Inconsistent Dataset

· Different tools have different filtering capabilities

· Solutions can only draw from public Facebook posts to protect privacy.

· Different tools have different access to the Twitter “firehose”.

· Crawlers and spam filters also affect dataset results.

New Behaviors

· Social media creates new behaviors that must be interpreted, and the value must be understood. Examples:

o A “Like” on Facebook.

o A re-tweet on Twitter

o A check-in on Foursquare

· Answers vary based on industry and business objective.

Disparate Sources

· Most brand conversations occur off the main website, outside the reach of traditional web analytics providers

· New apps generate data from an ever-increasing array of sources, each with different characteristics.

· Social analytics solutions are still new; few case studies from which to learn.

Language Limitations

· Industry terms

· Slang and abbreviations: LOL, OMG, TTYL, ROFL

· Irony and sarcasm

· Emoticons J

· Uneven support for global languages

Different Analytical Approaches by Vendors

Differing approaches to data collection affect results:

· Keyword-based is the simplest and least expensive, but least accurate.

· Natural language processing and algorithmic approaches are more sophisticated and expensive.