Request a Demo
 
“Avatar” still leads, blogs make up biggest block of Best Picture talk

I’ve been writing about Spiral16’s ongoing 2010 Oscar study this week. We are measuring the power of online buzz about the Academy Awards and by Sunday night, we’ll know if the amount of online chatter for the top three contenders had any connection to the ceremony’s Best Picture winner.

Two things about the way Spark is collecting data:

1. The search terms we used were the titles of the nominees and the words “best picture” to narrow the focus to sites that are discussing the nominees only in relation to the Oscars’ top prize. (To find out how to build better queries yourself, see Monday’s post.)
2. We are monitoring all Internet URLs in general and not merely social media websites. That way we can cast a wider net and see how press coverage and blog comments play into the scenario. There are a lot more contributors to online buzz than just social media websites.

Avatar is currently in the overall lead, so the graphic to the right represents the site type breakdown for URLs mentioning Avatar and “best picture.”

Below, however, we’ll combine the stats for all three movies and view the study as a whole.

This is how the breakdown of sites looks so far for frontrunners Avatar, The Hurt Locker, and Inglourious Basterds It’s actually very similar to the Avatar study:

  • Blogs have accounted for 41% of the total URLs collected. This is where most of the conversation about these three nominees and the term “best picture” is taking place.
  • 34% of the URLs came from News-oriented sites like the Los Angeles Times. Again, this is a pretty big share.
  • The Social category (Twitter, forums, microblogs) amounted to 18%.

    (Twitter itself is the most influential domain in the entire study with a whopping 9% share of the total URLs in the study so far. Although its share of the pie may be small, it’s the biggest individual slice. This illustrates just how saturated the Web is with 2010 Oscar talk right now and how widely it is spread out.)

  • Shopping websites and reference sites like Wikipedia comprise the General category, which only amounted to 4%.
  • YouTube and other Video sites accounted for 3% of the total volume. YouTube itself is the second biggest domain in the study, comprising about half of all video sites.
  • Also, it’s important to mention that each of these samplings are being done independently of each other. This means that if Avatar is mentioned in the same post as The Hurt Locker, that URL will be accepted as a relevant result in the individual studies for each movie. That way, one film is not able to “steal any votes” away from another.

    To illustrate just how widely spread out Oscar talk is all over the web, take a look at our 3D visual map of the Avatar study. The spheres represent URLs and the lines are links between them.

    Green spheres have an average of positive sentiment, red are average negative, and gray are URLs with an average neutral sentiment.

    This is a visual illustration of how many websites have Oscar talk related to these movies and how they all are fairly disconnected to each other. To explore a working Spark 3D Virtualization for yourself and see what this is really like, click here.

    Tomorrow, we’ll look at sentiment—some specific overt stuff and how sentiment relates to the big picture.

    Tags: , , , , , , , , , , , , , , ,
    Posted in Case Studies, Content Of Interest | No Comments »
    “Avatar” and “Hurt Locker” lead in online Oscar traffic, “Basterds” a distant third

    In yesterday’s blog, I went into detail about how we set up our 2010 Oscar prediction study. Using Spiral16’s Internet monitoring tool Spark, we aim to find out if there is any correlation between online chatter about the top three Best Picture candidates and the result of Sunday night’s awards ceremony.

    The overall percentage for each of the movies hasn’t changed that much since yesterday. Remember, the URLs deemed relevant for this study are only the ones that also mention “best picture.”

  • Avatar is still in the lead, gaining one percent since yesterday, with 43% of the conversation.
  • The Hurt Locker remains in second and goes down a percentage point, with a 32% share of online mentions.
  • Inglourious Basterds loses one point, going down to 25%.
  • Now let’s look at how the results differ in the three separate searches related to each movie. Looking at the Semantic Cloud, you can get a good idea of the language people that are using when writing about the Best Picture race. Although Inglourious Basterds has only 7% less of the share of conversation overall, it’s status as a distant third place is solidified by looking at the total count of words used.

  • In the Avatar insight, “hurt” is the fifth most-used word, “locker” is the sixth, and “basterds” is 27th. Look at the Semantic Cloud of words used in the Avatar study below.
  • In The Hurt Locker’s results, “avatar” is the sixth most used word and “basterds” is 26th.
  • In the Inglourious Basterds insight, “hurt,” “locker,” and “avatar” are three, four, and five.
  • Looking at this, it’s pretty clear that Basterds is running a distant third. It should also be noted that no other nominated movies’ keywords (Precious, Up, Up in the Air, The Blind Side, etc.) showed up before Basterds, further solidifying its place as the odds-on unrewarded bronze medalist .

    It’s also worth mentioning that The Hurt Locker has such a solid presence in the word cloud overall.

    Other interesting sidenotes:

    In all three studies, “oscar” outranks “academy” and “awards,” while “oscars” is right behind them.

    This is no big surprise to learn that people are using the slang terminology more than the official title of the ceremony.

