Decide which calendar answers the question
A travel business can record a booking today for a stay six months from now. Advertising runs on an exposure calendar, bookings arrive on a purchase calendar and revenue may be recognised on a stay calendar. Joining those series without a clear model of timing can create an impressive but misleading lag.
- Choose booking or stay date
- Measure cancellations
- Align the booking window
- Respect capacity
For acquisition planning, net bookings by booking date may be useful. For capacity and revenue management, stay-date demand matters too. Keep both calendars in the data contract and explain how they connect. The MMM data checklist is the starting point for consistent periods and outcomes.
A useful model starts with a consistent outcome and a decision it can inform. The marketing mix modelling learning path connects this idea to the assumptions, practical checks and next steps.
To work through the connected decisions, read Ecommerce MMM: separate promotions, media and returns and Adstock half-life: convert decay into a time assumption.
Separate a booking delay from media persistence
Imagine an advert runs in January, a customer books in February and travels in July. The January-to-February gap relates to the booking decision. The February-to-July gap relates to the product consumed. A single adstock parameter should not be interpreted as measuring both without an explicit model.
Compare booking-window distributions across products and seasons. Short breaks and long-haul trips may have very different timing. If the mix changes during a campaign, a national aggregate lag can shift even when the underlying response within each product is stable.
Work through cancellation maturity
A fictional campaign produces 1,000 bookings. After one month, 50 have cancelled; by the travel date, 150 have cancelled. Comparing a new campaign's 950 currently active bookings with an old campaign's 850 completed bookings gives the newer campaign an unfair advantage. Use a consistent maturity rule and label forecasts of eventual net bookings.
Cancellation policies, pricing and disruption events can change that maturity pattern. Retain data-as-of snapshots so historical evaluation uses the information available then. The adstock guide covers carryover transformations, but those transformations do not solve outcome revisions.
Plan spend against available inventory
When a route, property or departure is nearly full, an unconstrained demand increase may not translate into additional sales. It might shift demand to a substitute product, change prices or displace customers who would have booked anyway. Record the objective: occupancy, net revenue, contribution or a particular departure's demand.
Test a bounded allocation against capacity, price and cancellation scenarios. Do not use a national average response to justify a large increase on a sold-out departure. A useful report connects the proposed spend to the products and travel windows where capacity is available.
Before approving a change, review calendar alignment, maturity, product mix, disruption flags and the range of feasible capacity. If the model cannot distinguish those influences, use the uncertainty to design a narrower test instead of presenting a single precise return.
Sources and further practice
This article’s worked example is illustrative, not a measured client result. Use the linked methodology to check the assumptions before applying it.