Nodes and Edges: Visualizing Network Data 101

Visualizations of your social data can be extremely helpful in developing marketing strategies, but how can you go about creating these graphs?

Google Knowledge Graph, the Facebook Graph, and the Social Graph — today’s digital marketers and brands love to throw these terms around like they are ancient, tested, and proven concepts that everybody grasps and understands. However, the truth is often very different.

Recently, one of our clients asked me to show him his brand’s "Social Graph." Not being a social expert, I had to do a lot of research and internal inquiries to find some ways to actually "show" him. Turns out it is a little more complex than people often expect. It’s not a simple downloaded Excel sheet you can get that shows your "graph." Instead, Social Graphs are basically a network of nodes and edges — of entities and the connections between them.

Nodes and edges? Confused? I was, too, which is why I want to share a simple way to analyze your social graphs or networks in order to better understand and visualize them. We will focus today around analyzing Social Network Graphs, but this approach can be used for any type of network data like link networks and website structures.

Data Concepts

Before we dive head-first into one of those "fascinating" screenshot-powered, step-by-step guides, I want to quickly address the data concepts behind graph visualizations. Most graphs are powered by a two-dimensional data system consisting of two core items: nodes and edges.

Nodes are the entities we are evaluating (People, Pages, Handles, Groups, etc.) and edges are the connections between them (Likes, Following, Friendships, etc.). Most of the network data today is handled via GraphML files or .gdf (graph data format) files. Basically, these are simple text files that contain a list of all the nodes and the relationships/edges between them.

Tools

Several tools can visualize network data and the most exhaustive list I know of can be found here. In today’s examples we will use Gephi to visualize our social data. Why Gephi? It’s free, open-source, cross-platform, and easy to use. It also has one of the most appealing visual outputs compared to some of the other tools out there.

Data Sources

There are thousands of ways to extract social network data like the native APIs, Custom Applications, Excel tools, and more. One of the easiest for data extraction AND visualization is NodeXL. It allows for fairly effortless extraction from multiple networks (YouTube, Twitter, etc.) straight out of Excel. It also allows you to visualize and customize them directly inside Excel.

For Facebook data (and for today’s example) I’ll use an actual Facebook app called NetVizz. NetVizz allows you to export your own "Friend" and "Like" networks as well as "Page Like" networks. Examining Page Like Networks is a great way to analyze audience affinities and learn more about target audiences by understanding common interest and connections.

Getting Started

OK, first we have to determine what we want to analyze. Let’s say we are thinking about having a booth, sponsoring, or speaking at the next ClickZ Live conference and we are trying to determine some of the common interests among attendees. What else do they care about, what do they read, or who are they connected with? Having these insights will help us to better understand the audience and determine how and where to communicate with them.

Getting the Data

The first thing we would need to do is to get the Like Network for ClickZ Live. In order to do that, we need to find out the numeric Facebook ID for the conference. The easiest place to get it would be http://lookup-id.com; it’s a free (ad-supported) site that allows you to enter a Facebook URL and in return get the ID. Once you have the ID extracted (in our case, 82891330657) we would go to the NetVizz Facebook page and select "Page Like Network."

Now simply enter the numeric Facebook page ID of your choosing and select a depth of 2. This will take some extra time, but it gives you a broader set that goes to second-level likes.

After a few minutes of crawling you will see a link that allows you to download a GDF network file.

Note: If you are downloading data for pages with millions of likes, this can take a few hours. But since this is a server side crawl, you are able to have multiple crawls running simultaneously.

Importing the Data

Once you have downloaded your .gdf file, start up Gephi and import it via File->Open. On the Import report, just leave the default options and click "okay." You will be presented with a somewhat odd-looking bunch of lines and dots.

 

Enhancing the Data

One advantage of Gephi is the easy-to-use implementation of mathematical operations. For our data, we want to do two things.

1. Click on "average path length" on the right-hand side. This will calculate the distance and betweeness centrality of our nodes (their centrality within our chosen network). It will allow us to understand their importance relative to the other nodes. Once the calculation completes, just click "close."
2. On the right-hand side, run Modularity. Modularity uses a community detection algorithm that allows us to group related nodes together (we will color code them). Click "close" once the calculation is completed.

