Facebook topic data

New product ideation

Another arm of BIS provided media insights and intelligence to Fortune500 companies. This unit based in Paris needed a tool to automate tasks of data extraction and analysis for their clients. All incoming requests had been fulfilled manually, through searches in Newsdesk and monitoring alerts set up across all social channels.

I participated in a design sprint to create a prototype which would take Facebook topic data – anonymised by Datasift, a third party vendor – and present key demographics, sentiment, trending topics and keywords for use in analysis and trend forecasting.

This project didn’t advance past the prototype I created. News was that Facebook was reviewing its topic data policy and so while LexisNexis awaited the outcome, the project was put on the back burner. 

One of the key challenges was designing the initial search form. The search form dictated that multiple translations of a given search term could apply; not only English. In order to search accurately across the full dataset, translated keywords were vital. I championed that this should not be the onus on the user to think of this, or to also provide the various translations. For the purpose of testing, I wanted to present a way to automate this process.

But it still required the constraint of which language/s to be applied to provide the sample keywords.

There also existed the issue of disambiguation, as seen in the above. For example, how could the tool differentiate a search on keyword ‘apple’ as in fruit, or ‘apple’ as in the company. Again, for the purpose of testing, I provided keyword suggestion tool to facilitate this.

Another substantial part of the prototype was creating the journey from once a request is initiated, to what the user sees on a dashboard while it is being processed, and how the request manifests itself as visualisations.

Full version of the work-in-progress prototype. Onscreen prompts mimics the desired user journey.

Running alongside of this, I worked with the user researcher to help grow user panels for research and testing sessions. I organised an on-site visit with industry peers whom I had worked with previously. This helped facilitate a view of how some industries constructed their social media analysis and where their challenges lay.