In over a decade of FP&A forecasting for online travel marketplaces (OTAs), I’ve learned a lot about how they operate, specifically their predictability, optimisation, and goal setting. This article covers some key forecasting learnings from the perspective of an FP&A professional and how to potentially apply them to other online marketplaces.

The good news is that OTAs involve booking a service or experience, and these learnings can be extended to different industries, such as on-demand transportation, events, and food delivery.

What Is an OTA?

An OTA is a broker that matches consumers to suppliers for travel purposes – booking hotel stays, reserving car rentals, acquiring bus/train/airline tickets, or reserving destination experiences. The metrics and terms are relatively consistent across different OTAs regardless of booking type.

Learning №1: Forecasts always begin with the top-line data: transactions

Starting with transactions is crucial, as financial indicators lag while operational metrics lead. Revenue recognition policies may impact the timing of revenue recognition and supplier payments but may not correlate with booking timing. If the OTA is the merchant of record, revenue can be recognised either when the transaction occurs or when the consumer uses the service, which may be in the future. This discrepancy can mislead the FP&A forecaster unless the booking and consumption lags are consistent.

Revenue is the culmination of a funnel of factors starting with the transaction, the source of truth in financial ledgers. Note that site visitors (traffic or Monthly Active Users (MAU)) can be gamed by marketing channels, creating conversion variances and unreliability. The typical funnel includes transactions and key business drivers that multiply to get revenue from the transaction. These drivers vary across markets and seasons. For example, consider the funnel for hotel bookings:

Transactions * Average Length of Stay * Average Daily Rate * Commission Margin = Revenue

Seasons affect the traveller’s length of stay and prices, making them important drivers. For transportation services, the length of stay can be replaced by tickets/transactions and ADR by the average ticket price (ATP). And finally, ancillary services like insurance can be added via an attach rate, guiding the FP&A forecaster on purchases per transaction.

Conversely, a bottom-up EBITDA view without a top-line goal can lead to conservatism not aligning with the business direction, as OTAs are generally driven by the top line. This misalignment can cause disagreements and strain among top-level leadership. 

Learning №2: Segmentation is critical to an accurate forecast

After setting top-line targets, projected volumes and rates are shared with other FP&A teams (or partners that work with FP&A) but should be segmented by product and business function, which usually differ. For example, in online marketing, product expectations are broken down by channel, while supply managers parse products by different supplier types. The complexity of segmentation can increase with the number of channels, suppliers, and other factors.

The additional capability of international travel adds complexity. Domestic users travelling on domestic supply simplify operations, but internationalisation creates challenges with multiple currencies, purchasing power, and market dynamics. For example, managing multiple geographies can make creating a geography-based P&L problematic. In the USA, if all travel is local, all purchases are in US dollars, and seasonality is consistent. However, if there is additional supply in Mexico, then average prices can decline due to purchasing power, fees can change for payment processing, currency conversion can shift margins, and variable marketing costs can arise from regional differences.

Building forecasts for smaller or niche OTAs can be difficult without a history of performance or an expansive dataset. Limiting segmentation and using broad, simple forecasts will help these OTAs. As they grow and gather more data, they can scale the complexity of their forecasts.

Learning №3: Travel is an extremely seasonal industry

Worldwide top-line patterns from consumer behaviour in travel are surprisingly consistent. For example:

• In Mexico, one of the busiest booking and travel times is during Easter, mainly domestically.
• In Canada, consumers book all-inclusive Caribbean vacations in January due to winter.
• In Brazil, international tourists purchase bus tickets for Caribana in February.

Reviewing year-over-year trends allows for comparisons of weekly and monthly data to observe patterns and variances from marketing campaigns, special events (World Cup, Olympics, concerts), and supply changes. Unfortunately, these variances can lead to price swings, impacting margins. Similarly, one-time shock events, such as the pandemic or a military coup, need to be factored out from the seasonality. Note that automated data science techniques are not always sophisticated enough to take these variances into account and must have manual oversight to partially override history. Scenario Planning for different recovery levels enhances forecast resilience and identifies investment opportunities based on different growth trajectories.

Holiday season trends challenge the FP&A forecaster. Holidays like Easter shift across Q1 and Q2, while US Thanksgiving is the third Thursday in November. Labour Day is the first Monday in September in North America, but it is on May 1^st in many other countries.

As the calendar shifts from December to January, validating annual planning trends for year-over-year growth is essential. Most countries have holidays around this period, making it important to corroborate trends with daily and weekly forecasts. Management will seek to understand January's performance within the first week of the month.

Learning №4: Proper Investment in Planning Tools & Processes is Crucial

Segmentation by geography, currency, supply, product, and marketing channel requires robust planning systems to integrate data intersections early in a company’s history. In five late-stage travel companies I worked for, each struggled to project growth and profitability because of poor (or no) planning systems.

Complexity can easily overwhelm Excel and FP&A forecasters, causing files to fail due to size limitations. This complexity is compounded by multiple versions of the truth, which can occur without proper data governance or data integrity from multiple sources (i.e., is this the right number?).  Because OTAs are real-time businesses, regular audits and validation processes are crucial to align data quality across accounting, finance and technology organisations.

Over-investing early is always better than having systems fail from data overload later, but it may be a challenge for a smaller OTA that is just starting and does not have the resources to initially commit. However, if there is an early signal that the OTA will be successful, investment in tools, especially in FP&A, must become a top priority as soon as possible.

What Are the Key Takeaways for the FP&A Professionals?

Although OTAs are data-intensive and complex, they are predictably manageable with the right investments. By applying the insights from these forecast learnings, FP&A professionals in these and other online marketplaces can obtain the following benefits:

1. Enhanced predictability and strategic alignment: Accurate financial forecasts support the operating and Strategic Planning cycles, ensuring all departments work towards common financial and business goals. This minimises conflict and fosters cohesive growth.
2. Better decision-making through detailed segmentation: Segmentation by product, geography, and business function enables actionable insights. This detailed approach improves resource allocation through targeted marketing and manages complexities in international operations.
3. Robust planning tools are critical for long-term success: Investing in advanced planning tools and processes allows the FP&A team to handle large data volumes, perform real-time analyses, and maintain data integrity. Doing so early enough prevents data overload, ensuring the FP&A team and OTA remain agile and responsive, supporting sustained profitability and growth.