How to Set Up an OpenAI Ads Campaign: A Step-by-Step Guide

Quick answer: To set up an OpenAI ads campaign, first confirm that your business is eligible for ChatGPT advertising and that self-service Ads Manager is available to your legal entity. Create an advertiser account through OpenAI Ads Manager, complete business verification and billing, then create a campaign with an appropriate objective, budget, dates and targeting. Organize your ads into focused ad groups, add context hints that explain when your offer is relevant, write benefit-led ad variations, and send users to specific, accessible landing pages. Before activating the campaign, configure conversion measurement, validate tracking and review OpenAI’s current advertising policies. Submit the ads for review and verify that the campaign is approved and serving before evaluating performance.

OpenAI Ads Manager is a beta product, so available objectives, markets, targeting options and measurement features can change. Check the current Ads Manager availability guide before committing a budget.

What are OpenAI ads, and how are they different?

ChatGPT advertising introduces a different environment for reaching potential customers. Rather than relying exclusively on a typed search query, an advertiser can reach people while they explore options, compare products, research a service or work through a decision in a conversation.

In the current advertising experience, ads appear below ChatGPT responses and are distinguishable from the answer itself. An ad can include the advertiser's name and favicon, a title, descriptive copy, a landing page and an image. OpenAI uses the context of the conversation and, where ad personalization is enabled, selected signals from a user's broader ChatGPT experience to assess relevance.

That distinction matters for campaign strategy. Someone searching for “cheap flights from London to Dubai” has expressed a relatively explicit requirement. Someone asking ChatGPT to help plan a family holiday, compare destinations or understand the cost of a trip may reveal a broader decision-making context. An advertiser's opportunity is to present a genuinely useful offer when it fits that context—not to assume every conversation is a direct purchase signal.

ChatGPT ads compared with search and social advertising

These are strategic distinctions, not absolute boundaries. Search can support discovery, social ads can generate high-intent demand, and ChatGPT conversations can range from early research to purchase-oriented comparison.

OpenAI describes its ad selection as a relevance-weighted, second-price auction. Advertisers therefore should not treat the maximum bid as the only factor that matters. Relevance and the value of the ad to the user also matter in the selection process.

Before you start: the readiness checklist

The easiest way to waste time during campaign setup is to discover an eligibility, billing or landing-page problem after the campaign has been built. Resolve the following items first.

• Country and legal-entity eligibility: Confirm that the country where the business is legally established and billed is supported for self-service access. Being able to open the website or sign in does not necessarily mean your business can advertise through self-service. Some markets or categories may require a managed or partner-led route.

• Account ownership: Decide whether the advertiser account belongs to your own business or to a client. Agencies should establish a clear ownership and access model before creating the account.

• Billing: Have the business name, billing address, invoice email and supported payment method ready. Campaigns cannot deliver until the required billing setup is complete.

• Landing-page access: Make sure the destination works without a login, does not return errors to OpenAI's advertising crawler and clearly represents the advertised offer. OpenAI requires advertisers to allow OAI-AdsBot for landing-page validation.

• Creative assets: Prepare a recognizable brand name, favicon or logo, clear headlines, concise descriptions and suitable images.

• Measurement: Decide what counts as success—such as a qualified lead, completed booking or purchase—and implement the relevant tracking before launch.

• Policy review: Check whether your product, service, destination and claims are eligible. A technically correct campaign can still be rejected because of its category or landing-page content.

For travel businesses, this checklist should also include live inventory, transparent pricing, cancellation information, mobile usability and the ability to preserve booking details when a visitor moves between the landing page and booking engine.

Step 1: Create your advertiser account

Go to ads.openai.com and sign in with your existing OpenAI account or create one. You have two setup paths: enter your website URL and let Ads Manager help create an initial ad and campaign, or skip the initial ad and enter Ads Manager to build the campaign yourself. The second route is useful if you want to plan the campaign structure, measurement and budget before anything is submitted for review.

Understand the three names before you proceed

Ads Manager distinguishes between three names:

• Account Name: Your internal label for managing the advertiser account.

• Brand Name: The public name displayed alongside your ads.

• Legal Name: The registered business name used for billing and business verification.

The initial advertiser name fills all three fields, but you can edit them separately after account creation. Changing the Brand Name triggers a review and can pause ad serving until approval; changing the Legal Name restarts business review and may also interrupt delivery. By contrast, the account's country or region, currency and time zone are fixed after creation. Review these details carefully before submitting the setup form.

Complete the billing profile, add the required payment method, confirm your logo and invite teammates through Settings → Users. Assign Admin access only to people who need full account control; Members can manage campaigns, while Viewers can inspect account content without managing it. If you manage advertising for several clients, use a separate advertiser account for each client rather than placing unrelated businesses under one brand identity.