    With another search today, I also confirmed my suspicions from yesterday with data from Spark. I contended that:

    The Nielsen Company, in a post from yesterday, appears to have made a major goof in their flawed Best Picture study. Not only did Inglourious Basterds come in last in terms of total buzz—not a chance that is accurate—but in their data chart, the movie is spelled incorrectly as well.”

    Using two different queries, one for “inglourious basterds” and one for the more subtly misspelled title “inglorious basterds,” I discovered that 46% of all people referencing the film spelled it wrong like Neilsen did. This means their study didn’t pick up a whopping 55% of the traffic that spelled the film’s title right.

    We’ll have more data insight tomorrow, so stay tuned.

    Tags: , , , , , , , , , , , , , , ,
    Posted in Case Studies, Content Of Interest | 2 Comments »
    Can online monitoring predict the Oscars?

    With the Best Picture category at this year’s Academy Awards expanding to 10 nominees, this year’s race may be the least predictable in recent memory.

    Or not.

    Nobody really knows, since it is more than just the number of nominees that has changed. The Oscars are also employing a tiered voting system where Academy members rank the 10 films from best to worst and the scores determine the winner.

    Even though the Academy of Motion Picture Arts and Sciences are the only ones whose votes actually count, I thought I’d take a look at the amount of online chatter about some of the nominees to see if there is any correlation between the movies people are talking about and the end result of the Academy’s voting.

    Using Spiral16’s Internet monitoring platform Spark, I started tracking posts about the two movies that are generally considered to be the frontrunners—Avatar and The Hurt Locker—and the film that many say could be the spoiler: Inglourious Basterds.

    Remember, Spark is not limited to merely social media sites. Also, because Spark is not limited to RSS-generated results, it pulls in a huge amount of data.

    On Monday, I posted about setting up good search terms to get great results. This is absolutely key. For this study, I searched the title of each film and the words “best picture” to see only the posts that were equating the two in some fashion.

    For The Hurt Locker, I left out the word “the” and for Inglourious Basterds, I searched for both “inglourious” and “inglorious,” knowing that everybody will catch the obvious misspelling of “bastards” in the film’s title, but might miss the more subtle misspelling of “inglorious.”

    (The Nielsen Company, in a post from yesterday, appears to have made a major goof in their flawed Best Picture study. Not only did Inglourious Basterds come in last in terms of total buzz—not a chance that is accurate—but in their data chart, the movie is spelled incorrectly as well. Guess we know why it came in last! I’m also wondering how they solved the Up/Up in the Air problem, but that’s another story.)

    Thankful that they are not considered to be in the top three, I purposefully ignored nominees Up and Up in the Air because differentiating the two would be nearly impossible and lead to a lot of bad data. (You can add “Pixar” as a search term for Up, but how many people are actually mentioning the studio when they talk about the movie’s Best Picture chances? 50 percent? 40?)

    I set the date range to bring in results going back to Feb. 10, the day the ballots were mailed to members of the Academy. Hopefully, this brings back a total that is more relevant to the actual vote, as it reflects the online culture during the time when voters will be making up their minds.

    Here is the big-picture Best Picture overview as of noon CST today:

  • Avatar leads with 42 % of the three-movie pie. It is the highest-grossing film of all-time, after all, but keep in mind these URLs are only the ones that also mention “best picture.”
  • The Hurt Locker has the second biggest total share of online chatter, with 33%.
  • Inglourious Basterds comprises 26% of this three-picture study.
  • Below are some random snippets that bear out some of the bigger themes of this race.

    Although Avatar is the most talked about so far, it also is the film that’s been seen by the most people. That means we have to assume that not all the conversation surrounding the Best Picture race is about how it’s going to win. (Check out the very first conversation excerpt.)

    Inglourious Basterds’ spoiler chances are a popular topic, as you can see by the snippets in the third box below.

    Come back tomorrow for more interesting data in our 2010 Best Picture race case study. After the Oscar telecast on Monday night, we’ll know if there’s any connection between online chatter and who wins the big prize.

    Tags: , , , , , , , , , , , , , , ,
    Posted in Case Studies, Content Of Interest | 1 Comment »
    Define and Refine: 5 Steps to Build Better Searches

    One of the most frustrating elements of any Internet listening project can be setting up the search terms that will result in the most relevant data. If you’re not careful, valuable time can be wasted on a bad query while irrelevant results pour in. (For an example of bad queries, look at this post.)

    Your search is more than just a search. It’s the first step of a research project. Our job is to make your project successful, so we’ve had a lot of experience setting up and testing search queries to give clients the best possible statistics.

    Here are the steps you need to take to define and refine your query:

    1. Know what you’re looking for—have a defined topic.

    If you went into a library and told the librarian you were looking for stuff on history, the first question any good librarian would ask is “What kind of history?” Otherwise, you’d be poring through the entire breadth of human history just to find information on poodle skirts.

    It’s all about narrowing down the field. If you want to find out whether a used 2003 Ford Expedition truck is a good buy or not, you’re one step ahead of the game because you’re looking for a specific make, model, and year. A search for “Ford” isn’t going to do you any good. You have to define what you are looking for specifically because the search won’t do that for you. It gives you everything.

