There were social feeds to be seen throughout all the halls at NAB. It seems to be a requirement to engage with your listeners and viewers. Some people have stayed with a simple twitter feed, some have brought in feeds from Facebook; more recently all the integrators are expanding to bring in multiple feeds, adding graphics from the likes of Instagram. Twitter is by far the most popular and active stream for text messages.
There are two main problems when harvesting messages. The first is being able to connect to the feed in a way that you can handle it, the second is to be able to pick out the best items from the feed and not let the bad stuff through. The Twitter “firehose” can pour out huge amounts of data that most clients cannot handle, hundreds of millions of messages per day. For this reason, Twitter only allow normal connections to read up to 10% of the messages and only in batches of a few hundred. They have their trusted distributors that do have access to all the data, and who then can give clients a more tailored feed at a rate that they are able to handle. These distributors like DataSift, Gnip, MassRelevance can sell services and also run algorithms to give trending and other aggregated data. Other sites such as Topsy give an interactive interface for accessing instantly. There are a growing number of specialist companies to provide aggregation services. Because there are so many variations, there are a variety of solutions and a variety of pricing models to choose from. A more full list of integration services is shown here.
Different manufacturers have approached it in their own way. Chyron have had their Shout system working for over a year now. Initially it had the input from MassRelevance and never.no, but that seems to have been a format copied by others such as Telescope to provide more choices of supplier. The Shout system will bring in the stream for which Chyron charge a monthly payment, but the customer will also need to buy the Engage engine to map the fields on to the templates in the graphics system. Moderation is simple with a click and drag interface for moving messages about.
VizRT were also using never.no last year. This year they have their own new software package that connects to the tweet streams. It can use the public streams or it can connect to suppliers of larger amounts such as those above. Theirs is a one off price. Once the package is paid for, the insertion into the graphics comes for free. To get more numeric or aggregated data, they suggest the use of a connection to Trendrr which may cost you more. The system is more comprehensive in that the streams can come from a number of popular services, bringing input of graphics from Instagram directly into the templates. If a specialist stream can be manipulated into the same format, the stream could also be fed into the system. The moderation problem is handled by having the ability to attach five moderator screens who have the option of which groups of messages to put into the output graphics. Click and drag of messages is similar to the Chyron options.
The problem with most systems is that if you have a popular show, it become difficult to keep up with the number of messages arriving. Some sort of filtering needs to be done and this is where the skill lies. The options that the new distributors give are their skill in stripping out the rubbish and finding the important people to listen to. Some give aggregation to produce trending values. Some will say they can show the good and bad opinion levels around a topic, but this has not proven to be an easy task. The positive/negative terminology used in tweets is usually hard to decide. The location may be more useful; the number of tweets in particular cities or countries can be shown on the Viz graphics for example.
Sometimes it is not the top tweeter that you need to follow; you need to be following the conversation. The people with the most followers are not necessarily tweeting about your chosen topic; it may have built up from some totally separate subject. This can mean that a companies social team will be the ones to lead the searching. They will be talking to the followers of the programme or team and will be building up and promoting the conversations. They may drive conversations through polls and giveaways. They will be actively building the number of followers and should be able to give some structure to the twitter search. The tools required for this job are those that are springing up from smaller companies at NAB. never.no have been around for a while; others that I saw here were Quest on the Miranda stand, who are developing from other data feeds; Zoomph who have had much success in the Sports arena; BannisterLake are selling either a cloud based moderation and tweet out system or a small server that can be positioned in-house. Many of these companies have a cloud based filtering system. Some have moderation. These will generate a stream of the messages that you then need to feed into your graphics.
SocialNewsDesk is more a tool for expanding your social presence. It allows polls and tweets to be sent out of a company easily and can help generate a larger following on the main Facebook, twitter streams by asking questions, offering prizes etc.
What are the costs? It is difficult to compare some of these services as they are not comparing apples with apples. It will be best to try some of them and get a full quote for you circumstances. The main suppliers from Twitter will be charging a few dollars for every 10,000 tweets they send to you. It’s important to try and judge the number that you will be receiving. At higher levels it may be worthwhile to commit to a monthly contract which would give higher limits and some flexibility about trying new search terms. I suggest you try and work out a flexible contract that may be changed easily.
Take a look around the web sites to get a better view of the business. They are all trying to adapt quickly to new trends, so you’ll find that the initial choice is not always the best in the long term.