Ai Editorial: Handling each data source with dexterity

First Published, 1st March 2016

Ai Editorial: Having data feeds from multiple sources is paramount for airlines in their quest to understand what their customers will purchase next, writes Ai’s Ritesh Gupta


The complexity associated with tracking a user’s every digital activity is on the rise. Be it for a member of FFP, or someone who has booked with an airline or even a visitor to a desktop/ mobile site or mobile app, if a travel e-commerce brand isn’t blending fragments of customer data across various channels aptly then a competitor may act and garner the next booking.

So how many data points would an organization need to handle today?

Mark Ross-Smith, Chief Product Officer / Head of Loyalty, StayCorp, says there could be “in excess of 500 key data points” and an astute segmented program will have “over 1,000 unique, individual data points on each member”! Acknowledging the significance of data collection, he told me having data feeds from multiple sources is paramount for airlines in their quest to understand what their customers will purchase next. He added, “Data feeds come in all shapes and sizes, and the best sources originate from unlikely partners who have key metrics that might otherwise be unobtainable elsewhere.”

Overall, we’re seeing that the path to conversion in travel is not linear, and requires a specialised understanding of how to activate data at scale.  

Consolidating data sources

As for the difference between first and third party data, first party data comes from your own internal sources and is used by you. It can take the form of email addresses, offline or online purchase history, subscription and social data, loyalty data, interactions with your mobile app, etc. Third-party data, on the other hand, is someone else’s first-party data that you’re able to use for your marketing efforts through a direct relationship with the data source, indirect relationship with a data exchange, or that you pay for.

As an advertiser, there’s so much data available and people should use it all, but it needs to be used intelligently. For example, first-party data is great because it gives you consumers that already have an affinity for your brand. However, it’s limited and restricted, perhaps to only those who’ve visited your website; certainly you want and need to reach a larger audience. That’s where third-party data comes in: third-party data is great because you reach a huge audience, but they haven’t been to your website so they’re less likely to have an affinity for you. If you utilize only one of these data sets, you’ll always be missing key consumers. Rather, it’s the combination of first- and third-party data that can be very powerful if managed properly.

Right approach

At the end of the day, consolidating all data sources, learning as much as you can about your data assets/audiences, and deploying your data across all applicable channels is the absolute right approach. It’s not easy, nor it is accomplished overnight.

Also, data sources will include your hotel partners (especially independent hotels), telcos and niche social networking platforms but the flavor of the month is start-ups. Technology start-ups have some of the smartest folks driving innovation and disruption and through this they’re forced to find new revenue streams that break the traditional mould. In some cases -  the data retained by start-up companies can be worth more in the hands of an airline & FFP when rolled out over a larger member base. 

Unstructured data sets

Utilizing unstructured data sets can be a daunting task – but instead of burning through resources on data science teams trying to find the ‘one in a million’ breakthrough model – it can make sense to simply the objectives. By focusing on commercializing the data immediately with no fuss it will lead to greater revenue generation opportunities without the hassle of data scientists analytics and insights professionals. Treat each new data source individually and focus on formatting and structuring it so there are constant updates and that it remains accurate.  By then focusing on commercializing individual data points one at a time – FFPs can build out their marketing platform in baby steps.

Being in control

On a concluding note, organizations need to be in control of what data they need to gather. There is no point in accumulating inappropriate details about customers. Of course, accurate and reliable data is a vital component of working out correct customer profiles. So is data-quality technology. Also, one needs to do away with siloed departments and out-dated data. There is a need to create APIs to bring all your data into a central place. And pave way for a mechanism for data sources to “talk” to each other. In terms of sophistication, today there are search-based interfaces available to assess guest repository, and also offer real-time recognition say in a contact centre.

Still one can’t expect magic overnight. Talking of the final output, specialists do point out that combining probabilistic data with CRM and transactional (deterministic) data in way that you can use the data safely and effectively in different contexts isn’t easy yet. As things stand today, achieving a perfect single customer view may not be possible.

But at least by focusing on the right sources of data, one can lay base for a strong foundation. 

Follow Ai on Twitter: @Ai_Connects_Us