Dark Patterns in OTA Extranets

The term dark patterns was coined by the UX researcher Harry Brignull in 2010 to describe interface designs that steer people into choices they would not otherwise make: a pre-ticked box, a cancellation buried four screens deep, a comparison that hides the cheaper option. The literature that grew around it — Brignull's catalogue at deceptive.design, the US Federal Trade Commission's 2022 staff report Bringing Dark Patterns to Light, and Article 25 of the EU Digital Services Act — is written almost entirely about consumers.

Hotel operators are on the other side of the counter, but they are users too. A night auditor closing the day, a revenue manager pushing a rate change to 40 date cells, an owner reconciling a monthly commission statement: each is working through an interface that the platform designed and the property did not choose. This article applies the consumer literature's method to those interfaces. It describes patterns as observed design outcomes and prices what they cost, without claiming to know what any design team intended. The one line of attribution it borrows is the site's own, from the home page: “The interface isn't broken. It's working exactly as designed.”

Two boundaries. First, the patterns below are generalized from operator experience across the major extranets and channel-manager consoles; they are not a finding about a single named portal. Second, every dollar figure is illustrative and carries its assumptions in the sentence. The point is to give an operator a way to measure, not a number to repeat.

Six patterns recur. Each has a consumer-side cousin in the dark-patterns literature, and each converts to money through the same mechanism: a correction the property is entitled to make becomes expensive enough in time or attention that it is not made.

None of these is exotic. Each is a defensible product decision in isolation: fraud controls justify confirmation steps; statement formats are constrained by finance systems; defaults reduce onboarding friction. The audit question is not whether a pattern can be justified but whether its friction is symmetric — whether the operator's path to pay less is as short as the platform's path to be paid more.

Commission leakage is commission paid on revenue the property did not earn, or at a rate higher than the one it meant to accept. It arrives through four doors.

• Abandoned corrections. A guest leaves a night early; the property is owed a commission reduction on the unused night. If the correction path takes twenty minutes across several screens and a waiting period, and the amount is $29, a busy front desk will often let it stand. The home page names the mechanism: the time cost of correcting a $30 commission exceeds the recovery.
• Uncaught invoice errors. No-shows and cancellations that were not marked in time appear on the statement at full commission. If the statement cannot be joined to the PMS folio line by line, the error is found only by someone who goes looking.
• Defaults that outlive their purpose. A visibility booster set for a soft week, a campaign extended automatically, a member-discount tier enrolled during onboarding: each continues to charge until someone turns it off, and the interface rarely says so where the number is.
• Friction labor. Even corrections that are completed cost time — the hours of the most junior people in the building, spent on the platform's workflow rather than on the property's guests.

The pattern matters more than any single instance because the amounts are small and the volume is large. A $29 line is below the threshold at which a manager escalates and above the threshold at which an auditor bothers; across a year of reservations the sum belongs on the P&L, not in the anecdote file.

Illustration: a 120-key independent at a $160 ADR and 65% occupancy sells about 28,470 room-nights a year. Assume 40% arrive through OTAs at an 18% commission — about 11,390 OTA room-nights, roughly 5,180 OTA reservations at a 2.2-night average stay, and an annual commission bill near $328,000. The rates below are assumptions chosen to be conservative; substitute your own.

Two readings of that result. The first is that $7,500 is small against a $328,000 commission bill — about 2.3% — and an operator who expected a scandal will be underwhelmed. The second is that it is pure margin, it recurs every year, it is concentrated in the hours of the least senior staff, and it scales linearly: at 1,000 properties like this one it is $7.5M; at 10,000 it is $75M. The home page's order-of-magnitude claim of billions industry-wide depends on the abandonment rate and the property count, which is exactly why the measurement method below matters more than any headline figure. The abandonment rate is the number to know, and only your own statements can supply it.

The audit takes one person about two working days for a quarter's data and produces a number the owner can act on.

1. 1

   Pull the statements

   Export or download every OTA commission statement for the quarter, plus the reservation-level detail behind it. Note the format you receive — PDF, paged web view, CSV — and whether a reservation identifier joins to your PMS folio. The absence of a join is itself a finding.

2. 2

   Reconcile against the folio

   Match every commissioned reservation to a checked-out folio. Flag no-shows, cancellations, early departures, and rate changes where the commission base on the statement differs from the revenue you actually collected. That difference is the gross error rate.

3. 3

   Sample the corrections

   Take 30 flagged items. For each, record whether a correction was filed, whether it completed, how many days it took, and whether the next statement reflected it. The share never filed or never completed is your abandonment rate.

4. 4

   Time the workflow

   Have the person who normally does it perform ten corrections while you count screens, clicks, waits, and elapsed minutes. Do the same for a bulk rate change across a 30-day range. Record the platform's actual path lengths, not your estimate of them.

5. 5

   Price the abandoned ones

   Multiply the abandoned items by the commission at stake on each; add the loaded labor cost of the completed ones. Annualize. This is your leakage figure, on your own data, defensible in a conversation with a market manager.

6. 6

   Inventory the defaults

   List every program, promotion, and campaign currently active on each channel, with its start date, end date, and who enabled it. Anything nobody remembers enabling is a default; price its uplift from the date it began.

7. 7

   Report it as a line item

   Put the figure on the monthly distribution report beside commission, not beneath it. A leakage number that is tracked tends to fall; one that is anecdotal tends to persist.

The consumer-protection frame is arriving at business interfaces slowly. The FTC's 2022 staff report catalogued dark patterns across subscription, e-commerce, and privacy contexts and signalled enforcement under existing unfair-practice authority; its examples are consumer-facing, but its taxonomy — obstruction, misdirection, hidden information, asymmetric choice — transfers directly. In the EU, Article 25 of the Digital Services Act prohibits online platforms from designing interfaces that deceive or manipulate recipients of the service or materially distort their decisions; how far it reaches a business-user portal is untested. The Platform-to-Business Regulation is the instrument written for operators: it requires plain-language terms, at least 15 days' notice of changes, an internal complaint-handling system, and disclosure of the main ranking parameters and of any paid influence on ranking.

ADAPT proposes the following standards for any operator interface that claims compatibility with its protocols — including the protocol-native PMS that is the operator's own console. They are written so that compliance can be tested by an auditor with a stopwatch and an export, not argued about.

The register lists this audit as a proposed workshop topic. ADAPT proposes three things. First, that the standards above be adopted as the operator-interface conformance profile for protocol-native property systems, carried by the PMS vendor track and demonstrated in the Exchange Building pilot, where the reference PMS is the operator's own console rather than a portal designed by a counterparty. Second, that the measurement method in this article be published as a shared worksheet, so that operators can pool anonymized abandonment rates and replace the order-of-magnitude claim with a measured one. Third, that the same discipline apply to ADAPT's own surfaces: an alliance that asks platforms for symmetric friction must publish its correction paths, its statement formats, and its defaults, and invite the same stopwatch.

The underlying principle is the one that runs through every ADAPT proposal. With programmable settlement, the commission base is the revenue actually collected, the correction is a field change rather than a workflow, and the terms that priced a reservation travel with it. A statement that cannot be reconciled is a symptom of terms that were never encoded. The companion analysis of the agreements themselves is Onerous Distribution Agreement Clauses; the settlement mechanics are in Programmable Commerce for Hospitality.