    2. Think like a machine—narrow your topic into a concept.

    Let’s say you want to run a query about politicians who put a lot of political pork in the bills they submit to Congress. A machine doesn’t know what “pork” means. Machine language would happily spit back recipes for baconated grapefruit with the same enthusiasm as information about a spending bill. You’re liable to get a lot of irrelevant results unless you bear that in mind.

    3. Search for what people are saying, not what you want them to say.

    This is a common pitfall. If you are looking to define what customers who frequent movie concession stands are saying about the products there, don’t search for “concessions.” Chances are, few of the relevant URLs will contain this word. Use common language. You’ll get better results if you think about how people actually talk.

    People are more likely to be talking about the popcorn, hot dogs, or Cokes that they had at the movies. They may also talk about prices. All of these are important words to consider.
    Another example: Use “laid off” or “lay off” before you use “unemployed.”

    4. Find them and learn their language.

    Sometimes you have to do a little research before you start. You wouldn’t go to a mommy blog to find out what people think of the public option. Do a test queries to find out where the conversations about your topic are happening.

    Now it’s time to find out specifically what’s being said and how it relates to your concept. Read the blogs that cover your topic. Even better—use a tool like semantic analysis to learn the language that your targets are using. Now that you know the lingo, it’s time to refine your queries.

    5. Know that you may not find what you’re looking for.

    Expectation management is important. As shocking as it may be, there probably aren’t a lot of people talking about your original Cap’n Crunch/James Bond fan fiction or the rubberized oscillating spammer widget that you marketed to the Amish.

    That could mean many things. Are you not getting the right results because you haven’t used the right terms to find them or because the results you want aren’t there? If nobody is talking about your widget, look at it as an opportunity! Running a query on popular Internet widgets might reveal to you that the Amish aren’t real big on computers—and that’s valuable information.

    Tags: , , , , , , , , , , , , , , ,
    Posted in Content Of Interest, Resources | 1 Comment »
    New Design: A Look at Insight Explorer

    This weekend saw a new rollout for Spark that streamlined the user interface, making it easier to see and interact with important information at a glance. One of the most noticeable changes is that the Insight Analytics page is now known as Insight Explorer.

    This section of Spark is laid out in an intuitive fashion and has improved functionality. The new Insight Explorer section gives a more in-depth look at each data gathered for each individual insight by including a complete list of relevant URLs. These URLs can be sorted with the new global filter to find and highlight the material that is the most important to you.

    InsightExplorer

    The Bulk Options box can be checked to automatically select the individual boxes to the left of each URL. Use that option or select nodes individually and apply the Delete Selected Nodes feature or the Export Data icon.

    Insight Explorer is laid out in an easy-to-read format that gives you a snapshot of the most important information, with the ability to delve further into each URL. Here’s a glance at some of the new features:

    Site Type – Each site type is color-coded and has a corresponding abbreviation.

    SiteType

    Sentiment – The circle next to the site type graphic will either be colored, red, green, or gray to represent the negative, positive, or neutral average sentiment level of each URL.

    Page title – The title of the URL is displayed just to the right of the sentiment rating. It is bolded and in blue. If you click on the title, you will go to the Site Details page for that particular URL where more detailed information about that web page is shown.

    Domain – Under the title on the left side is the domain of the URL. If you click on the domain, a direct link to the specific URL will pop up in a new tab in your web browser. If you roll your cursor over the domain, the full URL will appear.

    Influence – This ranking orders each node in your insight according to Spark’s influence algorithm. This is the same proprietary algorithm as the Node Influence ranking on the Top Engagements Targets gadget.

    Summary – This displays text from the URL that was indexed from RSS feeds or search engine APIs, and is often—but not always—the first words of the post. Each site will be different.

    OptionsOptions
    • The Delete Node icon will delete the URL.
    • The Favorite icon highlights that site by adding a favorite icon next to the sentiment score. Favorites can be used to develop a Daily Growth Report. Click the icon a second time to un-favorite a site.
    • The Notes/Tasks icon allows you to add notes or tasks to the URL.

    If the options exist for that URL, one to four feature icons may appear along the bottom of each listing.

    • The Excerpt Available icon appears if Spark has captured at least one excerpt of overt sentiment. Click on the icon and a new box will appear below the listing with the text of the excerpt, its sentiment rating, and the date it was captured by Spark.

    ExcerptText

    • The Notes/Tasks icon appears if one or more notes or tasks have been assigned by a user within the insight. Clicking on this icon takes you to the Site Details page, where all tasks and notes assigned are displayed.

    • The Video Embed icon and External Comments icon will appear if a site offers API access to extra information such as videos or comment streams. Clicking on these icons takes you to the Site Details page, where the URL’s embedded video and comments section are displayed.

    Tags: , , , , , , , , , , , ,
    Posted in About Us, Company News, Works In Progress | No Comments »