All right, the last component I want to show you is the comparison tool. And what the comparison tool does is allow us to take this monitor and compare this monitor to other buzz monitors that we might have created. And so I've created another monitor for Starbucks competitors. So lets just pull that one up, just to give you a quick comparison. Starbucks versus it's competition, looks like Starbucks has about the same number of positive, but more negative conversation. If we look at Specific categories. Again, we'll go back to that Starbucks monitor. And what category? Positive, neutral, negative. We can see the difference over time. All right, so we've got our current monitor, the Starbucks monitor. And if we want to look at positive versus positive, looks like relatively similar, but if we want to compare negative versus negative, right? So Starbucks we see has had these higher spikes in negative conversation, whereas the competitors that we've identified, did not. So for Share of Voice, going to give us essentially a relative volume comparison. So there we are. So Starbucks seems to dwarf it's competition, in terms of the volume of conversation occurring around it. Okay, so comparing Starbucks conversation to the volume of conversation around the competition, notice that Starbucks is in terms of volume 90% versus 12%. So Starbucks has nine times the volume compared to the competition. So really dominating the conversation in this particular category. We have, at least based on the competitors that I had identified. If you want to look at how the volume itself is changing over time, then that's something that we're able to look at. And something that's probably worth noting. Okay, is, and this is something that's worth looking at for us, those two big spikes that we identified, the raced together conversation and the red cup conversation, big spikes for conversation in Starbuck. Notice no shift in terms of the conversation around the competitors. So, they are not capturing a share of voice around those events. When the competitors have spikes in volume, we dont see much of an aboration in terms of the volume of conversation around Starbucks. So that's probably, yeah, we're not seeing those competitors linked in terms of the conversation. There has been work that's found where conversation around competitors, specifically around product recalls, we have seen that competitive spillover. These events may not be triggering any conversation around those competitors. So that's the buzz monitor. Lots of built in capabilities. Let's go back to our home screen. And what I'd like to show you next is the social account monitor. Now social account monitor, this is going to be tied to the Twitter account. So very similar in terms of construction, we specify the Twitter handle that we're searching for and the timeframe that we're going to be looking at. And what's nice is if you're monitoring the conversations around for a specific brand, we can look at what the conversation is coming from the brand itself, that's what the social account monitor is going to do for us, as well as the conversation mentioning the brand. And so first on, essentially, our summary screen, and this is only looking over the last month or so, we see how many posts have gone out from Starbuck's, that's in this screen below. We can see metrics surrounding retweets and replies about that. We also get a measure of total impressions. Now, impressions to sent posts, that's based the followers that you have, okay? So, potential impressions calculated by adding senders, followers, and the followers of all users who have retweeted. So we're trying to get a measure of how many eyeballs we're getting. So in this case, whose following Starbucks and for all those people who have retweeted that post, how many followers do they have? Right, and so over the last month, looks like the biggest spike we got is up to 60 million as far as potential impressions. And this one will start to see a lot of interest around influence for marketing with social media because if we consider how many followers the brand has, how many followers influencers might have, that's potentially a lot of exposure. Right, yeah. We can again export to Excel, number of retweets, number of replies to each of the posts that have been pushed out by the Starbucks account. We can look at this engagement over time, broken down into retweets, replies and mentions. We can also look at follower growth over time, now this is only looking at a one month period. If we want to go back to let's say the beginning of 2015, to how our monitor was set up, we see from 2015 to the present, looks like Starbucks has grown from just over 7 million to just shy of 12 million followers. Very rapid growth. Seems to have tapered off, at least on Twitter. All right, and as you'll see on the side, we get very similar tools. The Explore tab allowing us to dig into the content itself. These are posts of individuals engaging with Starbucks. If we click on this similar type of information, cloud score basically neutral is the categorisation. What's the engagement sentiment, positive, neutral, negative. And so this is also using those automated tools. So based on the sentiment of the replys and mentions to that Twitter post. So that's the ability to look at the particular account. The last monitor type that I want to show you is the opinion monitor. Now the two that we've looked at so far rely on automated tools. So this is kind of the one size fits all model is you have a sentiment analysis algorithm, we're going to apply that sentiment analysis