How Location-Based Content Impacts Influence
As we mentioned in our previous post on disambiguation, identity is an important factor in finding influencers and correctly assigning them scores.
A valuable attribute of identity is location. Where the influencers live, work, or write can help determine interests, topics, and reach. Also, when we have location information about influencers, we’re able to disambiguate and resolve the identity of the individual, creating valuable demographic data points that we can provide to our clients.
In the new social media landscape, influencers’ ability to reach across geographic boundaries has increased, and the importance of geographic location has diminished. For example, an Arab Spring tweet can reverberate internationally, and a YouTube video can help a Korean pop group go global.
Yet traditional boundaries persist – political, economic, cultural, and linguistic barriers are often heavily connected to particular locations in the world. As consumers, our relationship to a place often determines whom we listen to and what we buy/sell.
Static vs. Dynamic Location
When we process opinions for our influence analysis platform, we are concerned with two types of location: static and dynamic.
For instance, mobile Twitter clients allow individuals to geotag their tweets when on vacation, a dynamic location, or they may specify a permanent (static) location. This poses an interesting problem – which one is more important?
While we may use information about where people are talking or tweeting from to come up with that influencer’s location, we prefer to resolve an influencer to a place where they are most closely associated (static location).
We believe that building campaigns and advocacy programs are about long-term relationships with people, not fleeting moments like business trips or vacations. We don’t need someone showing up our customer’s radar just because she took a weekend trip to Chicago, but really lives in San Diego.
Reliability and Specificity
Geolocation is notoriously difficult to pinpoint. While some sources (like geotagged tweets) are reliable indications of location, many other sources are not.
For example, user-provided data on blogs or social media can be unreliable. Some people say they live in “dreamville” or “In my own world”, while others list only the their country and others name their street corner. And, of course, some people forget to update their location when they move.
Whether out of a sense of humor or desire to remain anonymous, user-provided location is varied in accuracy and specificity. At Appinions, we prefer to resolve individuals to single states (or countries for international influencers).
In our system, we process opinions across a variety of sources. Some of these sources, such as Twitter, are identity-driven. We can reliably make use of the user-provide information about their location to figure out where they are located.
Other sources, such as news and blogs, provide us with opinions where the author is not location-bound. A blog or newspaper may pull a quote someone from halfway around the world, for example.
In those cases, we may know the location of the author or publication, but the location of the influencer it is a little less clear. We manage this ambiguity by looking at all the pieces of evidence associated with particular influencer persona. These include:
- Distribution of all the locations associated the publications in which that person has opinions;
- User-provided locations;
- Locations mentioned in influencer’s opinions; and
- Locations in user-specified meta data (such as biographical description).
We look at this evidence, normalize the information to a consistent level of specificity within a specific state or country (e.g. “Socal” = California) and assign a location to the persona. We then combine the information from those personas along with a measure of reliability associated with each source to pick an individual location for that individual.
We’re excited about the impact that our recently added the location value will bring to influencer export reports. Each influencer will now have his or her own location field, and 50% to 65% of our influencers have reliable location values with more being added each day.
As we work to improve the coverage and quality of this information, we will be further integrating the location information into the platform and providing the ability to filter influencer lists. We’re maniacally focused on supporting our clients with their influence marketing programs – that starts with finding the right influencers for the right place.
Image credit: zhrefch