By Michael Schäfers, Senior Data Engineer, TUI
In the travel and tourism industry, knowing your customer is a big advantage to creating a customer experience that surprises and delights people. In fact, 65 percent of consumers would become long-term customers of a brand that provides a positive experience throughout an entire customer journey, from initial interest, through purchase, and to customer service. But often, the data used to book travel, hotels, and attractions becomes siloed and disparate as individuals create more records of themselves using apps and third-party booking sites. Personal travel and business travel can also blur the full picture of a customer and block companies from understanding who they are marketing to.
Customer retention and revisits are a top priority for those in the travel and hospitality industry and data is a big part of driving those efforts. At TUI, we manage data for over 20 million customers in a modern data environment complete with a centralized data warehouse and cloud-native solutions, including Matillion ETL. Our mission is to create unforgettable moments for customers across the world and make their dreams come true. That means we need to ensure that the data for our customers is accurate so we can deliver unforgettable experiences.
A cloud data analytics platform comprising a data warehouse and ELT solution can get your data into an analytics-ready state to yield critical customer experience insights, faster.
ELT involves the same activities as traditional ETL, but in this case, the ‘E’ and ‘L’ portions of ELT are done in one move straight to the target data platform. ‘Extract’ works the same with ELT or ETL. ‘Load’ is the same, too, except in ELT you load data into your target before you transform it. But the ‘Transform’ activity is very different. Instead of transforming your data in your ETL engine/server, you use the power of your cloud data platform to process the raw data you extracted and loaded.
3 steps to improve customer experience in hospitality with ELT
Step 1: Create a deduplication algorithm in your data model
Our Analytics team at TUI has tackled the challenge of customer data deduplication with a tailored solution called THE DUDE (TUI Hotels Engine for DUplicate DEtection). THE DUDE helps us to find multiple representations of the same real-world customer. Duplicates may exist in a distributed IT infrastructure of a company operating worldwide and with a variety of booking channels.
Our goal is to identify these duplicates, then match and merge customer records even if there are huge variations in the records caused by spelling mistakes (e.g., “Jon Miler” instead of “John Miller”), a change of address, changes in emails or phone numbers, and even name changes. To achieve all this, we need to do the following five things:
Data Cleaning: Input data cleaning and harmonization/standardization
Strip special characters, ignore case, establish a common format for address data, and ensure data type homogeny
Candidate generation: We can’t compare every record with all other records
Establish scope of deduplication efforts. Identify pairs of customers that could possibly match
Similarity calculation: Apply measures to estimate the match quality
Map the output of string and date comparison to a numeric value. Grade or rank the commonality between fields.
Rule-based matching: Evaluate if a match candidate is classified as a matching
Threshold-based rules classify similarity values for different groups of attributes
Merging: Build a “Golden Record” (or System of Record) for all identified matches
Fill missing values, update attributes to the newest version
Step 2: Use ELT to transform data
Data deduplication is a highly complex task: On one hand, deduplication algorithms incorporate a lot of business and domain logic, such as similarity measures and matching rules. One will instantly get lost if trying to solve this complexity in a SQL statement. Additionally, a lot of computational power is required to evaluate and solve our complex matching calculations.
Matillion ETL’s clean and intuitive UI helps us to simplify this complexity with its broad set of transformation components. We can break down the huge problem visually into smaller tasks and solve each task with understandable and reusable bits of logic. The low-code UI also helps us manage the process accurately and holistically, as everything is commented with Matillion’s note feature and the process is visually arranged to make the comprehension of the individual steps much easier.
Next, the power of Matillion ETL and the scalability of the underlying cloud data warehouse helps us tame run times for the huge set customer data – with the capability to process even more happy guests in the future.
The following table compares our new ELT-powered approach with the previous solution, which was provided by an external supplier on legacy on-premises hardware.
|Previous Solution||THE DUDE|
|Knowledge/Process Control||Black box (external supplier)||Internal team, standard SQL, can adapt to business requirements|
|Infrastructure Requirements||On premises DB, ETL tool,
(existing tool stack with Matillion and CDW)
|Run Time||Overnight run||10 Minutes|
|Monthly Operational Costs||~ 1500 €||15 €|
As you can clearly see, Matillion ETL and our cloud data warehouse outperform the legacy approach in all aspects. In addition to huge advantages in understanding and maintaining the process, THE DUDE demonstrates impressive results in terms of costs and speed.
Step 3: Populate marketing, sales, and support platforms with accurate insights
With THE DUDE and its ability to match and merge customer duplicates, we can now see a complete picture of our customers for the first time. Valuable and timely data on the full customer lifecycle is crucial for all CRM-related tasks. We can now consider all previous bookings by our customers and determine if they prefer relaxed holidays on a sunny beach or if they strive for action at a thrilling ski resort.
We recently launched a successful project with our marketing and sales teams called the “Next Best Offer” (NBO). Combined with THE DUDE, our data scientists leverage algorithms to recommend tailored offers to our customers based on their previous holidays with TUI. In short, it is all about addressing the right offer to the right customer at the right time. We can already see the promising results of increasing customer retention rate.
Better data records mean better experiences for travelers
The more we know our customers, the better we can serve and anticipate their needs. We put our travelers first and want to make sure every holiday can be enjoyed to the fullest. To learn more about TUI Group, visit our website.
If you’re ready to learn how Matillion can help improve the customer experience at your business, sign up to attend a live weekly demo.
The post How Matillion ETL can help improve customer experience for travel and hospitality appeared first on Matillion.