NB: This is a guest perspective from Vivian Braun, Distribution Sector Lead for Business Analytics, EMEA, for IBM.
Organisations’ Big Data journey: it's the analysis, silly
Travel companies rarely complain that they don’t have enough data – but what they often lack is the ability to easily turn this data into valuable insight.
Having gained benefits from descriptive analytics - such as knowledge from Business Intelligence which can aggregate and report on historical data such as demand - marketers and others increasingly want more detailed and useful analysis. Desired outcomes include having the data tell them what is likely to happen next and how to influence outcomes in the desired direction.
Big Data Advanced Analytics helps organisations understand in great detail the needs and preferences of their customers, allowing them to address each customer with relevant, personalized offerings.
Predictive analytics provides analyses and models used to help predict possible outcomes of marketing initiatives and market trends (e.g., if/then scenarios, hypothesis testing, simulations, etc.).
Prescriptive analytics provides sophisticated models used to recommend next steps or actions based on analysis across complex criteria and data (e.g., prescribing specific marketing or customer engagement strategies based on trade-offs).
Below, we will look at how travel providers – and their customers – can benefit from useful analysis of available data streams.
Did the travel industry lose the personal touch?
In the ‘olden days’, products and services were primarily sold face-to-face. Store staff got to know their regular customers – their preferences, their budget and how to entice them to spend that little bit extra. Sellers would use gut feel and experience to tailor their proposition and address each individual accordingly.
Over the course of the 20th century, self-service took over … mass marketing became a key driver of sales. The tactics were determined by “what do I have to sell, and what – roughly speaking – is my target market”.
In the travel industry, face-to-face specialist providers continued to exist, but large travel providers began to dominate with a combination of bulk catalogue marketing and telephone order taking, and later, advertising and selling via their websites.
Many have focused on cutting sales costs by encouraging customers to purchase travel on distributor websites, but have thereby given away control over that all important ‘first impression’ touch point.
What happened to the personalised, targeted proposition?? And does its absence perhaps correlate directly with the margin pressures that many in the industry are facing?
In recent years, new travel industry business models have expanded rapidly, with online travel agents and travel-enabled search engines making their mark on the landscape. They have huge amounts and a wide variety of customer data flowing through their systems.
Some clearly aim to capitalise on this, using analytics to understand travel customers’ preferences and adapt their offering, thereby placing themselves in a position to shift business and margins away from traditional travel providers.
Travel providers are data rich
Contrary to many industries, players in the travel industry possess a vast amount of data about their customers. You can’t travel by air without showing ID, any international travel requires it, many hotels ask for it, car rental firms, etc.
This means that the organisation knows the names, addresses of the purchaser, and often the full family demographics or corporate association. Online browsing and bookings generate a host of information on what pages the customer lingers on, how they work their way through the site, at what point they dump a budding ‘basket’, and what combination of offerings makes them most quickly press the “purchase now” button.
Much the same data is available through call centre or brick & mortar contact records. Simple additional questions – where did you hear about us – help define the shopping journey. Often, there will also be prompted or unprompted customer feedback.
In other words, organisations know the who, how, what and sometimes why of the engagements, and can track this over the life time relationship with that individual, given the right system. There is actually no excuse for treating customers like big heterogeneous groups, because the technology exists to aggregate and parse in a continuous loop.
Making the data speak
What is required to provide a personalised travel experience (from pre to post travel) is the ability to capture and analyse the available data, and then act upon the insight.
With Big Data Advanced Analytics, specifically predictive modelling, organisations can find patterns and trends in the data, and affinities between e.g. different combinations of travel components, prices and customers.
Furthermore, predictive modelling can indicate the likelihood of an individual accepting offering A vs offering B.
This is aided not only by the analysis of the interaction history of one certain individual – of course, not all customers have prior history with a given provider – but also the aggregation of interactions with other individuals that display the same characteristics. This can be used to draw strong assumptions about an individual, placing them in a ‘micro segment’.
Case study: A universally known rental car company uses predictive analytics (IBM SPSS) to segment customers, which reveals where to focus marketing spend.
More accurate targeting lead to a comparative reduction in e-mail marketing costs by 42 percent. At the same time, the improved insight into customer activity drives loyalty by enabling timely, relevant and personalised communications.Armed with such insight, the travel provider can predict with a greater degree of certainty the actions / offers to which individuals will most likely respond.
For example, a hotel customer may be price focused, but willing to pay a premium for specific services or luxuries in the hotel room or along the journey.
