How SXSW staff monitor social media to track rumors, and identify safety and crowd issues during the festival

The SXSW Music, Interactive and Film festival attracts a sea of people to Austin, Texas every year, and they come armed with smartphones and social media profiles. The result is a massive stream of social media updates that break news and signal important events.

“Sometimes a band won’t make it to the festival, but they’ll tweet about it before they call us or their management,” says Melissa Golding, SXSW’s Digital Content Manager. “A lot of the time, we can verify rumors on social media faster than we can call out into the field.”

But in order to verify rumors, and spot an emerging issue, Golding and her colleagues realized they need better ways to quickly separate the signal from the noise. The volume of tweets, Facebook posts and other social updates makes it hard to extract insights.

Golding and her team turned to Hootsuite and Nexalogy to help SXSW make sense of its social data and conversations — in real-time.

The two companies collaborated and the result was the combination of the Hootsuite social media dashboard with a custom social analytics and conversations dashboard built by Nexalogy.

SXSW teams used Hootsuite to monitor hashtags, filter streams and engage with people. The Nexalogy dashboard (below) visualized the social discussion, and also delivered automated alerts centered on specific categories, such as safety, police and crowd.

The custom monitoring dashboard Nexlaogy built for SXSW

The custom monitoring dashboard Nexlaogy built for SXSW

“We comb through thousands of tweets on any given day,” Golding said. “The visualization Nexalogy provided highlighted things as they happened, which we were then able to set up as searches in Hootsuite to track and archive. Without the visualzation, we wouldn’t necessarily be able to tell if certain words were trending and needed our attention.”

Golding’s digital content team used Hootsuite and Nexalogy dashboards constantly. Staff at the SXSW Communications Centre (seen below) also used the tools to monitor activity.

“The two teams are in constant contact, making changes to our filters or setting up new searches based on information we’re getting from our boots on the ground, social media, press and other entities we work with,” she said.


Inside the SXSW Communications Centre, with the Nexalogy dashboard front and center.

Once the festival was underway, they customized the Nexalogy dashboard and Hootsuite filters to look at specific keywords, hashtags and other items.

“Were able to make some changes on the fly to really fine tune the filters which brought us useful information such as what people were saying around things like lines, crowds, rumors about Rihanna, etc.,” Golding said.

The result was SXSW could quickly spot emerging issues, such as lines that may be starting to get too long, or places where the WiFi needed to be stronger. They then dispatched people based on what they saw happening on social media.

“Nexalogy helped us manage our event in real time by cutting through the noise and letting the things we had filtered for come to the surface where we could easily digest the information,” Golding said.

An analysis of the C2MTL speakers, concepts and quotes that stood out

Even before C2MTL started, Cindy Gallop made her presence felt.

Gallop is the founder of Make Love Not Porn. She’s an engaged and savvy social media user, and a great speaker. All of those reasons, among others, led her to be one of the main stage speakers at C2MTL, a unique and remarkable annual business conference that bills itself as being “somewhere between genius and insanity.”

As the big data partner for the event, Nexalogy gathered and analyzed all of the social media data and conversations taking place on the #C2MTL hashtag.

We used this data to create a dashboard showing key people, concepts and tweets trending, and to provide a daily briefing for media at the event. (Nexalogy CEO and founder Claude Théoret also spoke on the Garage stage.)

Beginning with the day before C2MTL, we saw that Cindy Gallop was a social force to be reckoned with. She began using the hashtag even before departing for the conference:

After landing in Montreal, she tweeted her hotel room, promoted her talk the next day, connected with other participants over drinks. Then, once on site, she connected in real life with people, and interacted with anyone who reached out to her on Twitter. She also fell in love with shoes designed by Montrealer Anastasia Radevich and grabbed a pair to wear on stage for her speech.

That of course resulted in more tweets:


Gallop wasn’t just active — she was interactive.

That stood out when we looked at an interaction map in the Nexalogy platform. It shows the activity between Twitter users on the #C2MTL hashtag. Gallop was represented in a big way, meaning she was very active and influential. But she was also talking with people, rather than just being talked about.

The size of the below dot representing Gallop’s Twitter account shows that by the end of the event, she was being mentioned and interacted with more than just about any other speaker. The blue color of the dot indicates that she was sending and receiving tweets – meaning she was the focus of real conversation. (Green dots indicate a user that is being mentioned, but isn’t interacting as much.)


Ideas that Caught on

C2MTL is a conference of ideas, but which ones stood out? Using a lexical map, we were able see the key concepts taking hold, and their relationship with each other. Not surprisingly, some of the top social media memes at C2MTL were the result of things said on stage. Here are three speaker quotes that stood out.

1. Mad Men to Math Men

SAP CMO Jonathan Becher hit home with this statement:

We saw it emerge as its own cluster on the lexical map, and it stayed strong throughout the event. This is illustrated by the connections between words, and the size of each node:


In the end this was one of the most talked about quotes from C2MTL.

2. Broken iPad

Noreena Hertz had many quotable moments in her talk on the first day. One that jumped out was an anecdote she told about her young niece:

We saw from the lexical map that it ended up being discussed by many people who were not connected to the #C2MTL hashtag, meaning Hertz’s iPad comment spread farther than the conference community: 4

3. Doggy Daycare

Zappos’ Tony Hseih had many great insights in his presentation, including, “A great brand is a story that never stops unfolding.” But we soon saw that his biggest meme was… doggy daycare:

When you think about it, this makes sense: it was by far the most requested amenity from his employees. So why wouldn’t it also be the most talked about part of his presentation? It seems humans everywhere love doggy daycare.


