What's youre research say?

[ 11/10/10 11:38 AM] Michael.Wu.Ph.D: Do you mean multiple regression?

[ 11/10/10 11:38 AM] Michael.Wu.Ph.D: yes, they are very useful

[ 11/10/10 11:38 AM] Paul G - Lithium: So we have about 2mins left, do we have any final questions for Michael?

[ 11/10/10 11:39 AM] Darius.Dunlap: I've always done that with case loads and quality data, so it just makes sense to have more responsive and less responsive curves/projections.

[ 11/10/10 11:39 AM] Paul G - Lithium: Of course we can take topics over to the Group Chat area when we are done here too

[ 11/10/10 11:39 AM] Darius.Dunlap: General question: I lost connection mid-way, will the audio be avialable for replay? [ 11/10/10 11:39 AM] Michael.Wu.Ph.D: yes

[ 11/10/10 11:40 AM] Drew.Jordan: @dr Wu...thank you, would you weight a "power user" differently?

[ 11/10/10 11:40 AM] Michael.Wu.Ph.D: currently we don't but we are starting to collect data that would enable us to do that [ 11/10/10 11:41 AM] Michael.Wu.Ph.D: for example, we need to cross correlate twitter handle with FB names, with email account etc.

[ 11/10/10 11:41 AM] Michael.Wu.Ph.D: that is in the pipe

[ 11/10/10 11:41 AM] Paul G - Lithium: The other thing I would say Angie is that depending on your implementation with us, it is possible to connect community users with your crm users very easily, you have enough data points, but outside in the social web

[ 11/10/10 11:41 AM] Paul G - Lithium: there is often not enough data

[ 11/10/10 11:41 AM] Drew.Jordan: what I mean, is if someone has said 100 nice things about your product, and now dislikes it is there any sort of predictive data that he would likely make 100 negative comments, or would he likely go away? (I think that you may have just answered this)

[ 11/10/10 11:42 AM] Paul G - Lithium: Okay guys - we are at time here

[ 11/10/10 11:42 AM] Paul G - Lithium: Michael will be in the Group chat area for a while if you have more questions [ 11/10/10 11:42 AM] Drew.Jordan: thank you

[ 11/10/10 11:42 AM] Scott.Rogers: thank you

[ 11/10/10 11:42 AM] Darius.Dunlap: Thanks again!

[ 11/10/10 11:42 AM] Paul G - Lithium: and of course you can read Michaels blog or follow him on twitter @Micha8elWu [ 11/10/10 11:42 AM] Michael.Wu.Ph.D: we can continue in the group chat  

[ 11/10/10 11:42 AM] Paul G - Lithium: Thanks so much Michael and everyone for your time and questions

[ 11/10/10 11:43 AM] Scott.Rogers: advice ...do follow Michael

[ 11/10/10 11:43 AM] Michael.Wu.Ph.D: thx

[ 11/10/10 11:43 AM] Paul G - Lithium: thanks everyone!

[ 11/10/10 11:43 AM] Michael.Wu.Ph.D: @Drew, i will answer your Q in the group chat area