Performance Max for Travel Goals 2026: Hotel Guide

Hotel advertising breaks most of the habits that work everywhere else. The same room is sold by you and by five OTAs at once, the decision takes weeks, and a meaningful share of what you book gets cancelled before anyone checks in. Run Performance Max for travel goals with default settings and you get a campaign optimizing toward an event that may never happen, learning from data that arrives three weeks late.

This guide covers what the travel subtype actually is in 2026: how property assets differ from ordinary assets, what you genuinely need Hotel Center for, which bid strategies exist, how to set conversion windows that match a long booking cycle, and how to keep cancellations from poisoning Smart Bidding. Everything here is about working with the ad system correctly — accurate feeds, honest pricing, clean measurement.

Performance Max for travel goals: what makes the subtype different

Technically it is the same PERFORMANCE_MAX campaign type with one extra ingredient: an asset set of type HOTEL_PROPERTY attached to the campaign, where each property is described not by copy and imagery but by a Google place_id. If that asset set is not attached, what you have is a plain Performance Max campaign pointing at a hotel website.

With property assets connected, three things change:

Google knows which building you are advertising. Not “a hotel brand” but a specific property with an address, a Maps listing, reviews and photos. That unlocks travel surfaces where your ad appears in the context of that property.
Suggested assets become available. The system can propose headlines, descriptions and images assembled from property data. It is best-effort: some properties return nothing, and minimum asset requirements still have to be met by hand.
Budget is allocated automatically across channels — Search, Display, YouTube, Discover, Maps — inside one campaign, with final URL expansion in play.

The practical headline: Hotel Center access is not required. To run the travel subtype you need verified properties in Google Business Profile and their place IDs, nothing more. Hotel Center solves a different problem — showing live rates and availability in the booking module and competing on price against OTAs. These two mechanics get conflated constantly, and the confusion costs weeks.

Two setups, and how to choose

Setup	What it needs	What you get	Who it fits
PMax travel goals, no Hotel Center	Verified Business Profile listing, `place_id`, a `HOTEL_PROPERTY` asset set	Cross-channel reach, ads tied to a specific property, optimization toward bookings on your own site	Hotels with their own booking engine, small groups, management companies
Hotel Center with a rate feed	Hotel Center account, price and availability feed, link to Google Ads	Rates and availability inside the hotel module, price comparison against OTAs, free booking links	Properties competing head-on with aggregators on price
Both	All of the above	Price comparison plus upper-funnel reach	Groups of 10+ properties with someone who owns feed operations

If you are starting from zero, start with the first row. It needs no feed, it can be live in a day, and it produces the data you need to decide whether Hotel Center is worth the operational overhead.

Building property assets

The sequence is the same whether you work in the UI or the API:

Create an asset set of type HOTEL_PROPERTY. It is the container for every property this campaign advertises.
Create one asset per property. The only required field is place_id. Get it from the Place ID Finder in Maps Platform documentation, or export it from Business Profile if your properties are already verified.
Link the assets to the set and point the campaign’s hotel_property_asset_set field at it.
Link each property asset to an asset group with field type HOTEL_PROPERTY. This is the step teams skip. Without it the campaign exists but behaves like regular PMax.
Meet minimum asset requirements in the group: headlines, long headlines, descriptions, images in three ratios, a logo, and video if you have it.

Travel asset suggestions are allowlisted and best-effort. Treat whatever comes back as a draft: verify star rating, amenities, distance to the centre. Accuracy of the ad is the advertiser’s responsibility, not the generator’s, and inaccurate property claims are the fastest route to disapprovals and refund requests.

Getting place IDs right

Three reliable routes: the Place ID Finder for one-offs, the Places API by name and coordinates for groups, and a Business Profile API export when you manage the listings. The classic group-level mistake is grabbing the place ID of a neighbouring business in the same building — the hotel restaurant, the spa, the conference venue. Validate every ID against the address and the business category before upload.

Account structure: how many asset groups

The logic is the same as in standard Performance Max — an asset group is a unit of creative and audience meaning, not a micromanagement lever. The general principle is covered in our breakdown of asset groups and listing groups in Performance Max; what follows is travel-specific.

