For a decade, hospitality's most defended metric was direct booking share. It was a proxy for brand strength, margin protection and independence from intermediary economics. Skift and Phocuswright have both tracked how, after the pandemic, that share rebuilt itself against OTA dominance — until roughly 2024, when the trend line began to soften again in ways that did not appear in any traditional attribution model.
The reason is not price parity or aggressive OTA marketing. It is that a new intermediary layer has emerged between the guest and every property: the AI assistant. When a guest asks an assistant to plan a trip, the assistant does not present ten options with rate transparency. It selects — often two or three — and the properties it selects tend to be the ones whose canonical presence on the web is machine-legible. In most cases today, that presence is an OTA listing.
The churn is silent because it never registers as a lost booking. There was no abandoned cart, no failed checkout, no email sequence to reactivate. The comparison the direct channel was built to win did not take place. The demand was resolved before the property's own funnel had a chance to run.
This changes what 'defending direct' means. Rate parity clauses and loyalty programmes were built to shape choices at the end of a comparison. In a single-answer world, the leverage moves upstream: the property either exists as a coherent, first-person entity that a model can quote, or it exists as a row inside someone else's aggregation.
Direct share, in this cycle, will not be defended by better commissions. It will be defended by being answerable — in the property's own voice, on the property's own surface, in the moment the AI layer decides whom to name.
"The next OTA cycle will not be won by commissions. It will be won by being the answer the model already has."