Last week I co-hosted a webinar entitled Boosting Your Brand Marketing with Facebook Topic Data Insights. On the webinar we talked about how to use Facebook topic data for use cases such as measuring brand health, campaign analysis and identifying viral content. If you missed the webinar you can catch up with it here and you can read in detail how to use PYLON for Facebook Topic Data to get these insights on our developer site.
On the day we got a lot of great questions about some of the finer points of Facebook topic data, I’ve reproduced some of those here (and the answers, obvs) as I thought a wider audience might find them interesting. So, in no particular order:
How is sentiment being calculated merely based on Facebook data?
Sentiment is assigned to the posts by Facebook on a three point scale (positive, negative, neutral) based on text analytics.
How are you measuring brand health? Is it only the volume of discussion? Or volume of discussion with positive sentiment?
Facebook topic data can provide insights into some of the key components of a brand health analysis including both share of voice and sentiment analysis. You can compare the volume of stories for brand and their models and also the sentiment associated with those stories. You can also create custom tags and scoring rules to further monitor your brand attractiveness and loyalty among your target audience group.
Can you extract feature level sentiment?
Yes! You can create custom tags to identify features in posts and find the sentiment associated with those posts.
Does Facebook provide demographic data?
Facebook topic data includes self-declared age, gender and geographic data.
Do you have data for Thailand and other ASEAN countries?
We do have data for Thailand and all other ASEAN countries (Indonesia, Malaysia, Philippines, Singapore, Brunei, Cambodia, Laos and Myanmar). The complete list of available countries can be found in our FAQ.
Is this in real time, so you can adjust the campaign to use the insights while it’s live?
The data is recorded in real-time. You can make adjustments to your campaigns and update your recording filters to see your campaign in action in real-time.
What is the process to actually analyze this data? Are the platforms shown available within the DataSift platform or are they created by data analysts independently?
The analysis results from your API query are returned in JSON format. The sample charts and graphs shown in the webinar were visualized by importing the data into a third party data visualization tool. If you are looking for insights to help your marketing (rather than providing a product or service to others) then we recommend you get in touch with one of our certified PYLON partners, who will be able to provide you with easy access to Facebook topic data insights. You can find them on our partner portal.
Will the celebrities names associated with the product be picked up automatically by the system or we have to query them by name?
Celebrities names are usually created as topics by Facebook’s Graph. You can query top topics and filter the categories of actor, musician etc. to find out which celebrities are being referenced in relation a product. Alternatively, if you are looking to track specific celebrities you can create custom tags using our VEDO rules engine to tag them.
Is sentiment available for different languages?
Facebook topic data includes sentiment detection for seven different languages. They are English, French, German, Italian, Portuguese, Spanish & Turkish. If you have your own sentiment detection using text analytics for other languages that can be applied via VEDO.
Are we able to drive down to the specific comment text level when viewing audience engagement?
In order to protect the privacy of the people on Facebook, topic data is aggregated and anonymized. However, we do provide a sample of verbatim text from Super Public posts to validate your analysis results. Super Public text samples come from posts that are made public, posted by an author with the follow setting turned on, and not posted on another person’s timeline.
Is it possible extract an anonymized timeline of shares of a specific URL?
Yes! You can query for the specific url and see how engagements changed over time. This is alluded to in the Speed of Content Spread section of this article on our developer site.
Is there a taxonomy available that could help us build classifications in a more automated way, instead of having to provide many different keywords when writing a filter?
Yes! You can classify using the topics from the Facebook graph, these come pre-populated in the data. For detailed information on topics, please see this in-depth guide.