One property per asset group when you run up to 15–20 properties with distinct economics. You see which hotel absorbs budget and you can change creative surgically.
Cluster by market or class at larger scale: resort five-star, urban four-star in capitals, aparthotels. Inside a cluster the promise and the creative match, so grouping does not blur the message.
Split into separate campaigns only when economics or budget differ. Travel campaigns cannot use shared budgets, so splitting campaigns directly splits learning. Three campaigns at $30/day learn worse than one at $90.

Build audience signals from first-party data: past guests, newsletter subscribers, visitors to specific destination pages. For a hotel that beats any off-the-shelf interest segment, because repeat stays and lookalikes of past guests are where cheap bookings actually come from.

Bidding: what is available, what to pick

The travel subtype supports exactly two strategies: Maximize Conversions and Maximize Conversion Value, each with optional tCPA or tROAS targets. Commission-per-stay bidding, familiar from Hotel Ads, does not apply here. Portfolio strategies referenced by resource name are not supported either — the strategy lives on the campaign.

A sequence that holds up in practice:

Launch on Maximize Conversions with no target. Give it two to three weeks and at least 30–50 bookings before you constrain it. Any earlier and the target is simply a reach ceiling.
Move to value. As soon as you pass real booking revenue rather than a flat placeholder, switch to Maximize Conversion Value. A midweek single night and a week in a suite differ by an order of magnitude, and conversion-count optimization is blind to that.
Add tROAS last, anchored on the trailing 30-day actual, moving no more than 10–15% per iteration. Aggressive target hikes hurt more in travel than elsewhere: the campaign drops the long tail of destinations that was generating the learning data.

As a rough orientation, meaningful stabilization comes after roughly 50 conversions or three conversion cycles, whichever is later. In travel the second condition almost always dominates. Treat that as a benchmark to sanity-check against your own numbers, not a promise.

The travel trap: long lag and cancellations

A June booking for October, and a cancellation three days before arrival, are both normal — and both break default settings.

Conversion window

The default 30 days is often too short for hotels. If your CRM shows a median click-to-payment lag above 14 days, widen the window; otherwise the system cannot see a chunk of its own results and systematically undervalues upper-funnel activity. The trade-off is slower feedback. A workable compromise is 30–45 days for urban properties and 60–90 for resorts with early-booking behaviour. Decide from the conversion lag report, not from instinct.

Cancellations and real value

Optimizing toward confirmed bookings while ignoring cancellations teaches the system to find people who book easily and cancel just as easily. Two mechanisms work:

Conversion adjustments. Send a retraction on cancellation and a restatement when the amount changes. Timing limits and mechanics are covered in our piece on conversion adjustments in Google Ads.
Pass expected value instead of face value. If a segment historically cancels 25% of the time, pass 75% of the amount. That is more honest input for Smart Bidding than a full amount followed by a wave of retractions.

If you cannot run adjustments at all, at minimum separate primary from secondary conversions: non-refundable paid booking as primary, booking created as secondary for observation. The reasoning is laid out in our guide to primary and secondary conversions.

If you are a travel affiliate, not a hotel

Property assets are tied to properties you or your client operate, so an affiliate without rights to the property cannot use this subtype. The realistic alternatives:

Search campaigns against intent demand — destinations, trip types, seasonal queries. Structure and formats are covered in travel search campaigns and Things to do formats.
Standard PMax for lead generation with your own first-party data in the audience signals.
Video and Demand Gen for demand creation on long-consideration destinations — but only if you can measure upper-funnel contribution instead of judging it on last click. For seasonal on-sale dates where delivery has to land on a specific day, YouTube reservation campaigns with guaranteed reach are worth pricing out.

Your destination also has to hold up: all-in pricing including fees, clear cancellation terms, and a landing page that delivers what the ad promised. If you run many markets and need a managed account with real history and support behind it, see how agency Google Ads accounts are structured — that removes operational friction, but it does not substitute for the settings above. The rest of what we run sits in the PPC Rebels service stack.

Measurement beyond cost per booking

Three numbers that tell you more than CPA in travel:

Cost per stayed night, calculated post-checkout from PMS data. The only metric cancellations cannot inflate.
Direct share of the booking mix. The point of PMax for a hotel is to take volume back from OTAs, not to add a paid layer on top of the same demand. If direct bookings rise while total volume is flat, you are paying for demand you already had.
Incrementality. The dependable test is geo-based: hold out a set of markets and compare. Mechanics are in our write-up on incrementality and geo experiments. This matters most for brand queries, where part of the paid click is bought back from your own organic traffic.

