We’ve arrived at the sixth and final post in our series ‘getting the most from Facebook topic data’, and this time we’re going to look at how you can productize Facebook topic data to get the best results for your clients.
PYLON comes with many out-of-the-box features, but products that enable self-service exploration of Facebook topic data can deliver rich new insights about audiences and the topics they engage with.
The most successful products built on PYLON focus on use cases that best leverage the functionality of PYLON and the richness of Facebook topic data and this is how we recommend approaching building a product on top of PYLON.
Getting started: prerequisites & prototyping
Aggregated and anonymized data – the data you get from a non-public network such as Facebook – is fundamentally different from the data that is made available from public networks. It is best to explore the data by focusing on a real-world customer question, following these tips:
Ensure recordings have sufficient volume – in general, it’s easier to get insights out of a recording that contains at least 100,000 interactions. This volume enables deep analysis across the available dimensions. Brands that do not attract this quantity of brand mentions and product names should think about casting a wider net at an industry level.
Start analysis with a wide filter and then narrow in – when working with Facebook topic data, avoid starting analysis with very focused and detailed queries that may not return any data. Instead, start with broad queries and work through the data to the level necessary to retrieve the required insights.
Get the most from out-of-the-box features- it’s also important to make sure you leverage the full potential of PYLON’s out-of-the-box features. These are designed with the express intention of extracting insights from non-public data, and deploying these will stand you in good stead.
Think about differentiation
Everyone that uses PYLON has the exact same set of data attributes to leverage, such as topics, demographics, links and hashtags. To create a unique product to suit your own specific needs, differentiation is essential.
Apply context – Using VEDO, DataSift’s classification engine, you can apply domain expertise to group topics, keywords, links and other attributes into themes and categories.
Create multi-dimensional views – Facebook topic data allows for the creation of interactive multi-dimensional views of the data which brings the various attributes, such as age and topics, together.
Make filter creation simple – you should always start with a simple set of broad filtering criteria and refine primary filtering criteria over time as information is learned from the dataset. This avoids the need to ask end customers to exhaustively elaborate their filtering terms up front, which can be tedious and time-consuming.
Dashboards vs. drill-downs – although users may find the high-level dashboards useful, the real value is in the specific elements that drive the summary metrics, so incorporating drill-down functionality into products and highlighting all dimensions of the dataset will of real benefit.
Archive insights –PYLON’s privacy controls mean recorded interaction data is retained for 32 days, but the results of analysis queries can be permanently stored. Doing so builds an archive of insights, which can be used to measure the evolution of the data and results, and puts marketers in a unique place where they have a set of historical insights that no one else has access to.
Approaches to commercialization
An initial point to consider when commercializing your product is what level of service your customers will want. A managed services approach can help develop expertise on PYLON and define a product development roadmap. This could take the shape of writing filters and performing analysis for end users, or running custom projects to help end users gather actionable insights from Facebook topic data.
While every company is different, most customers expect to be able to record consistent quantities of data and execute analysis against that data consistently. The PYLON platform has a number of controls in place to manage data and query volumes:
PYLON identities and data volume packaging – a PYLON identity is a bucket or folder that contains a group of recordings for a particular customer. Multiple identities can be created inside a PYLON account; create as many identities as needed in order to manage recordings across multiple customers.
In order to guarantee a particular daily capacity level to a customer, ensure that the sum total of the daily allowances granted to each customer does not exceed the account’s total daily capacity. If the account limit is exceeded, create headroom for volume spikes during events or peak times using the sum of identify caps.
Managing analysis queries – once it has been determined what kind of visualizations are to be built, it is a good idea to sketch out the number of PYLON analysis queries required to build each view. Combine that with a projection of how frequently customers will create those views to get a sense of how many analysis queries will be required to serve customers on an hourly basis. As part of this exercise, consider usage patterns in existing products in terms of session length and peak usage times of day.
There’s no need to start from scratch
The final point to consider when productizing Facebook topic data is that there is no need to start from scratch. DataSift maintains client libraries, which wrap the PYLON API in a variety of languages including Java, PHP, Python, Ruby and Node.js. These can save time and development effort, while DataSift also makes a variety of tools available for working with PYLON, details of which can be found on our dev site.
Productizing Facebook topic data is an important step, and by applying the techniques outlined above, you will be on the way towards building a product on top of PYLON and helping your customers find insights that improve the way they do business.
Best of luck with it and remember we are always here to answer your questions on any aspect of PYLON or Facebook Topic Data. If you haven’t seen Facebook topic data in action already, sign up for our free 14-day trial.