Rather than offering this individual a typical low budget option with which they won’t be satisfied, or a high budget option which they cannot afford, the provider can take out the ‘unwanted’ cost elements (e.g. the silk sheets, the airport pick-up) and add a free bottle of wine in the hotel room, if the analysis shows this to be the winning combination for this individual. This personalised approach helps keeps the costs down and increases the revenue and customer satisfaction.
There’s more to personalization than content
Another important driver of success is knowing when and how to target a prospective customer with a certain offering.
For example, the customer may make their vacation plans over the weekend, and therefore save only emails received towards the end of the week. Once the travel has been booked, the person may respond favourably to little add-on suggestions sent by text message while they’re at work – or a phone call in the evening with a special, supplementary deal.
The provider would be able to define the details of these tactics through the modelling of countless, similar shopping experiences, with characteristics similar to this individual’s, using triangulation of models to get a better understanding of propensity related to intersection of all elements. Additional data points that can be thrown into the mix include weather forecasts, advertising by local tourist boards, popular TV shows driving a demand for certain destinations or activities….
Case study: A European capital’s Tourism Board uses IBM’s predictive analysis tools to understand how, when and why tourists visit attractions across the city.
It uses the insights to influence tourist behaviour through targeted marketing promotions and to design a more accessible, enjoyable tourist experience. By examining data from the city's visitor pass—which grants access to transportation and tourist attractions—the board uncovered eye-opening temporal and spatial patterns in how people explore the city.
Certain museums see their heaviest traffic in the morning, so the board runs time-sensitive offers to distribute crowds more evenly. Some tourist sites are visited only in combination with other area attractions, so it makes sense to promote them as a group.
As a result, sales of the city card were increased by 26 percent by running promotional campaigns that cater to the specific travel styles of different demographics. The solution also supported a 14 percent boost in hotel bookings and a 7 percent increase in international airport arrivals by focusing on travellers most likely to visit.
External as well as internal data can be modelled and enhance the accuracy of the predictions. Obviously, a very potent source of what’s hot and what isn’t, is social media. User generated content is providing an endless source of insight into drivers of behaviour, segment characteristics and emerging trends.
Modelling all this data together can give definitions of segments, affinities (between offering components, and offerings/customers) and contact preferences, and predict propensity to purchase.
Think omni-channel, multi-media, anytime
In this way, a travel provider can deliver personalised, continuous and relevant interactions, based on multi-mode /interaction history across any available touchpoint or channel. The customers are unlikely to make their research and purchases in a linear, single-media manner.
To take advantage of an ‘always on’ shopping mindset, distributors should hone their micro-segmentation schemes and target consumers with the highest propensity to accept a given offer by understanding all customers’ interactions across time horizons (historic, real-time and projected future), analysing all forms of data to infer behaviour, and delivering to individuals the optimal action/offer, in their desired format, at the right time.
The smart provider will consider their ‘inventory’ of travel options when defining what to promote to whom; if resources are scarce, advertise only to those individuals who have a propensity to pay a premium for them, if there’s surplus inventory, advertise more broadly and price according to your price sensitivity insights.
Case study: A travel services company operating in Spain, Portugal and Central America uses IBM SPSS software to gain insight into customer buying behaviour both on its website and at its travel agencies.
It uses this information to better target its products to customer needs. For example, the solution discovers and analyses customer buying patterns such as purchase types, timing and customer age.
The solution can also identify correlations among the physical location of its travel agencies, the type of agency and the type of purchases made. The travel company then uses these patterns and correlations to improve its targeted marketing campaigns and develop new methods for marketing and sales. Agency productivity went up by 10-15 percent, which resulted in higher profitability rates, while increasing customer loyalty.
Another thing to keep in mind is that each member of the travel value chain does not hold all data on all customers. Collaboration – insight sharing between partners – can lead to a better combined offering, as defined by the customer, who would then be more willing to pay a premium, and come back again for more. This is particularly necessitated by the customer’s jumping between channels and media. Clearly, data protection legislation would have to be respected, but even aggregate and/or anonymised data can provide useful, actionable insights.
Turning insight into action
Generating this insight is pointless unless deployed into the organisation’s processes and decision making. This can be achieved in different ways.
As mentioned previously, it is important to not only analyse, but also execute across channels (web, travel shop, phone…) and indeed, at the point of delivery, be it a hotel, theme park, airport, in-flight etc.