If video sits in the mix, add the branded searches metric to your reporting — it shows how many people went looking for the property by name after exposure. How to turn it on and how not to over-read it is covered in branded searches as a demand metric.

Launch checklist

Verify every property in Google Business Profile; export and validate each place_id.
Create the HOTEL_PROPERTY asset set, add property assets, attach the set to the campaign.
Link property assets to asset groups as a separate, verified step.
Meet asset minimums: 5 headlines, 5 long headlines, 5 descriptions, images at 1:1, 1.91:1 and 4:5, logo, video.
Configure conversions: paid confirmed booking as primary, window based on CRM lag, real revenue passed through.
Enable conversion adjustments for cancellations and amount changes.
Upload audience signals: 24 months of guests, subscribers, destination-page visitors.
Launch on Maximize Conversions with no target; budget at least 10–15 expected CPAs per day.
Change nothing for 14 days except obvious breakage.
After 30–50 bookings switch to value, then layer tROAS in 10–15% steps.

Five expensive mistakes

Property assets created but never linked to an asset group. The campaign spends like standard PMax while you wait for travel inventory.
Optimizing on “booking started”. You will find people who start and abandon. Optimize on completed payment.
One flat value for every booking. Value-based bidding loses its entire point and the campaign chases cheap nights.
Splitting into micro-campaigns per property. Each learns alone, none gets enough data. Consolidate while the economics allow it.
Judging peak season on last click. Brand search takes all the credit and demand-building channels look unprofitable. Cross-check with attribution models and with experiments.

Performance Max for travel goals is not a switch you flip and wait for bookings. It is three things done correctly: clean property assets, honest value passed back including cancellations, and a conversion window that matches how long people actually take to decide. Get any one of them wrong and sophisticated automation turns into an expensive generator of cancelled reservations.

FAQ

Do I need Hotel Center to run Performance Max for travel goals?

No. The subtype runs on property assets with place IDs. Hotel Center is for showing live rates and availability in the hotel booking module and competing on price with OTAs.

How is a property asset different from a normal asset?

A normal asset is copy, an image or a video. A property asset is a pointer to a specific property via place_id. It does not replace creative: headline, description and image minimums still apply.

Can I bid on commission per stay like in Hotel Ads?

Not in this subtype. You get Maximize Conversions and Maximize Conversion Value, with optional tCPA and tROAS.

How many properties belong in one campaign?

The technical limit is generous; the practical limit is economic. Group properties with similar average value and a similar promise; separate them when economics or budget ownership differ.

What conversion window should a hotel use?

Start from the median click-to-payment lag in your CRM. Urban properties typically land at 30–45 days, resorts with early booking at 60–90. Use the conversion lag report to pick the number rather than copying a default.

How do I handle cancellations?

Conversion adjustments: retraction on cancellation, restatement when the amount changes. If that integration does not exist yet, pass expected value discounted by the historical cancellation rate for that segment.

Why is my campaign not appearing on travel surfaces?

Most often the property assets were never linked to the asset group with the HOTEL_PROPERTY field type. Second most common: the property is unverified in Business Profile, or the place ID points at the wrong business.

Should I keep a separate brand search campaign?

Usually yes. Keeping brand queries in a dedicated exact-match campaign lets you control the bid and see what brand traffic truly costs. Otherwise PMax absorbs brand demand and the whole campaign looks better than it is.

Do I need to upload video?

If you do not, the system will generate video from your images and text. For a hotel that is almost always worse than a real walkthrough, because the atmosphere is exactly what the guest is buying.

How soon can I draw conclusions?

Not before one full conversion cycle plus the learning period has passed — in travel that is typically three to six weeks. Week-two decisions are almost always decisions about noise.

What if booking volume is too low to learn?

Move one step up the funnel: optimize toward a micro-conversion that correlates with booking, such as an availability check or a payment-page view, build volume, then switch back. Consolidate campaigns instead of splitting them further.

Where do I see which asset group drives bookings?

PMax reporting breaks out asset groups and channels, so a one-property-per-group setup gives you a near-complete channel picture per hotel. For clean creative comparisons, use Performance Max asset experiments.

Performance Max for Travel Goals in 2026: Property Assets, Place IDs and Direct Bookings