For my talk at the Columbia Business School Alumni Club on the evening of June 16th I’m preparing a case study of five local Times Square Restaurants.  As my readers may have guessed, I refine my ideas as I write and convey information – it’s as if the act of sharing the information with others, gives me more information.  I’ll stay away from Sentiment Analysis for the time being in this case study.

Clearly, I’ll be developing this idea more – and I welcome your input.

I also find considering other points of view are often helpful in getting a better result in Social Media Monitoring -  if  I can’t talk about what I am measuring and how, it’s difficult to get that information – and I bet that’s pretty symptomatic of many in this new, emerging field of  Social Monitoring – they don’t want to talk about and show the actual data they collect – which actually keeps the industry from evolving as quickly as it could or should.

My hypothesis starting out .. was that  I’m going to find  more Social Media Activity at Havana Central’s Times Square restaurant location on W.46th (because we have a community strategist, Cecilia Pineda Feret, and we’re actively doing Social Media Outreach and an owner that is supportive of Social Media) than the others (consisting of Carmines, Bond 45, Heartland Brewery/ChopShop and Blue Fin).  As I don’t know the management of the other restaurants in this study, I do not know for sure what they are doing or not doing with Social Media – except what I found in relation to looking at specific local restaurants in Times Square, Manhattan – not the chain or franchise.  My other suspicion, is that overall, I would not find that much Social Media activity in any of the 5 restaurants since Social Media is still emerging in this sector (dining).

Using Radian6, I crafted very targeted queries for each restaurant that included the FourSquare URL, the restaurant name + “Times Square”, and various references to meeting, dining, drinking and music.  I excluded references to Mayor Bloomberg dining at Blue Fin after the bomb attempt recently, and any references to police activity which sometimes crops up.

Here’s the first chart – over the last 30 days Havana Central Times Square had close to double the average number of social media references as the other 4 restaurants, and close to a third of the total references, period.

One of the problems I ran into was writing queries that reflected just social media activity or outreach – a form of marketing -  (a meetup, tweetup, FourSquare checkin, perhaps a tweet that some one was at the restaurant) vs. any information about a particular restaurant location that was online.   I reasoned that if I just focused on social media activities, I’d probably find next to nothing on any of the locations, even Havana Central would not show all that much – so I decided to open up the query to music, menus and dining in general – reasoning that act of showing up in Social Media is … Social Media (even though it might not have been caused by outreach).

According to Radian6, Havana Central had visitors who on Twitter that had the most followers (opportunity to see) – but the numbers of followers were much closer between Havana Central and Carmines and that might who happened to drop by and tweet there.

I think it’s worth comparing these results to the Roger Smith Hotel – because they went all out on Social Media and have made a name for themselves as the Social Media Hotel – something no restaurant has done yet.

One thing a restaurant has that Roger Smith does not – is the propensity to buy – just about anyone that who says they are at a restaurant is probably going to be a customer and will be ordering.

That might not necessarily be true at a hotel (on the other hand, when people do spend money at a hotel, as those who are staying there, they usually spend much more – case in point – just the cost of the room, not to mention eating and purchasing other items and  services there).

In either case, Restaurant or Hotel – the more social media activity detected (via Monitoring, in this case using Radian6) the more money is likely to be spent at that location.    Let’s step back a second and think about why that is.   Today, it is still next to impossible to tie Social Media activity (a Tweet, Checking, Post, Blog Post, Photo, etc) to actual purchases, but it is clear that such activities are proxies for actual purchases of services, or meals, or rooms – even if we can’t always prove it yet (but will in the near future).

So with that in mind, you would think the more people check in using FourSquare to a restaurant, or to hotel, for that matter – the better – as in both cases, people are going to more likely than not be purchasing something – and as more people use Social Media – more people will be “checking in”.

We know Roger Smith Hotel (and Roger Smith life) has been active almost two years now – I want to compare these five restaurants using Radian6, and I can’t go back more than 30 days with a new profile – so we’ll look at Roger Smith’s last 30 days.

Roger Smith Hotel has 11 times more “buzz” than Havana Central (and they work pretty tirelessly on achieving that, as well) and more than 3 times the amount of buzz of all 5 restaurants, combined, over the last 30 days.