Step 2: Understand the campaign structure

OpenAI Ads Manager uses three levels: campaign, ad group and ad. Each level has a different job, and keeping them distinct makes reporting and optimization more meaningful.

The campaign answers what business outcome are we pursuing? The ad group answers which related need or use case are we addressing? The ad answers what message should the potential customer see?

Avoid building one campaign that combines unrelated services simply because they belong to the same company. An OTA selling flights, hotels and car rentals may need separate campaigns if the economics, landing pages, seasonality or objectives differ. Within a campaign, create ad groups for meaningful intent themes and use multiple ads to test different value propositions for that same theme. OpenAI recommends separating materially different business goals, regions and product categories where appropriate.

Step 3: Choose the campaign objective

The objective determines how delivery is optimized and which billing approach applies. OpenAI's current campaign setup documentation lists Views, Clicks and Conversions. Choose based on what you need to learn or achieve, not simply because one option sounds more advanced.

For a new advertiser without reliable ChatGPT Ads performance data, Clicks can provide a practical starting point: it helps establish whether the offer earns traffic and whether the landing page can convert it. Views may make sense when awareness is the primary goal and the campaign can be assessed on reach and subsequent effects. Conversions is more appropriate when the event is implemented correctly and there is enough signal to evaluate results.

Why the objective matters: OpenAI states that a campaign's objective cannot simply be changed after creation. If you want to move from Clicks to Conversions, create a new campaign with the new objective, then configure the conversion event and relevant settings. Plan for that possibility from the beginning rather than treating the objective as a setting you can freely switch later.

Step 4: Set the budget, bids, dates and targeting

Daily budget or campaign-total budget?

A daily budget is an average delivery target across the Sunday-to-Saturday budget week in your account's time zone. It is not necessarily a hard cap for each individual day. Under the current published limits, spend can reach up to twice the selected daily budget on an individual day, while the total over the applicable seven-day period will not exceed seven times that daily budget, subject to the platform's rules for budget changes.

For example, a campaign with a daily budget of $50 does not promise to spend exactly $50 every day. Delivery may be higher on one day and lower on another. If the campaign spends its weekly allowance early, it may stop delivering until the next budget week even though its status remains Active. A campaign-total budget instead defines the amount available over the campaign's full duration and gives you a stricter total-spend boundary.

For a first test, a daily budget can make pacing easier to observe. The current minimum daily budget depends on the billing currency and is listed in OpenAI's campaign setup documentation. Check the current amount in the account rather than relying on a figure copied from an older guide.

Set bids with delivery in mind

OpenAI uses a relevance-weighted, second-price auction. The maximum bid is a ceiling, not a guarantee that you will pay that amount on every impression or click. A competitive bid can help delivery, but the auction also considers relevance. An advertiser should therefore avoid assuming that simply raising the bid will fix weak creative, poor context alignment or a landing page that does not fulfil the ad's promise.

Set the maximum bid based on your economics. For a travel booking business, the useful ceiling depends on the margin per booking, the proportion of leads that become confirmed bookings, cancellations, supplier costs and repeat value. A click that looks affordable in isolation may be unprofitable if the booking funnel has a low completion rate or the product has thin margins.

Choose countries, locations and platforms

Country targeting determines where ads are eligible to appear. Some new self-service accounts may initially be limited to their home country until identity verification and the required spend conditions are met. Adding a country in campaign settings does not necessarily override this restriction. For non-product-feed campaigns, additional supported targeting can include states or regions, cities, markets and postal codes where available. Platform options currently include supported Android, iOS and web placements; use the options actually presented in your account.

Custom audiences can be included or excluded where supported. Treat them as a controlled test rather than a default requirement: audience eligibility and matching thresholds can vary, and some regions have limitations. Confirm the current rules before relying on retargeting or customer-list targeting.

Avoid the first-campaign start-date mistake

The first campaign created during signup may not expose start and end dates before publication. Those controls can be available when you edit the campaign afterward, but publishing before setting a future start date can allow approved ads to begin serving—and spending—earlier than intended.

If your campaign must start on a specific date, do not publish it merely to reach the scheduling controls. Follow the current setup flow, configure the schedule before publication where possible, and confirm the final campaign status, dates and billing readiness before activating it.

Step 5: Build ad groups around context hints

Context hints help the system understand when an ad may be relevant. They are additional descriptions of your offering, audience need or use case, entered at the ad-group level. They are not exact-match keywords, audience restrictions or instructions that guarantee placement. OpenAI says the system can recognize relevance without the user repeating the exact wording of a hint.

Consider a travel agency offering flexible flight bookings. A weak hint might be:

Flights, travel, holidays, cheap tickets, airline deals.

This is a disconnected list. It says little about the specific customer problem the business solves.

A stronger hint would be:

Flexible flight options for families planning international holidays who need to compare departure times, baggage allowances and change conditions before booking.