After you run both of these, nothing will change visually, but we can now perform operations against these calculations.

Visualizing the Results

This is the fun part. Now that we have run our calculations, let’s start by sizing the nodes. On the top left side select the Nodes tab, then select the diamond icon (size) and choose "betweeness centrality." The minimum and maximum sizing depends on the size of your set; for this small example I would recommend minimum 10 and maximum 50. Choose "apply" and you should see that the nodes have adjusted their sizes.

Next, choose the Partition tab in the top left corner. Then select "nodes" and hit the green arrows in order to refresh the options. You should see the Modularity class option. This is the data we got from our community detection algorithm. Once you select this and hit "apply," the nodes will be colored based on the results of our community detection algorithm, according to their common attributes and relation to each other.

Now let’s give our results that awesome look. Underneath the Partitions and Ranking window on the left is a Layout option. This allows you to use different algorithms to lay out the nodes and edges. The best one for this type of data is Force Atlas. Simply select it, check "prevent overlay" and press "apply." You should be left with a view similar to mine below, which clearly displays the major and minor nodes as well as the connections between them:

But what are they? Use the three little icons highlighted above in yellow to reveal your metrics: The first one will show the labels; in the second one use the dropdown and choose node size; then use the slider (third one) to find a fitting size.

At this point it should like this:

There are a ton of adjustments you can make to sizes, colors, etc. to graph your data and see what’s really happening in a brand’s social network but this is not bad for five minutes of work. Now you can start to zoom in, move and highlight nodes. As an example, when I hover over the ClickZ Live node, I can clearly see the biggest affinities:

Playing around with the data a bit reveals some interesting connections. During this exercise, for instance, I discovered some patterns from pages that indicate they either paid for their likes or made all their employees like their clients’ pages (but I won’t call them out publicly; can you find them?).

Another insight from my ClickZ Live example is that comScore is the biggest common denominator outside of ClickZ’s own properties.

There are countless deeper analysis models you can apply in Gephi and build on the existing data such as PageRank and Clustering.

The visualization below is what I eventually sent to my client to show him his brand’s social network. I generated it using Force Atlas, Page Rank, and Modularity and then added some transparency in the Preview Dialog.

I hope this inspires you to perform this type of visualization and get some great insights into your brand’s graph data. Questions? Feel free to message me at @nxfxcom.

(via)

Data Deconstruction: Understanding Complex Social Influences

When you deconstruct and simplify your complex data patterns, you can better understand the data and build better social media marketing programs. What are you waiting for?

With the holidays approaching, I’m hearing the familiar descriptors in campaign and interaction data charts: no doubt you’re hearing them, too. "This time of year, we typically see a dip (or rise — I hear both) in traffic as business slows (or ramps up.)" I started wondering: How much of this is accepted as belief, and how much is based on data? And more importantly, how can data deconstruction and modelling be used to extract really useful information, supporting better-performing social campaigns and programs?

A few years back (actually, quite a bit more than a few) I worked for NASA|JPL on the Voyager spacecraft. In order to prepare for planetary fly-bys, we’d simulate maneuvers, use of cameras, power up the various detectors, etc. As we did this, we captured data: my team was responsible for understanding how ambient temperature affected the spacecraft’s radio receivers. As we turned instruments on and off, and as we rolled the spacecraft’s receivers toward and away from the sun, we’d record the temperature of the receivers, gaining information that was critical to keeping them locked on the Earth.

So you’re wondering: what does this have to do with understanding the performance of my marketing campaigns on the social Web? A lot, actually. Like the temperature of the receiver, your campaigns are impacted by some things you do directly, and by some other things that occur and thereby impact your programs indirectly: external events, competitive announcements, etc. I wrote an article a while back about the importance of launching campaigns on specific days of the week on this same topic.

In the case of a metrics following marketing campaigns — tallying social data around the interaction of customers with each other in a discussion forum — there are of course direct drivers of behavior: if you change the offer, if you alter the forum navigation or change creative elements, then you’re likely to see a change in traffic, conversion, bounce…the set of metrics that depend most closely on the parameters or attributes you just changed. No news there.

