A metasearch budget: cost per click, cost per stay and what the hotel keeps

A hotel metasearch budget is built from four numbers the hotel owns: clicks, conversion, length of stay and cancellation. Multiply clicks by the price bid per click, divide by confirmed stays, and compare that cost per stay with what the same booking would cost on another channel. Pay-per-stay models bill a chosen percentage after checkout instead.

The click invoice arrives at the start of the month and the reservations screen shows fewer bookings than the clicks suggested. Nobody in the building can say whether the gap was the price, the landing page, or the guest who cancelled three days after booking. This article fixes that. After reading, you will be able to build a one-page worksheet from clicks to net revenue, compare metasearch with another channel on the same basis, and run a paid test with a written stop rule.

What are the two ways to pay for metasearch?#

A hotel pays metasearch either per click, whether or not the click becomes a booking, or per stay, as a percentage of the booking value. Some placements cost nothing per click at all.

First, the term. A metasearch site compares rates from several sellers for the same dates and sends the traveler to the seller's own page to finish the booking. The comparison site does not take the reservation. If you are new to the mechanics, start with how metasearch works and how rates reach comparison sites.

Under cost per click (CPC), the advertiser is charged the amount it bid for each click, and the click is paid whether or not it turns into a booking. Under cost per acquisition (CPA), also called pay per stay, the advertiser pays a pre-defined percentage of the booking value when a booking is made. Both models are listed on trivago's advertiser page (opens in a new tab) (trivago, "Get listed", 2026-11-07).

The third route is free. Google's Hotel Center Help (opens in a new tab) states there is no cost for clicks on free booking links, which are ranked by their usefulness to users and not by bids, show the site or hotel name with the room rate, and send the user to the landing page the partner provides (Google, 2026-11-01). The same page notes that a hotel using a reservation system may already be connected to Google through its provider, so the first check is what the hotel's own systems already publish.

The distinction matters for the budget because of timing. CPC spend leaves the account before any revenue arrives; CPA spend is billed after the stay. The same booking therefore lands in different months and carries different risk.

How much does trivago charge hotels, and what does Google charge?#

Neither publishes a price list a hotel can budget from. trivago offers CPC and CPA models where the amount comes from the hotel's own bid or the commission rate it chooses, and Google states that clicks on free booking links carry no cost.

trivago's advertiser page says it currently offers two marketplace models, CPC and CPA, and in both a user who clicks an advertised rate is sent to the advertiser's own website to complete the booking (trivago, "Get listed", 2026-11-07). Under CPC the advertiser is charged the amount it bid for each click; under CPA it pays a pre-defined percentage of the booking value when a booking is made.

The pay-per-stay detail sits in trivago Business Studio's help centre, which describes the pay-per-stay campaign in Rate Connect, trivago's tool for showing hotels' own website rates, as a CPA campaign. The hotel chooses one of the available commission rates, pays it on a guest's direct booking after the guest has checked out, and can choose a higher rate to influence how much exposure its website rates get. Billing for Rate Connect campaigns happens at the beginning of every month, and whether pay-per-stay is available at all depends on the hotel's connectivity provider (trivago, 2026-11-07). The entry steps are covered in claiming a trivago listing.

On the Google side, the only figure in this article is zero: no cost for clicks on free booking links, ranked by usefulness to users rather than bids (Google, 2026-11-01). Do not read a country rule into either page; Google's help pages state none, so the hotel checks availability inside its own account or with its provider rather than from a blog post. The same logic applies to Hotel Center and hotel ads on Google.

The practical consequence is that a budget is a set of assumptions about the hotel's own bid, conversion and cancellation, not a published tariff. Write the assumptions down next to the number.

How do you build the worksheet from click to net revenue?#

Run the numbers in one order: clicks, spend, bookings, room nights, gross revenue, confirmed stays, net revenue, then cost per confirmed stay. Every input comes from the hotel, and the output is a single cost per stay that can be compared with another channel.

The hotel supplies six inputs: clicks bought, price per click, conversion from click to booking, average length of stay, average daily rate, and the share of bookings cancelled. The order matters. Clicks times price gives spend. Clicks times conversion gives bookings. Bookings times length of stay gives room nights. Room nights times rate gives gross room revenue. Gross revenue reduced by the cancellation share gives net room revenue and confirmed stays.

Two outputs, both useful: cost per confirmed stay, which is spend divided by confirmed stays, and cost as a share of net room revenue, which is spend divided by net room revenue. The first compares channels; the second tests whether the margin survives. A third output suits long-stay properties: cost per occupied room night, which is spend divided by the room nights actually stayed.

Keep one currency and one period through the whole sheet. Mixing a monthly click invoice with an annual revenue figure is the most common arithmetic error in these budgets. Recompute for each source market or device if the account allows it, because conversion and length of stay rarely match across them, and an average of two different answers is not an answer.

