The true cost of OTA dependency is not the commission line

Ask a general manager what their online travel agency relationship costs, and the answer will almost always be a commission percentage. It is a reasonable answer. It is also the smallest component of the number they were asked for.

Across the distribution diagnostics we have run over the past three years, commission has accounted for somewhere between 45% and 60% of the total economic cost of intermediated distribution. The remainder is real, recurring and material — but it never appears on an invoice, which is why it is almost never managed. This article sets out the four unbilled costs, how to quantify each of them, and what a defensible channel mix target actually looks like.

Cost one: rate suppression

The most expensive consequence of high OTA share is not what you pay per booking. It is what you are able to charge in the first place. A property that derives the majority of its demand from price-sorted marketplace results is competing in an environment engineered to make price the primary variable, and it will price accordingly.

This shows up in the data as a persistent gap between a property's rate position in OTA-sourced bookings and its rate position in direct bookings for equivalent dates and room types. In our engagement sample the direct-channel ADR premium ranged from 6% to 19%, with a median of 11.4%. That premium is not a pricing trick — the same room, on the same date, sells for more when the guest arrives through a channel where the property controls the presentation, the inclusions and the comparison set.

Direct-channel ADR premium on matched date and room-type pairs, five-property engagement sample

At 80% OTA share, an 11% rate suppression applied to four fifths of room nights is worth roughly 8.8% of total room revenue. That is a cost of comparable magnitude to the commission itself, and it is entirely invisible in the management accounts.

Cost two: guest data forfeiture

An intermediated booking does not transfer the guest relationship to the property. It transfers a room night. The email address is masked or absent, the booking history sits in someone else's system, and the property has no consented channel through which to reach that guest again.

The cost of this is best expressed as forgone repeat revenue. In properties with a functioning guest data asset and an active lifecycle programme, we typically observe repeat guest ratios between 18% and 27%. In properties with high OTA dependency and no data capture, the figure sits between 3% and 6%. The gap is not explained by product quality or guest satisfaction — several of the low-repeat properties in our sample carried superior review scores. It is explained by the absence of any mechanism to invite a past guest back.

Cost three: loyalty transfer

Marketplace loyalty programmes — Genius, Rewards, and their equivalents — are frequently enrolled into once and then never re-examined. The mechanics are worth stating plainly: the property funds a discount, the marketplace owns the loyalty relationship that discount creates, and the guest becomes more likely to return to the marketplace rather than to the property.

The margin arithmetic is straightforward and unattractive. A 20% commission with a 10% loyalty discount applied on top does not cost 30% — it costs 28% of the gross rate, because the commission is calculated on the discounted amount. That is still a rate of distribution cost that very few operators would approve if it were presented to them as a single figure rather than as two separate programme enrolments.

The more damaging effect is behavioural. Loyalty-programme members are, by construction, being trained to begin their next search on the marketplace. The property is paying to increase the probability that its future demand arrives intermediated.

Cost four: demand substitution

This is the cost operators find hardest to accept, because it requires acknowledging that a substantial proportion of intermediated bookings were never incremental. In path analysis across two engagements, 31% and 27% respectively of OTA reservations were preceded within 48 hours by a branded search for the property by name, or by a direct visit to the property website.

These are guests who had already selected the property. The marketplace did not generate that demand; it intercepted it, usually by outbidding the property on its own brand terms, and then charged a commission for the introduction. Every one of those bookings represents commission paid on demand the property generated itself.

Assembling the true cost

Combining the four components against reported commission produces a materially different picture. The table below models a property at 80% OTA share with a 20% blended commission — a common profile in the independent segment.

True distribution cost model, 80% OTA share, 20% blended commission, expressed as % of total room revenue

A property reporting 16% distribution cost is carrying something closer to 32%. This is not an argument that marketplaces destroy value — they do not, and for genuinely incremental demand into unfamiliar markets they remain the most efficient acquisition channel available. It is an argument that the cost is being understated by roughly a factor of two, and that decisions made on the understated figure will systematically over-allocate to the intermediated channel.

What a defensible target looks like

The correct objective is not zero OTA share. Properties that pursue it usually discover that the marginal cost of replacing genuinely incremental marketplace demand with paid acquisition exceeds the commission they were avoiding. The objective is a channel mix in which the marginal cost of each channel is known, and the allocation to each is a deliberate decision rather than an accumulated one.

In practice, the properties we have worked with settle between 40% and 55% OTA share, with the marketplace retained for genuine reach into unfamiliar feeder markets and need-date fill, and the direct channel carrying repeat guests, branded demand and contracted business. What changes is not that the marketplace disappears. It is that it stops being the default recipient of demand the property could have served itself.