But you’ll also see other changes: the effects of holidays, or an event in the news, a product recall, etc. It impacts your results, so you have to explain it. Even better — after all, if it happened once it will probably happen again — you can try to understand these factors and build them into your planning model. How? You can use the use the same techniques I used at JPL to pick apart the various contributions of each of these factors. The benefit? Rather than assuming, you’ll know if what you are seeing is because of the holidays or instead an indication of an underlying cause that you really need to pay attention to.

Here’s how it works: Take a look at the charts in Figure 1: on the left is a typical representation of weekly social interaction in a community: weekends are the low points, with visitors more active during the week. To the right are two different "trend" interpretations of the pattern in the data.

In the center chart you see a typical linear trend line: Excel will do this easily. The point here is that by choosing the start/end points it would be just as easy to redraw the trend line sloping downwards or flat as is was to draw it upwards. Now take a look at the chart on the far right. Instead of a liner trend, the chart is split into two parts, with a flat trend line fitted to the each of these. The insight is this: if you strongly believe there are no influences on your results other than the changes you are making, then the data likely falls into a single set and the center chart (the slope of the line notwithstanding) is the right analysis. But that’s not always the case, and splitting data set may help you spot external influences.

If there are also external events occurring, splitting your data may be better approach: deconstruct your data into sets of components and see. Artist Piet Mondrian made creative use of deconstruction: start with the complexity of real life and reduce it — deconstruct it — to more clearly show the fundamental elements that collectively contribute to the complexity observed. Check out the progression from left-to-right, as the artist’s early work slowly gave way to a radically deconstructed view of the world.

To deconstruct social marketing data, start by separating local periodic cycles (weekly traffic patterns, for examples) from the larger global (external) influences. Create a model for the local cycle — in this case the weekly up and down that repeats regularly and lay it over the observed data. In the chart shown on the left in Figure 2 I’ve created a curve (using the polynomial trend line option) that fits the weekly data. On the right, the cycle has been plotted (red) over the original data using the baselines from Figure 1, far-right chart. The fit looks good, especially when compared with the single linear trend.

A couple of notes: nothing about this analysis says it any more "correct" than a single trend (or any other tracking model). Rather, the idea is to look at your data in more than one way, and see if the alternative views produce insights that lead to a better understanding of what is actually happening. Ultimately, the decision is yours: it’s your "marketing gut" that has to check the validity of the results.

Take a look at the charts in Figure 3: I’ve removed the original data, leaving only the model (red, chart on the right) representing the curves that were created based on the model of the local cycle and suspected discontinuity (the separation from on data set to the next). On the right, the model itself is de-emphasized (now shown in gray) to highlight the chart feature of interest: the discontinuity in the data between the two baselines. The task now is to explain this, and if it’s material to predict its impact on future campaigns.

Why go to all this trouble? Rather than trying to make sense of a graph like the original chart containing all of the complexity of the real world, it’s often easier to look for a simplifying model and then deduce why the specific characteristics of that simplified model do (or do not!) apply to your actual results. In the case described in this article, a half-year of cyclical data has been reduced to a single repeating curve and one discontinuity — a step change — in the data. This helps to zero in on the data itself and ask the important question: "What happened right here?" If an answer is found, that provides a big clue into understanding external events that may impact future campaigns. This is information that can be used to make better (in the quantitative sense) statements about how campaigns actually perform, and therefore to make better decisions on how and when to run them.

(via)

Dark Data and Measuring Invisible Impact in Your Facebook Community

In digital communities, good brand managers spend their time observing and interacting with the hundreds or maybe even thousands of fans who engage daily by commenting and posting. These vocal fans are incredibly important, but they represent only a part of the story. Over 80 percent of brand fans are lurkers – quiet fans that neither comment nor post.

As the largest social community, creating Facebook posts that appeal to your silent majority is as important as engaging your loyal brand advocates. The silent majority refers to fans that are being reached by your brand, but don’t take the actions to Like, comment, reply or share that we typically see from top advocates. They’re statistically some of your most informed and loyal customers, potential future Superfans, and the primary providers of the social currency needed to foster your active advocates: validation.

To appeal to these fans, brands need to gain insight into what they aren’t doing. Marketers need to start focusing on three key metrics to read the dark data that often slips through the cracks for this fan segment: clicks, churn and engagement over time.

Measuring silent fans: Clicks, churn, and engagement over time.