First and foremost, establish what decisions made by departments and individuals across the business could be improved if they had a given customer insight. Don’t start with ‘what data do we have, how can we use it’.
Timely insight generated by predictive modelling of data from any source and tailored to the decision maker’s responsibilities can be fed directly to their device of choice. This enables them to combine modelled insight with their industry experience and make strategic and tactical decisions about the design of offerings, at a segment level rather than personalised (if handling large numbers of customers).
By integrating the Big Data Advanced Analytics findings into marketing campaign tools, the customers will benefit from the continuously refreshed insight into what the ‘next best action’ is.
Only the costs of the delivery mechanics (e.g. if using print catalogues, there may be a limit to how many versions can cost efficiently be produced) set the limit for the amount of tailoring that can take place.
Today, the more likely media is online, with mobile growing most rapidly. The digital campaign management tool can deliver the offering, the price and the visual interface based on the way the visitor browses the site, the click-stream and pace, what is selected… each visitor can be fed with the proposition that the analytics has predicted that they are most likely to accept.
This can also apply to e.g. call centre scripts, so that the service agent leads the caller down the most advantageous route, depending on how the conversation evolves.
Case study: A leading Asian airline has gained insight into travel and purchasing trends and patterns underlying the behaviour of millions of website visitors and more than 25 million loyalty members who accumulate points via purchases from the airline and its partners.
The IBM SPSS solution mines and unifies customer information and achieves a single customer view from which patterns and trends arise, enabling the airline to dissect and analyse the data to gain a deep understanding of its customer behaviour.
For instance, they can now make correlations between users’ demographics data, such as age and gender, with purchase history and web behaviour, such as browsing preferences, rates of disengagement or conversion and frequency of visits. Additional data from surveys is incorporated into the analysis for further insights. A 1.6% year-to-year growth in sales revenue of transportation services from personalised online marketing based on predictive analysis is expected.
Refining the action with prescriptive analytics
Automation doesn’t mean that the ‘old fashioned’ gut feel, experience and common business sense have to be ignored.
Business rules such as “don’t target the same person more than once per week”, “don’t advertise free alcohol to under-age customers – even if they are predicted to accept the offer!”, or “don’t sent differing price offers to multiple members of the same household” can be applied to the execution tools.
As mentioned above, the vendor would also balance the amount of offers and the recipients according to availability – or determine whether securing additional availability will increase profitability – and whether the current objective is to entice new users, introduce a new offering, optimise short term revenue, grow satisfaction levels etc.
With prescriptive analytics tools, such factors are taken into consideration, i.e. the rules, trade-off preferences and the common sense are built into the solution. Alternative actions or decisions given a complex set of requirements, objectives and constraints – as defined by the business – determine the best solution or outcome among various choices.
Case study: A major German tour operator created a hotel pricing system based on IBM Business Analytics and Optimisation technologies, ensuring prices are precisely tailored to customer behaviour.
The software automatically structures and clusters historical booking data, and all pricing-relevant information is made continually available to the pricing specialists via an intuitive user interface. They define the desired margin for a particular destination region and the solution then automatically calculates all the possible combinations and dependencies until the optimum result is achieved.
The solution automatically forecasts which group of customers will drive demand for particular accommodation services at each point of the season to be priced. An entire destination region of 50-100 hotels can be priced in less than 20% of the time required by manual pricing methods. Pricing consistency (i.e. ensuring a four-star hotel is always less expensive than a five-star) is now 100% guaranteed.
Summary
Customers today value personalised interactions and offerings just as much as they always did. In fact, they have become more demanding and can pick and choose in this competitive market, using consumer technology to gain visibility and interact like never before. Organisations can provide a personalised experience by using systems and processes to:
Understand, measure and analyse customer behaviour across all interactions and all touch points.
Develop micro segments.
Infer buying behaviour and identify best action(s) for segments and individual customers.
Deliver personalised offers and recommendations in real-time to drive basket size, satisfaction, retention, and even brand evangelism.
Incorporate company objectives, rule, targets and constraints into the offerings.
Organisations are recognising that information is a competitive advantage. By capturing a variety of data from inside and outside of the organisation, and applying Big Data Advanced Analytics, travel providers can better understand consumers and their behavioural drivers. Advanced – predictive and prescriptive – analytics can be used to develop targeted marketing, tailored offering and personalised shopping and travel experiences for the target customers.
NB1: This is a guest perspective from Vivian Braun, Distribution Sector Lead for Business Analytics, EMEA.
NB2: Computer analysis image courtesy of
Shutterstock.