The stronger version describes a concrete use case, the product's relevance and the decision the customer is trying to make. It does not ask the system to show the ad only to a particular demographic or force delivery in a specific city; those controls belong in supported campaign targeting.

For a travel business, separate hints might describe business travel with changeable fares, multi-city international trips, family holidays with checked baggage, or travellers comparing hotel locations before choosing a property. Each hint should reflect an actual offer and a landing page capable of supporting it.

This is where high-intent keyword and campaign structure work remains valuable even though ChatGPT ads are not simply keyword-targeted search ads. The discipline of distinguishing customer needs, offers and landing pages still applies; the implementation is different.

Step 6: Write the ads

A ChatGPT ad needs to make its value clear quickly. OpenAI recommends specific, benefit-focused copy and distinct creative variations rather than multiple versions that merely rearrange the same words. The system needs meaningful alternatives that express different reasons the offer could be useful.

The stronger examples are not guaranteed to outperform the weaker ones, they illustrate the principle of explaining a specific benefit. The right message depends on the actual product, the audience need and what the destination page delivers.

Write several distinct variations for each ad group. One can emphasize convenience, another explain a differentiating feature, and another reduce a common source of uncertainty. Keep claims verifiable. Avoid suggesting that fares are guaranteed, availability is unlimited or a saving is certain unless the offer genuinely supports that statement.

The landing page is part of the ad experience, not a separate afterthought. An ad about flexible flight changes should not lead to a generic homepage where the visitor must search again to discover the conditions. A hotel comparison ad should lead to relevant inventory or a useful destination page, not a page that makes the user restart the task. For more detail on message development, see Teckgeekz's guide to ad creative and copy strategy.

Image and landing-page requirements

Use an image that supports the offer and brand, not a decorative asset unrelated to the message. OpenAI's current bulk-upload guidance describes publicly accessible square PNG or JPG image assets, with a maximum size of 1200 × 1200 pixels. Confirm the format and dimensions shown in the live ad builder, because requirements may change.

The destination must be reachable by OpenAI's crawler. At minimum, allow OAI-AdsBot to access the advertised landing page. Review robots.txt, CDN and firewall rules, bot protection, CAPTCHA or JavaScript challenges, authentication and rate limiting. A page that loads normally for a human can still fail automated validation if a security layer blocks the crawler.

For deeper travel-specific guidance, connect this work to landing page optimization for travel PPC.

Step 7: Configure measurement before launch

Do not wait for the first click to decide how you will measure success. Before launch, document the event that matters, confirm how it will be collected and test the entire journey from the ad click to the final action.

Pixel, Conversions API and conversion events

OpenAI supports conversion measurement through its JavaScript Pixel and Conversions API. The Pixel can capture supported events in the browser, while server-side event delivery can provide a complementary measurement path. Where appropriate, use both according to OpenAI's current implementation guidance, with event deduplication and consent requirements handled correctly.

Choose events that reflect meaningful business outcomes. A travel advertiser might track a search initiated, a booking flow started and a booking completed, but should not treat all three as equivalent business results. If the primary goal is a completed booking, ensure the conversion event represents an actual completed booking rather than a visit to the confirmation URL that can be refreshed or revisited.

Preserve attribution through redirects

One of the most common implementation failures happens when a click passes through a redirect, a tracking service, a regional URL or a booking engine and loses its attribution parameters. Test the full route, including mobile redirects, cross-domain transitions and any third-party booking supplier.

Add appropriate UTM parameters to landing-page URLs so GA4 and other analytics tools can identify traffic from ChatGPT Ads. Use consistent campaign naming and check that the parameters survive redirects. UTM tags help you analyze sessions and downstream behavior, but they do not replace the platform's own conversion tracking.

Connect the platform view with GA4

Use Ads Manager to review impressions, clicks, spend, click-through rate, average CPC, average CPM and attributed conversions where measurement is configured. Use GA4 and your booking or CRM system to investigate engagement, qualified leads, completed bookings and revenue.

Expect differences between reporting systems. They can use different attribution rules, event definitions, reporting windows and data-processing schedules. OpenAI's primary Conversions metric currently represents click-through conversions; view-through conversions, when available in exports, are supplemental and should not simply be added to that number. Impressions, clicks and CTR refresh approximately every 15 minutes, while spend reporting can be delayed by several hours.

A difference between Ads Manager and GA4 does not automatically mean one system is broken. Compare like with like: date range, time zone, conversion event, attribution window and traffic filters. For travel businesses, reconcile platform conversions against confirmed bookings and cancellations wherever possible.

Step 8: Complete the pre-launch review and activate

Before activation, check the account and campaign as one connected system.

• Confirm business verification and billing are complete.

• Verify the campaign objective, budget type, bid settings and dates.

• Check the country, location and platform settings.

• Ensure each ad group has a clear theme and accurate context hints.