If you can’t measure it, you can’t manage it. Even though this is dark data, measurement is still possible. Some of this information is available immediately within the data Facebook already provides, and some, like fan history and tracking engagements over time, is more readily available through 3^rdparty technology providers. There are a number of supporting metrics like your subscription rate; ratio of fan replies and Likes to fan comments; and interaction on fan shares that you can layer in to give you a more nuanced perspective, but the basic lurker landscape can be derived by looking at clicks, churn, and engagement over time. 

Clicks are a key metric provided by Facebook on a post by post basis, as well as a weekly aggregate. They refer to fans who have clicked into a News Feed post but then didn’t otherwise engaged with that post. With Facebook’s new definition of their Engagement metric that now tracks clicks, this metric can be used as a powerful proxy to determine lurkers’ dark activity.

The most effective way to monitors clicks is looking at clicks/engagements and clicks/reach. Engagement is a catch all that includes shares, Likes, comments, and clicks. Watching how clicks as a ratio of engagements and of reach perform across different post types will give you a good indication of what best hooks your lurkers, and may vary greatly from what hooks your active advocates. Be aware though, that the occasional uncropped recipe photo or hotel pool shot might still appeal to lurkers in the feed without requiring a click through.

To combat lost clicks, one way to start measuring dark data from lurkers would be to work teaser photos into your post mix. Teaser photos are visually grabbing, intentionally cropped photos that hint at a punch line or fuller story that can be reached by clicking through to an externally hosted image or page. Compelling clicks in this manner provides a positive fan experience while presenting a very low obstacle to engagement.

Churn rate, or the rate of fans leaving your page, is a key measurement to determine what silent fans like and don’t like about your brand. Measuring your churn rate regularly and tying it to specific posts or campaigns can act as quantitative analysis that will, over time, allow you to create content that appeals to your fan base as a whole. Since dark data pertains to the majority of your fan base, overall churn rate provided by Facebook serves as a useful reflection of what resonates with your silent majority.

Timing of engagement is a crucial piece of dark data when measuring individuals in your Facebook community. For active fans, use a technology provider or manual research to see how much time elapsed between when they initially Liked your page and when they first engaged. A long average time suggests you’re bringing lurkers along for an enjoyable ride until they’re ready to post, but if the vast majority of your fans engage immediately or not at all, you’re likely not appealing to your passive fans.

Measuring timing between engagements and week over week retention from active fans can also provide valuable dark data on what resonates with these fans; if they start to post more frequently or go through inactive periods, mapping that data to the types of content posted can inform brands about how to stimulate the best discussion with varying segments of their fan base.

Using Dark Data for Good

Once you’ve identified your silent fans, keep track of new engagements as a percentage of overall engagement. Pay attention to the content and initiatives that reliably bring back hardcore fans week over week, and see if it differs from the efforts that get newbies to engage or lurkers to look. This unique perspective can be informative beyond Facebook to influence underlying values of your brand marketing.

Think of dark data as a type of sonar - you’re constantly sending out signals, and now learning to pay attention to what comes back as a method of measuring the shape of your community. As such, benchmarking and reliable measurement over time is essential.

(via)

Related articles

• 5 Things to Think About When Creating a Social Media Content Strategy
• Social Media Branding: 16 Tips to Create a Consistent, Relevant and Trusted Social Brand
• 3 Ways to Identify Brand Advocates
• Nielsen: TV and Online Advertising Will Merge by 2020
• How to Analyze Your Facebook Metrics to Improve Your Marketing

Facebook Testing “Hover Metrics” And Other New Behavioral Data Collection

Earlier this week the head of Facebook analytics, Ken Rudin, said that the company may begin capturing even more data about how users behave on the social site. The data collected could include minute interactions or more specialized information than has previously been tracked.

According to a report in the Wall Street Journal, Rudin told a conference in New York that the company would potentially be collecting data on, for example, cursor movements and “hovering”:

The social network may start collecting data on minute user interactions with its content, such as how long a user’s cursor hovers over a certain part of its website, or whether a user’s newsfeed is visible at a given moment on the screen of his or her mobile phone, Facebook analytics chief Ken Rudin said Tuesday during an interview.

Facebook is testing various scenarios. It’s unclear whether any of them will be implemented, however. The additional behavioral data, including “hover” metrics, could be used for both site design and ultimately ad-targeting purposes.