These figures are invented for illustration and are not a benchmark. Replace every one with your own.

Why do cancellations and length of stay change the answer?#

Both sit between the click and the money. A cancellation removes revenue but not the click cost, and a longer stay spreads the same click over more room nights.

Under CPC the click is paid at the moment of the click, so a later cancellation leaves the spend in place and the revenue gone. The cost per confirmed stay rises while the invoice does not change. Under CPA the timing is different: trivago does not bill a direct booking cancelled on or before the check-in date, unless the hotel keeps part or all of the booking value (trivago Business Studio help, 2026-11-07).

Length of stay changes the comparison between hotels, not just between channels. Two hotels paying the same price per click get different costs per room night if one sells three-night stays and the other sells one-night stays.

Count cancellations at the same point in time on every channel. A booking counted on the day it is made and a booking counted after checkout are two different numbers wearing the same name. Use the conversion of metasearch clicks specifically, not the average conversion of all website traffic, which usually includes direct visitors who were already going to book.

Re-run the sheet with a worse cancellation share and a shorter stay, and see whether the decision still holds. If it only works under the optimistic pair, it does not work.

What would the same booking have cost on another channel?#

Put the alternative on the same line: cost per confirmed stay, in the same currency, at the same cancellation point. If metasearch costs more per confirmed stay than the alternative for the same room and dates, the click is not paying for itself.

Take the commission rate from the hotel's own contract, not from an article, and apply it to the same net room revenue the worksheet produced. Add the lines the commission hides: payment processing, connectivity or channel-manager charges, and any per-booking fee the hotel actually sees on an invoice. Use only figures that appear on the hotel's own documents.

Compare like with like: same room type, same rate plan, same dates, same cancellation policy. A flexible rate and a non-refundable rate are not the same product. Compare the guest as well as the cost: a channel that brings a first-time guest and a channel that reaches people who already know the hotel are not interchangeable, even at the same cost per stay. The wider method is covered in comparing distribution channels by what the hotel keeps.

Write the decision rule before looking at the result, so the number does not get interpreted after the fact. Keep the worksheet for the next budget round: the value of this exercise is the comparison over several months, not a single reading.

How do you run a small test with dates, budget and a stop rule?#

Fix three things before the first click: a budget, an end date, and a stop rule that says what result ends the test early. Then check the landing page, because the click goes to the hotel's own site.

Write the stop rule as a sentence with a number in it. For example: stop when cost per confirmed stay passes a stated ceiling, or when a stated number of confirmed stays has been counted, whichever comes first. Set the budget as money, not as a target: the amount the hotel is willing to lose in order to learn the conversion figure.

Define what counts as a confirmed stay, and define it the same way for the comparison channel: after check-in, after checkout, or after the cancellation window closes. Check the landing page first. trivago sends the clicking user to the advertiser's website to complete the booking (trivago, "Get listed", 2026-11-07), so a slow page or a rate that differs from the advertised one wastes the click before any budget question arises.

Review weekly in the same worksheet, and write down what changed: bid, rate, page, dates. A result without its change log cannot be repeated. End the test in one of three ways: continue at the same settings, change one setting and continue, or stop. All three are results; only drifting is not.

When is metasearch not worth it yet?#

Metasearch is not worth paying for while the click lands on a page that cannot complete a booking, or while the hotel cannot measure a confirmed stay.

The first blocker is technical: no booking engine, or a booking page that cannot finish the reservation, means the click is bought and the guest is handed to someone else. The second is commercial: a rate on the hotel's own site that is visibly worse than the rate on a booking site means the traveler compares on the comparison site and leaves through the cheaper door.

The third is statistical: too little volume to conclude anything inside the budget. If the test cannot produce enough confirmed stays to read, the money buys an anecdote. The fourth is margin: a cost per confirmed stay that takes a large share of net room revenue may still be worth running, but not before the rate or the page is fixed. The fifth is ownership: without one named person reviewing weekly, the test does not stop, it just stops being watched.

The free route comes first. Google states there is no cost for clicks on free booking links (Google, 2026-11-01), so where they are available to the hotel, they are the cheaper way to learn whether the page converts before any click is bought.

What to do next#

Open the worksheet today, before any budget is approved. Fill in your own clicks, conversion, length of stay and cancellation, then write the stop rule and the comparison channel's cost per confirmed stay into the same sheet. The ideas fit together in one order: the payment model tells you when the money leaves, the worksheet tells you what a confirmed stay costs, the comparison tells you whether that is acceptable, and the stop rule tells you when to stop asking.

If the hotel's rates already live in one inventory, publishing them to metasearch engines is a smaller step than it looks: AI Otel's metasearch feeds are built, with no accepted connection yet, so rates can be published to metasearch engines from the same inventory. If you want to see how that works, request a briefing.