When you buy an airplane ticket, what comes with it? These days, it’s pretty much just a ticket. However, most airlines give you the option to purchase additional items, such as an upgraded seat, Wi-fi, beverages and more. These products and services are all in addition to your original purchase and are known as ancillaries.
While ancillaries were originally seen as “hidden fees,” industry leaders have fought to change that perception. Now, products that are presented at or before checkout provide additional value to customers through fresh and innovative product offerings. Given the increasingly hyper-transparent economy, the need to evolve is critical for companies to drive organic revenue growth.
Besides airlines, other industries that currently leverage ancillary techniques include automotive, hospitality, cruise and car rental companies. Regardless of the industry, customers who do not regularly purchase ancillary products fall into the following buckets:
- Customers who were offered the product and chose not to purchase it.
- Customers who were not offered the product at all or were eligible to receive it for free but were not aware of the product.
Both instances point out to companies not offering customers an end-to-end personalized travel experience through exclusive and tailored offers at personalized price points. However, how can companies target each of these customer groups and drive organic ancillary revenue growth at the same time?
For starters, leveraging a Classification and Regression Tree (CART) model allows companies to segment customers by identifying significant attributes that drive purchase behavior, and determining statistically significant break points for each attribute.
This method involves several key design considerations, including:
- Pre-segmentation of customers based on business considerations – this helps to avoid ambiguity and aligns the set of attributes with pre-existing conditions that determine eligibility of certain purchases, such as members of loyalty programs, credit card status, etc.;
- The choice of target variable – this helps identify groups of customers that exhibit similar purchase behavior for a particular combination of ancillary products (for example there are 16 possible values, including No-Purchase for a set of 4 ancillary products);
- Set of Attributes – encompasses a range of customer attributes such as the number of loyalty points, credit card tier or history of previous ancillary purchase behavior and trip context attributes such as season, party size, business versus leisure, or proximity of departure.
Another useful method for behavior analysis is through a Market Basket Analysis (MBA), which helps to identify sets of association rules across market basket items, i.e., the set of ancillary products. Market Basket Analysis answers questions such as, “How often are certain items purchased?” and “How are items purchased together in a basket?” This gives companies sets of association rules they can use to make business decisions and recommend combinations that will entice customers to purchase additional products based on predicted anchor purchases. Key design considerations for a MBA include:
- Identify key filtering criteria for association rules – this is driven by a foundational analysis across the following key MBA metrics: Support (e.g. percentage of rentals who booked a Mid-Size Car AND bought Fuel AND Upgraded), Confidence (e.g., of the rentals who booked a Mid-Size, AND bought Fuel, the percent who also Upgraded), and Lift (e.g., rentals who booked a Mid-Size and bought Fuel are more likely to Upgrade compared to the rest of rentals);
- Evaluate trade-offs for “edge case” association rules – this is relevant in the case of excluding association rules with low support metrics that are being bought in low volumes since it hinders the ability to drive future promotions on these sets of ancillary products.
Finally, it is important for companies to run a Market Response Model (MRM) to measure customer price sensitivity when it comes to ancillaries. The key design considerations for a MRM are:
- MRM Normalization – which helps to isolate the effect of ancillaries price on demand and is essential to ensure price elasticity estimates are not biased by outside factors. Users must control for factors, such as seasonality and trends, that impact both demand and price;
- MRM functional form – provides preference towards selling certain products based on profit margins, favoring a willingness-to-pay model (particularly for service offerings in automotive);
- Price – Demand Observation – this accounts for both the aggregation and imputation of transactional data. Aggregation controls the need for imputation and in general provides more stable estimates of elasticity. Imputation is associated with a lack of purchase and requires a fill-in algorithm based on similarities between products offerings.
One Fortune 500 automotive retailer that implemented the above practical considerations saw a substantial financial increase. Prior to implementation, the company viewed new vehicles as commodities since other dealers offered the same products. To enhance the customer experience, the auto retailer strategically reduced customers’ time in the dealership while selling them more Finance & Insurance products like warranties and GAP insurance.
Revenue Analytics analyzed over 10 million rows of sales data on of roughly 500,000 add-on products using CART, MBA and MRM analyses to identify “customers like you also bought this” providing customers with fewer – but more appropriate – options on F&I products to buy. The results were undeniable, with the automotive retailer reaping tens of millions of dollars in organic revenue.
Clearly, ancillaries can be a huge revenue driver for businesses. However, the only way to determine the best course of action and eliminate the unknowns is through predictive analytics. The above models allow companies to analyze millions of rows of data in a timely manner, producing actionable pathways to lift bottom-line revenue and put the most relevant offer (product and price) in front of the right customer at the right time.
Ancillaries don’t have to be hidden fees or a burden on customers. By presenting them in an attractive manner with scientific, analytics-backed decisions, companies will see significant revenue uplift.