First Published on 18th December, 2017
Ai Editorial: Counting on computing power, algorithms, airline-specific IT offerings, front-end technology and customer profiling will lend a new dimension to personalisation, writes Ai’s Ritesh Gupta
Airlines are attempting ways to differentiate their offerings, and serve them to today’s always-connected travellers as per their preferences and the propensity to spend.
There are 5 factors that would contribute in taking web personalisation deeper from here on:
1. Computing power: It includes both the personal device being used by travellers including mobile devices as well as the power underlying the server infrastructure readily available in the cloud, says Kenneth Purcell, CEO, iSeatz. These elements are become cheaper with the passage of time and the trend continues. The mobile phone is a highly personal device, and offers an opportunity to capitalise on certain innate features, such as location. So the blend of device plus user profiling can result in contextual recommendations. So information such as the time of the day, weather, location etc. can be used for personalisation. A mobile device has a massive computational power, plus it has camera, location-specific features.
As for moving infrastructure into the cloud, it is time to leverage today’s technology at a much lower cost, and avail the benefit of scaling it up. Airlines consider factors such as such as security (the role of technologies such as encryption and tokenisation comes into the picture) as well as connecting legacy applications to the cloud at enterprise scale. Travel searching is heavy, and organizations are moving swiftly in this direction. Among airlines, American Airlines has decided to migrate to cloud a quota of their crucial applications, including aa.com, mobile app and network of check-in kiosks. The plan is to enable developers to swiftly set up and modify application functionalities for American’s passengers. These customer-facing systems will be on cloud. The cloud business model that the airline has chosen is a hybrid one.
2. Algorithms that run the sort order pertaining to the results that are being shown and how does that intersect with business rules, so that two of them in harmony as per the objective laid and outcome expected by an organisation are getting sophisticated in web personalisation. They can also be categorized – basic one such as showing what’s popular on the site to what’s new. Then there is collaborative filtering. Depending upon a user’s engagement with various products, say destinations chosen, bundles or ancillaries, they are clubbed into a group of users with similar likes and dislikes. Recommendations are crafted accordingly. Also, the sort of data that helps in personalising digital experience includes the source of traffic or acquisition data, anonymous visitor data, profile data as well as real-time interaction with the website. The process of serving anonymous passengers starts with some level of contextualisation – once a prospective traveller accesses a website or a mobile app, enters city-pair, dates, type of travel (family, with kids etc.), then algorithms can match them against pre-set customer segments and serve offers accordingly. At a deeper level, airlines can also focus on precise preferences, adding them for each user via deep behavioural tracking (a bunch of factors are considered - mouse movement, scrolling etc. + IP address, geo-location, device type etc. + other signals) to sharpen algorithms and make them even more relevant.
3. Sophistication of airline-specific engines: For any airline if their systems or engines get smarter over a period of time, then they are bound to come up with better recommendations or offers. And travel technology companies are looking at using data better, for instance, letting merchandising rules deliver better results. Similarly, the industry is looking at consistency in terms of what they have to offer, another engine - for shopping and pricing – would be the way to go forward, capitalising on all sorts of data – loyalty, merchandising, fare, schedule, availability etc.
4. Front-end technology: it is making rapid advancement in the industry. “This is significant in terms showing the different search results, how the entire page is rendered and paving way for segmentation all the way to user experience. So looking at the APIs, all of this needs to support sorting of the inventory, that is being outsourced, is done in a way that it is relevant for the user and the front-end is a layer on top of it. This would include using 3rd party tools or working them in-house to set up front-end in a more personalised way,” says Purcell.
E-commerce specialists point out that the efficacy of content management systems is coming to the fore when it comes to managing, personalizing, publishing, viewing and comparing different page versions. How to create create large web applications that use data which can change over time without the need to reload the full website? (Speed is an important element – a case study of how travel search engine Wego counted upon Google’s open source initiative called Accelerated Mobile Pages (AMP). In case of Wego, page-load speeds came down from more than 11 seconds to less than one second. AMP pages are stored in Google’s cache servers and load in milliseconds).
5. Preparing for customer profiling and 1to1 personalisation: “For 1to1 personalisation, factors such as “too hard to do it” or “too much storage is required” need to be done away with. The truth of the matter is there are lots of tools out there that make it easier to do it (whether its analytics provider tailoring the user experience on an individual basis or an organisation decides to develop the infrastructure in-house, open source frameworks pave way for the same, and even storage isn’t an issue today),” Gillian Morris, CEO, Hitlist told Ai in a recent interview.
It is imperative to bank on 1st party data. “(Data strategy) It’s not about how much data you have (and big data is inherently a vague term - how big is big?), but rather the quality of the data you’ re using. Travel companies that focus on loose intent signals from many different providers are acting on weak cues that might be misleading. The ideal situation would be to generate enough data within your own user ecosystem to truly understand where and why people are planning to travel. Google, Facebook, and theoretically Apple have the biggest leg up here,” asserted Morris.
As for the journey of personalisation, as explained in this article, start with segmentation and make steady progress to rules-driven personalisation. This means setting up and further reworking on business rules that are utilised against clusters of visitors, based on information one can garner about users. The second major component is progressing toward algorithmic personalisation, where one initiates with a relatively broad set of recommendations to ones that are specifically meant for an individual.
Interplay of all 5 aspects
Eventually, the interplay of all these 5 aspects – computing power, algorithms, personalisation and front-end technology come into play to deliver a relevant, contextual, personalised experience. For instance, the benefits of knowing a customer – not only steps up the conversion rate, but it also means less time spent on browsing, taking a decision faster on an airline’ site, and this would also cut down on the server cost.
Follow Ai on Twitter: @Ai_Connects_Us