Even if adopted it’s unlikely that any of these micro-metrics would find their way into explicit ad targeting or analytics from a marketer perspective. Quite the opposite: Facebook has been trying to streamline its advertising programs. Along those lines the company recently introduced a simplified ad creation flow using a marketing objectives-based framework.

The company has been criticized for not delivering more clarity around ROI to marketers. Accordingly Facebook is trying to simplify and offer new ROI tools that make advertiser value more obvious and transparent.

(via)

Related articles

• How to Analyze Your Facebook Metrics to Improve Your Marketing
• How to Build an Analytics Package to Measure Off-Site Content Distribution
• The Forgotten Social Media Metric: You
• 5 Pinterest Metrics Every Marketer Should Track
• New Facebook APIs Will Provide More Detailed Data to Brands

New AdWords Estimated Total Conversions Tracks Consumer Purchases Across Devices

Starting today and over the next few weeks, Google AdWords will roll out a major reporting update to conversion tracking called Estimated Total Conversions. This feature provides estimates of conversions that take multiple devices to complete and adds this data to the conversion reporting we see today.

Following the launch of enhanced campaigns this year, search advertisers have combined mobile and desktops with the ability to further modify bids by mobile and other targeting factors. One gap in reporting and comprehension of the campaigns effectiveness has been the limited data on how consumers are navigating and converting via multiple device options.

What is a Cross-Device Conversion?

Consumers constant connectivity has enabled them to browse, shop, and interact with businesses on the go and from multiple devices.

A September 2013 Google study found that more than 90 percent of multi-device consumers move sequentially between several screens like mobile to desktop, or mobile to tablet to complete a transaction online. Google found that a high percentage of converters actually jumped from desktop to desktop too, presuming a work desktop to home desktop computer.

How Estimated Total Conversions Works

Google calculates cross-device conversions for a particular advertiser based on how their customers convert when they are logged in. They then use this as the basis for extrapolating out to the complete data set to form an estimate of what total conversions that cross devices might look like. This data is only used in aggregate and not personally identifiable.

(via)

From Keyword to Products Sold

Here are some thoughts to leave you with when you optimize your campaigns:

• Understanding this relationship overall, and even at a category/subcategory level, can be helpful when thinking about pausing or activating sets of keywords
• Use this data to find out what things consumers are buying with various keyword sets. If, for example, you find that a large number of people buy smartphones after searching for the keyword "digital camera", you might have a cross-sell or optimization opportunity. The same can be said when you're out of stock where you might redirect someone
• Consider the variances in your data set by search engine. This is true for more things than the keyword to product purchased relationship. It can include bids, ad copy, landing page, and other performance metrics.

Leveraging data is one of the reasons why we all love paid search. Finding new and interesting ways to cut that data up to identify optimization worthy insights is what will ultimately set apart the best performing campaigns.

(via)

Related articles

• Items to Consider in Competitive SEO Analysis
• How To Assign Value to Your Brand's Digital Properties
• Social Media and SEO: They Love Each Other
• 12 essential PPC landing page success factors
• Make Keywords Fit a Marketing Messaging

5 Ways to Make Web Analytics Data More Insightful

Fortune 500 companies can expect to push more than a gigabyte a day in raw web analytics data, which can be easily tripled for media companies. Big data is anything anyone ever talks about anymore, so the C-suite has never been more interested in integrated analytics, shining a spotlight on the web analyst team to deliver more than just pretty charts and high-level talking points.

Pulling the information from web analytics software should be less than 10 percent of the work, with an overwhelming 90 percent of time dedicated to deriving insights your organization can use to drive change.

Facebook To Roll Out Email- and Phone Number-Based Ad Targeting Next Week

Facebook will be launching new features next week that allow advertisers to target their ads to customers based on contact information that the advertiser has already collected. It’s a way for businesses to connect their Facebook ads with the customer lists they may have built up elsewhere.

Inside Facebook first reported on the feature after tipsters saw it go live temporarily. A Facebook spokesperson confirmed this afternoon that it’s a real product, and she walked me through how the program will work.

