Parrot Analytics utilizes a stringent validation process such that unless a data stream has been verified to be 100% attributed to a title, it will not be included. If we encounter an ambiguous data stream, we prefer to err on the side of caution and not include it. While our manual verification steps takes a significant amount of time, these ensure that our demand measurement is the most accurate in the world.
For all shows, especially those with ambiguous or generic names (e.g. "Scandal", or "Class"), rigorous processes are in place to ensure that all activity that is captured is directly relevant to the show itself, rather than any other references.
For example, a specific question may be how we differentiate between "Class" (the TV show), and "class" (a generic word). To answer this question, let us examine the specific on-boarding and validation process that is applied.
Title Validation Process
To validate a title, the following are just some of the quality assurance steps that are taken by our engineers prior to a title being fully on-boarded and available in a customer's subscription:
- We plug into a host of metadata providers to automatically extract a) show-specific metadata like episodes, and b) demand expression platform tags like social media handles, hashtags, video streaming channels and tags, community pages (etc.) for each show.
- We haven't found a combination of sources above that are exhaustive, so we also collate and update our own databases that we can interrogate, especially for blogs, fan sites and the like (e.g. the US version of House of Cards picks up traffic to this Wikipedia page and not this one).
- For downloading/streaming sites and peer-to-peer protocols, we also operate our own multilingual system that checks specifically for each show at an episode level.
- Individual title validation requires a manual verification step for every data stream in the first instance. This ensures the data streams are directly and accurately attributed to that show. This is the most painful (but necessary) part of the process.
Steps 1-3 are designed to be quite stringent, which means unless we have pre-verified a data stream to be 100% attributed to a given show, it will not be included. In addition, the manual verification step is critical to ensure the highest levels of accuracy.
Where a source is ambiguous, we prefer to err on the safe side and not include a specific stream. There are instances where we may be missing out on some show-specific data (mostly conversational data) that simply cannot be 100% attributed to a given show. We are, however, always working to capture more data streams; our process as stated minimizes any serious errors by ensuring streams are not on-boarded unless we are able to verify its attribution fully.
Once all the title validation steps have been completed in their entirety, the title is fully on-boarded and made available in our customer's Demand Portal account.