algorithm for every monitor that you create. But in some cases that's not going to be entirely appropriate and so that's where the opinion tool comes in. You can use it for sentiment classification, to say positive, neutral, negative but you can also use it to classify the topics of posts. All right, so let's say I'm Starbucks and I want to understand the volume of conversation around different topics, around service quality, around pricing, around products, around jobs. Well I can create those different categories, but I've got to train the algorithm to put posts in the categories that they belong to. Well the way that this has to work is somebody's going to sit down at the Crimson Hexagon platform. All right, so let's open up this opinion monitor. All right, so this is the opinion monitor we're creating and I haven't trained it yet. All I've said is we're going to use the search term Starbucks. So step one in this algorithm, it's called Bright View, is going to be creating the categories that are of interest to you. Step two is the training period. So that individual assigned to doing the work within Crimson Hexagon is going to train the algorithm by manually categorizing the posts that come in. Once you've categorized enough of them, and the more you categorize the better, Crimson Hexagon's algorithm is going to automatically categorize the remaining posts based on the similarity of those posts to the ones that you've classified. And so let's just begin walking through the training. The tools that you'll see on the side are going to be very similar. All right, so first we have to begin by building a category. All right, so let's say what I'm interested is the products or the drinks. Do the posts mention the products or drinks? We might also be interested in service quality. And perhaps the last category that we'll look at will be the facilities. Maybe these are the categories that I wanted to find. Okay, so are they talking about drinks, are they talking about service quality, are they talking about our facilities, or is it unrelated to what I'm looking for? All right, and then what we have to do is just go through and drag and drop. All right, so there's a little girl in line at Starbucks with a shirt that says My Daddy is My ATM. All right, it doesn't really relate to my product, my service quality, my facilities. So we're going to categorized that as off topic. We're not really interested in it. All right, I need strawberry acai refresher with lemonade from Starbucks. All right, well that's one of our products and drinks, let's categorize that one. We've got a Starbucks coupon. Someone likes their frappucino. Now I'm going to loosely categorize that as a product or drink since it's mentioning that. Chilly mocha. All right, that's one of our drinks. I'm at a Starbucks in Des Moines. All right, well, that's one of our facilities. My mom just got so hyped about the wine and beer at Starbucks. That's the products that we carry. And so as we start to categorize these, notice how this percentage starts to fill up. All right, I'm going to drop some of these categories, since we don't seem to have too many posts pulling that in. So let's change these slightly. All right, and we'll just add one more category, let's say Starbucks. So let's say it's jobs or corporate. So, management going that would fit under that. All right, so I'm going to take some time to just do this categorization and you'll see as this monitor fills up, we'll be able to actually implement the algorithm. All right, so what I've done is, just for a demonstration purposes, done our classification. And notice, I had to get to at least ten in each of these different categories. We get the 100% complete. Once we've done that, I can click on run Bright View. What that's going to do for us, that's going to automate the assignment of other posts to these different categories. So if we click on run Bright View and then it's going to take some time for that classification to occur. Now, we can always go back and train the Crimson Hexagon platform with additional data. That data being provided by a user. And so if you wanted to classify positive, neutral, negative, you can do a classification based on that. If you wanted to classify emotion, you can do a classification based on that. The only limitation, in a sense, is your categorization scheme, you cannot have a single post fall into more than one category. But the Bright View algorithm gives also a lot of flexibility in terms of tailoring what we are looking for. If you are looking for a retailer or if you are a retailer and you wan't to understand why are people shopping here versus shopping online. Why do they come into this store versus go to our website. Well we can build categories based on the conversation that's happening on social media. So those are the capabilities of the Crimson Hexagon platform itself. As I mentioned at the start, a lot of power built into the automated analysis tools that you get under the foresight platform, the buzz monitor, as well as the social account monitor. If you need additional customization that's where the opinion monitor comes into play. Opinion monitor requires that someone sits down and do that training period. And the more training, the better but as few as ten posts per category, and it's able to get up and running. So in the next module, we're going to take a look at data that we've pulled out of Crimson Hexagon and doing some basic analysis with it as far as spotting deviations in behavior, places where we may want to drill down a little bit further to find out what's going on.