But what is even more compelling is the amount of twitter followers (the opportunity any follower is going to see a tweet or re tweet is small – but when you have close to 700,000 Twitter followers who could have seen an offer or message – that is impressive, and can not be ignored (see below).

We know “OTS” or Opportunity To See isn’t the same thing as money in the bank but it starts to count a lot who might potentially see a post or listing.  Want to know who is making up the bulk of the OTS opportunities for Roger Smith Hotel over the last 30 days?   Think Paul Baron (the same guy I met a few months ago) – the “Agent of Change” and @Jakrose, etc – you can sort the River of News by Twitter Followers (too bad it doesn’t do Facebook Friends as well – oh well – next year).

Last year I spoke with Brian Simpson who started Social Media outreach at Roger Smith Hotel and he told me that he could trace as much as 20,000.00 per month in additional income to the hotel as a result of the Social Media Work he and his team are doing there – and I bet that he’s not captured everything – and he listed out what he captures – including Lily’s Restaurant – etc.

When you add all the people who stay at Roger Smith – like @iJustine and @ChrisBrogan, you kinda figure – that a good turn comes back – over and over – and mind you – the charts above just show activity over the last 30 days – the cumulative charts over the last two years would show far more.

But wanted to end this post on the starting point – the missed opportunities all the restaurants in this study have – there’s so many possibilities to get people coming in and saying they are “in the restaurant” which means – they’re spending money there – creating value – it’s a wonder that more focus hasn’t been put on Social Media, before now.

And here’s the activity over the last month as a trend chart

The peaks in activity at Roger Smith Hotel is closely tied to events – in this case (May 9th/10th) a Clambake on the 16th Floor Terrace and a Digital Somethings panel – I bet Brian is able to tie every activity at the hotel to increased income – and everyone who Tweets is usually with someone else, several people.

So lets take another step back and go to the restaurants – which are all comparable in prices for a meal – let’s eyeball it and say a typical meal for 1 is between $30-$40 dollars (in the “moderate range” even at Blue Fin, which is fairly expensive – I have been there.

Assume that your social media activities generates 100 people a month that come in due to social media – check in – and they are with at least two other people – (average reservation is usually for 3 people) – that means every check in is worth close to 100 bucks – every tweet could be worth that much as it’s a proxy for activity – we know these people who are tweeting are eating and drinking – and not all of the people in a party tweet – one will, maybe the others won’t.

If we look at 100 people (reasonable in Havana Central’s case – as we’ve documented that much, at least, in all three restaurant locations) per month – we get 10,000 dollars per month due to Social Media Activity – and that’s not counting all the people in the Follower and Friends stream who see the message – even if they don’t act on it.

Finally, the missed opportunity is in finding ways to distinguish people who tweet and check in who were coming in anyway (we should encourage this – give them a drink – that is what I helped come up with) vs those who would not ordinarily come in  – but will now – because they will get a discount via a promotion – or  a friend ate there and liked it (recommendation).

I don’t think there is any doubt Social Media Activity translates into Money – for now, we’re still using proxies, but in the future – we’ll have the exact dollar amounts – even now, we have a pretty good idea about Social Media ROI – and if you don’t believe me ask Brian Simpson over at Roger Smith Hotel – he’ll tell you.

But I can not end this post by throwing out how we can start recording new activity – a new visit vs repeat visitors (just like we do in Web Analytics).

Ok, say we go back to Havana Central – and talk about Twitter and FourSquare check ins / tweets.   Say we use Radian6 to export a River of News for as long back as we can go (in this case, over 6 months) and de-duplicate – so we have a list of all the people who have checked in – but only one instance of each.

Every month – we get a list of all those who have checked in or tweeted – and compare it to the list I have just mentioned above – if the tweet or checkin comes from a new account/handle (someone has not checked in before, or tweeted before about Havana Central) it counts as one acquisition  (or $100.00 as I suggested – it will all equal out even if it turns out sometimes less is spent and other times, more).

In this case, over the last month – these Twitter/FourSquare accounts are logged – next month if any of these show up they are tracked as customer loyalty while new visitors are treated as aquisitions.