All About the Jameses

C2MTL lived up to its billing as an event all about commerce and creativity. The top two concepts discussed, aside from C2MTL itself, were “business” and “creativity.”

When looking at the core cluster of event concepts below, we see that C2MTL is also about ideas, innovation, and social media, among other concepts:


But what about that big concept labeled as ”Jame”? It has to do with two people.

The first is filmmaker James Cameron, who spoke and also announced that he and Cirque du Soleil would launch a world touring production of Avatar. The second “Jame” is James Murphy, the former LCD Soundsystem front man who did a DJ set one night.

The two Jameses generated a lot of discussion, and our system captured the root of their first names to indicate that.

That leads us to offer a bit of data-driven advice for next year’s event: More people named James!

(And more Cindy Gallop.)

What happens when social data meet statistical analysis

As the old saying goes, sometimes people miss the forest for the trees.

It means they’re too focused on a small aspect and miss the bigger picture. When it comes to analyzing social media, however, the only way to gain true insights is to examine the forest and the trees.

You need to recognize the big patterns (the forest) as well as investigate specific pieces of content that make up the whole (the trees).

One of our best tools for enabling customers to see the forest and the trees is the lexical map.

The Lexical Map

A lexical map is what happens when social data meet statistical analysis. The map uses a large data set — tweets, Facebook posts, blog posts etc. — and illustrates the concepts that emerge, and how they’re connected to each other.

Concepts are automatically sized based on the volume of data behind them. The more times a word is repeated in the data, the bigger it is on the map. Here’s an example of a lexical map at what we might call the forest level:


A lexical map of Twitter discussions about the Montreal Canadiens hockey team.

“When we collect the data the system reads every word,” says Nexalogy Chief Analyst Zach Devereaux. “As it’s processing something such as blog posts or tweets, it’s assigning values to the words used, determining how frequently they appear within the data set. It also analyzes how often a term is used in conjunction with other terms and then visualizes the network based on their connections.”

The clusters of concepts are a result of this process. With the map generated, it’s up to our analysts to begin interpreting and scoring the data. They zoom in on concepts and read the content associated with them.

Conversation Sculpting

Nexalogy CEO Claude Théoret says that social data is “a raw material that can be collected, stored, carved and sculpted into new objects with a purpose.”

That’s what our team does with a lexical map: we sculpt the results into something that reveals key insights. This process involves using our software to view the content (such as tweets or blog posts) associated with specific concepts.

“It’s about making sure you have a data set that represents the problem at hand,” says Mark Goren, Nexalogy Director of Client Services. “We’re looking for authentic conversations and voices that enable us to derive insights from the aggregate of all the data. With the map, we see the relationships not just between the words themselves but the content behind it. That’s very powerful.”

Ultimately, he says, “we sculpt the data to hone in on the content that helps us solve the client’s problem.”

Here are two of examples of doing that.

Customer Insights: Terra20

Terra20 is a retailer of environmentally friendly products with stores in Ontario. Prior to opening its first location in Ottawa, Terra20 asked Nexalogy to analyze blog posts from female bloggers in the target market. The goal was to understand more about what local mothers care about, how they talk about environmental and health issues, and to determine a smart strategy to reach them on their own terms.

Nexalogy designed a blog search to pull in posts that contained mentions of motherhood and parenting in an eco-friendly context. The team then examined the data on a lexical map to extract the top concepts from these blog posts. What emerged was a picture of concerns and passions that fell into four categories: family, the environment, home, and health and product purchases. You can see them labeled on the below map:

Screen Shot 2014-04-16 at 4.17.43 PM.png

The team was able to drill down and see what the people were talking about in each area, and to even generate lists of the specific brands and products they were talking about.

This, along with additional information about the womens’ views regarding green products, was key intelligence for Terra20’s social and product launch strategies. It helped them ensure a successful launch.

Needle in the Haystack: Petro-Canada

Another project where a lexical map delivered real insight was when Petro-Canada engaged Nexalogy to provide a detailed examination of its reputation on social media. There was simply too much information on blogs and elsewhere for them to process it and extract key insights efficiently.

Nexalogy conducted a search of Canadian bloggers and the conversations taking place around Petro-Canada, its projects and issues.

By closely examining the map and the conversations represented, Nexalogy was able to identify an emerging issue prior to the Winter Olympics in Vancouver. The relevant concepts are highlighted in this map:

Nexalogy identified an ongoing and accelerating discussion taking place around native land rights and the Games. The map showed how the relevant concepts were connected, and Nexalogy analysts dug deeper to read the specific blog posts.

“The discussion around land rights wasn’t the biggest conversation on the map — but that’s the point,” says Goren. “It was an early warning of something important, and it came long before this issue hit the press and the public consciousness. The map enabled us to locate that faint but essential signal.”

The result was that Petro-Canada’s risk management team was able to refocus their efforts around this emerging issue, and avoid a possible public relations issue.

You can quickly generate a lexical map based on activity in your Twitter timeline by using our free NexaMe tool or HootSuite app. If you want a more detailed demo, just get in touch.