Again, it starts with a customer list that a business has already created — for example if I’ve given my email address to the bookstore on my block so that I can hear about future sales and events. Businesses will be able to upload those lists of email addresses, phone numbers, and user IDs to Facebook, though the data will be hashed first so that Facebook doesn’t have access to that information. Meanwhile, Facebook’s user data will be similarly hashed, so the company can compare both sets of hashed data, creating a list of users whose contact information matches up with what the advertiser uploaded.

After that, businesses will have the option target their ads at that group, or they can further target their content towards a certain demographic (say, females between 25 and 45) within the group. The simplest use case: Most businesses have loyal customers who aren’t Facebook fans, so they can create an ad for those customers asking them to become fans. Advertisers can also offer deals — an auto repair shop could tell customers that they’ll get a free oil change if they become a fan. It’s applicable beyond brick-and-mortar businesses too — an app developer could target lapsed users with an ad outlining the features in a new update.

This is just about giving advertisers more targeting options, Facebook says — businesses won’t have access to any additional user data.

Facebook’s spokesperson says the feature has been in private testing, and that in one early campaign, a financial services company was able to double its fan base in two weeks, at a lower cost-per-fan than ever before. Which makes sense — the company was reaching out to people who were probably fans already, just not in the Facebook sense.

Starting next week, this targeting feature will be available to all “managed” advertisers (i.e., the ones who receive support from Facebook).

via

Related articles

• Facebook's New Post Targeting Options Launching Soon, Learn How To Use Them Now
• Targeting Facebook Posts: The Good, Bad, and Intrusive
• 7 Ways to Prospect for New Customers With LinkedIn
• Is Instagram Right for Your Business?

Five Tips for Developing an Adaptable Approach to Online Marketing

1. Start With a Goal

If you are going to go to the trouble of creating content and publishing it online, you need to have a specific goal in mind. As Odden

 asks, “If there’s no reason for it, then why do it?”

The specific goals of your online marketing strategy depend on your business, your customers, and numerous variables, but you must have goals. “If a corporation is publishing content just for the sake of publishing content, then they’re not that different from a spammer,” says Odden. So, don’t be a spammer! Have a goal.

2. Develop a Hypothesis

With your goal in mind, you begin selecting and implementing the tactics you will use to achieve it. Those tactics may include creating a blog, launching a podcast, devising cool infographics, or doing something crazy that no one has ever considered.

No matter which tactic you choose, you need to have a hypothesis about what’s going to happen. How exactly are things going to work? Developing a hypothesis requires you to think things through and to have a way of figuring out whether your efforts have borne fruit.

3. Monitor and Measure Progress

Your hypothesis said, “If we do X, our customers will do Y and that will result in Z.” You did X. Did Y and Z follow?

To answer that question, Odden says, “We need a methodology for monitoring our progress.” A methodology will have at least two components: Key Performance Indicators (KPIs)—the actions or events you want to measure—and some way of collecting data. The beauty of online marketing is that, by virtue of its being online, we can collect mountains of data. The difficult part is figuring out what type of data is meaningful and what is noise. You will need to look closely at what measurable actions are most closely correlated to which measurable results.

4. Analyze and Revise

You have a goal and a hypothesis. You’ve identified ways of measuring your progress. And you’re actively gathering data to “measure actual impact.” Now, you need to analyze that data and draw some conclusions. Is our strategy working? If so, why? How could we make it work better? If not, why not? What do we need to do differently? The purpose here is, as Odden insists, ”to extract insight to further refine our goals.”

Going through this whole process is pointless unless you take the insights you’ve gathered, reapply them to your efforts, and strive for continuous improvement.

5. Adaptation and Optimization

It’s a cliché to say, “Change is the only constant.” However, that cliché happens to be true. An adaptable approach is one that allows you to adapt to change. The only way to do that is to clearly perceive what is happening and to adjust your actions to fit the changing circumstances. Developing and test hypotheses, and learn from these activities. By doing so, you are also following the path of optimization, which Lee describes as “continuous effort to make better, make perfect.”

And isn’t that the path we should all be on?

Related articles

• Why Measuring Social Media MATTERS

  ---

• Why Online Brand Marketers Better Get an Interest in Pinterest, Fast

  ---

• Facebook is Easy if You Master These 2 Tips

  ---

• Digital Marketing Strategy Development: 12 Common Problems

  ---