The list of visitors who have returned in the following month gets handed over to the community strategist to reach out to – these are encouraged to come back in.  Meanwhile, the new visitors are also treated differently – and when possible – are asked how and why came in the first place – this information is put into a CRM (a spreadsheet will do as an interim solution).

That’s how you start building ROI – that’s how you start showing the results of what your doing – this is where it all starts adding up.  You need to build these lists and keep updating them  – then you have something you can take to the BANK.

Sentiment Analysis, as Matt Rhodes writes in theThe Problem with Automated Sentiment Analysis today that no tool (he reviewed) consistently distinguishes between positive and negative conversations.  What he doesn’t mention, and hasn’t considered is that many other applications of similar types of content analysis where the same problems are showing up – such as Copyrighted material on YouTube – it’s not sentiment analysis – but the same kind of arguments are coming up from people who’s content is being yanked off of YouTube because Automated Bots can not tell the difference between parody and remixing of content (which is allowed) and reusing copyrighted content – which is protected by the owner of the copyright (say, a television show like Sex in the City).   At the Re/Mix Festival in Dumbo I dropped by yesterday, I heard about this and more.

The same kind of analysis, more or less is done with Google Voice on telephone calls, and those are maybe, OK, if there isn’t noise in the background, but are horrible if there is.   At any rate, I look at Sentiment Analysis software as a technological “answer” to massive production of content that people don’t have time to read and sift through – it is coming in too fast for that.   Sure, if you really care about what your reading, if your brand that really cares about what people are saying about you – you’ll probably have to have human listening – but who ever said we should not have humans listening?  Why do we think that technology is the answer to everything?

Also, JaTin reports extreme gamers spend 48 hours a week on average gaming (farmville/fishville)? –  I think it’s far more than that for some people – try 100 hours a week- there are people I know of that spend that much time on it.  JaTin also reports on how FourSquare is being used in some unique ways for Social Justice.

And there was a big story on how Geo Services Mismatch Places Up To 40 Percent Of The Time in TechCrunch – funny, but the challenges of getting geolocation listings correct sound almost as bad as the problems of getting sites search friendly, for SEO – it’s like a new technology makes us have to go back and reconsider how we’re storing and presenting information that didn’t matter before the technology matured became real time.

Finally, the debate of “who owns” Social Media seems to have been resolved this year – the winner – Public Relations.

And if your into Iron Man 2 – guess what’s coming out soon – I think there’s even a Arc Generator built into the shoe (ha).

I missed the TechCrunch Hackathon in Manhattan last weekend (I’m not a programmer and it did not make sense to go) though TechCrunch covered the TechCrunch Hackathon in detail in a post titled The Best Of The TechCrunch Disrupt Hackathon.

Found some very interesting stuff being developed at  TechCrunch Hackathon – the tools and platforms have matured to the level where an application can be spawned by a talented developer in a matter of hours (though such an application may still have limited capabilities and not always be able to handle massive amounts of traffic or complicated queries / questions in a uniform manner yet).

Last weekend I mentioned TechCrunch Disrupt also reminded me of AlwaysON! Conferences  (both are three days and have many of the same speakers)- but certainly – this “HackaThon” is one way TechCrunch and AlwaysON! are very different -don’t think AlwaysON! conferences have software development happening at the conference.

One of the winners of TechCrunch Disrupt is a Hack created last weekend called “Twitter Demographics” – and while it doesn’t 100% work and seems limited on what it will report on – is truly amazing in what does do – take the query “oilspill” (the platform does not appear to work well yet with multi word queries – well, what can you expect with a hack created on the fly in a dozen hours or so?).  Twitter Demographics was one of the winners of the TechCrunch Hackathon – here’s what TechCrunch says about Twitter Demographics.

Twitter Demographics
Twitter Demographics mashes up Tweets with geolocation and demographic data. So if you search for a specific keyword on the application, you can see the percentage of Twitterers who Tweeted about the keyword within certain income ranges and even by political party, giving more info not just about where they are but who they are.

Read more: http://techcrunch.com/2010/05/29/best-disrupt-hackathon/#ixzz0pLQpNng5

I tried to get something to display in Twitter Demographics related to the Gulf Coast Oil Spill and managed to get a readout on “oilspill” though “oil spill” and “gulf coast” did not work.

Twitter Demographics on the Gulf Coast Oil Spill

A couple of things to consider here:

1. We don’t know the size of the data set being used – I think  a fraction of the Twitter Firehose is actually being used as the report only processes 12000 or so records – it maybe that all the mentions being using are associated with income, political affiliation and location ahead of time – not sure.
2. We don’t know how many of the records processed actually pertain to “oilspill” and the percentages of the pie chart in the diagram above are based on that number of records – what if only a few hundred records pertained but the rest were ignored?
3. So many queries you might want to know about do not work, that your fairly limited in what you can ask Twitter Demographics about today.   Certainly this would be great research for Brands – but it’s hard to get it to work well, yet.
4. Twitter Demographics is a Hack – it probably can’t handle a lot of people asking for information at one time.
5. The demographics and political data are somewhat open to question as to reliability and we need to know more about how that data is collected and used – having said that – what is assemb

Assuming the data Twitter Demographics assembled on what is essentially a hashtag about the Gulf Coast Oil Spill has any validity, here’s some insights that it suggests.

First, more than two thirds (68%) of the Twitter conversation about the Gulf Coast Oil Spill is from Republicans (and given the sharply polarized nature of our politics is almost certain to be negative to President Obama’s handling on the crisis and his leadership or lack of in this situation.  Without this “hack” it would have almost be impossible to make that particular insight.

Just the other day I noticed in my RSS feeds just how negative the conversations online have turned on Obama’s handling of the Gulf Coast Oil Spill Crisis – even though, in all honestly, there’s probably not much more he could do in terms of remedies to the situation than he has already done.  Could the Obama Administration have prevented the Oil Spill – technically – yes, without issuing permits to allow drilling – the disaster in the Gulf Coast would not have happened, and the latest set of permits come after Obama was voted in.

Let’s put it this way – there’s plenty of blame to  go around.  On the other hand, to foresee and fix every possible abuse and mistake in a agencies that have a long history of being over permissive – is a lot to expect, especially in a year.   I’m not saying the Obama Administration isn’t responsible – but it’s not the only place to look.

And, it’s  not as if the Federal Government / Coast Guard, etc (as Paul Krugman pointed out a few weeks ago it was the Republicans in the Bush Administration who crippled the ability of government agencies to respond to situations like this Oil Spill and 15 months in power is probably not enough time to correct for 8 years of more lax rules and abuse) can actually stop the oil spill all by itself – and while it could respond perhaps a little more forcibly – it probably needs the same entities that caused the problem in the first, to stop the Oil Spill – which is going on.

Also, Krugman points our to just how much Katrina Revisionism is going on now.

So, for Twitter Demographics to gauge close to 70% of the brouhaha about #OilSpill is from Republicans almost exposes the nature of the attack as a coordinated effort to make the Gulf Coast Oil Spill Obama’s and the Democrats Katrina – even if we suspected as much – we could never prove it before – but now, possible, a simple hack like this, exposes it.  Wow!  And Midterm Elections are coming up, so there’s even more incentive to find more fault with how this situation is being handled.  Nuff said.

And, if I’m reading the pie chart demographics correctly (with that rainbow effect, it’s not easy to separate one category from another when there are so many and the colors of adjacent categories are pretty close) a significant chunk of the audience for this #oilspill conversation is coming from the “Mining Industry” … we’ll, isn’t that interesting?

Assuming I’m correct, it would imply the entities that created this mess are among the most vocal voices here, along with Real Estate, Rental Leasing (oil rigs?) and Insurance.

Just to sum this post up – I’m in no way saying the Gulf Coast Oil Spill is handled correctly or not - and I’m not saying that Twitter Demographics is accurate, or not - and I have thrown out a lot of questions about the data and reliability of this information – but ……. what is in front of me is very interesting picture – and all done through a TechCrunch Disrupt Hack last weekend …. and if that isn’t “Disruptive”, I don’t know what would be.

You?