• Review every title, description, image and landing-page URL.

• Test the destination on mobile and desktop, including any redirects.

• Confirm that OAI-AdsBot can reach the destination.

• Validate the Pixel, Conversions API and UTM parameters.

• Check the ad and campaign review status.

If you want to build first and launch later, save the campaign in a paused state where the workflow supports it, and verify that it is actually paused. A draft or a campaign awaiting review is not the same as a campaign that has been deliberately scheduled for a future date.

Submit the ads for review and inspect their status in Ads Manager. Approval alone does not guarantee delivery: billing, account readiness, campaign status, dates and other delivery requirements must also be satisfied. If an ad is marked as not serving, investigate the stated reason before making unrelated changes.

After launch: what to monitor and when to optimize

During the first days, distinguish between a setup problem and a performance problem. If impressions and clicks are absent, check review status, billing, dates, targeting and bid competitiveness first. If the campaign is delivering but performance is weak, investigate the message, context hints, audience or location coverage, landing page and conversion tracking.

Do not make several major changes simultaneously. If you rewrite all ads, replace the landing page and change the budget at the same time, it becomes difficult to identify which change influenced performance. Make changes in a controlled order and keep a record of what changed and when.

A sensible learning window depends on traffic volume, spend, the sales cycle and the conversion event. There is no universal number of days that guarantees enough data. A low-volume B2B service may need longer to produce qualified leads than a high-volume consumer product needs to produce clicks. In travel, bookings can also lag the first visit while customers compare dates, prices and alternatives.

When should you move from Clicks to Conversions?

Consider a new Conversions campaign when:

1. The conversion event is correctly implemented and tested.

2. You have enough reliable event data to evaluate performance.

3. The landing page and checkout process work consistently.

4. You can judge success using a business metric such as cost per completed booking or qualified lead—not just click volume.

Because the objective is locked after campaign creation, the transition requires a new campaign rather than a simple objective edit. Keep the original campaign's reporting for comparison and make sure the new campaign has the correct conversion event and billing setup.

Advanced options: when to use them

As the platform develops, additional features can support more sophisticated campaign operations. Their availability may vary by advertiser, market and beta access.

Product-feed campaigns have their own setup rules. OpenAI says product information in feeds is eligible for product ads during the current beta and does not automatically make those products appear in organic ChatGPT conversations. Product feeds should therefore be evaluated as a paid catalog workflow, not as a shortcut to organic visibility.

Policies and ad review: what can get rejected?

Review the official OpenAI Ad Policies before investing time in creative production. OpenAI applies restrictions to ad content, destination pages and the contexts in which ads can appear. The policy currently identifies travel and experiences among the consumer categories supported during the initial rollout, while several regulated categories require specific approval or are not supported. Eligibility can vary by category and market.

Common reasons to investigate include:

• Category restrictions: The product or service is not eligible in the selected market, or it requires approval the advertiser does not have.

• Mismatch between ad and destination: The creative promotes one offer while the landing page introduces a different or disallowed offering.

• Unsupported claims: The ad makes misleading promises, guarantees or claims the business cannot substantiate.

• Crawler access failures: The page is blocked by robots.txt, a firewall, a CDN rule, a CAPTCHA, authentication or another bot-protection layer.

• Brand or interface confusion: The ad imitates ChatGPT or another OpenAI interface in a way that could make users believe the advertisement is part of the product.

• Sensitive or unsafe context: The placement would be unsuitable under OpenAI's brand-safety and user-trust rules.

Travel is not categorically excluded; it is listed as a supported consumer vertical in the current policy. That does not mean every travel-related offer is automatically eligible. Review the exact product, claims, destination and market before launch. If an ad is rejected, use the rejection reason to identify whether the problem lies in the creative, the landing page, account eligibility or the product category, and use the available review or appeal process when appropriate.

Common mistakes to avoid

Most early campaign problems are not caused by a lack of advertising features. They come from mismatches between the objective, the customer need, the offer and the measurement setup.

Treating context hints like keywords. A long list of broad terms is less useful than a precise description of when a product genuinely helps. Hints should explain relevance, not attempt to force delivery.

Sending every ad to the homepage. A homepage makes users repeat work the ad should have helped them complete. Match the destination to the need described in the ad.

Choosing Conversions before validating the event. An optimization system cannot compensate for a broken or poorly defined conversion signal. Test the event before building a campaign around it.

Assuming the daily budget is a hard daily cap. Budget pacing can move spend across days within the published weekly limits. If your finance team needs a firm campaign-wide ceiling, evaluate the campaign-total budget and the account's billing controls.

Publishing before checking the schedule. A first campaign created during signup may be able to serve after review before you have configured a desired future start date.

Making too many changes at once. Changing bids, hints, copy, budget and landing pages together destroys the clarity of your experiment.

Judging success on clicks alone. Cheap traffic is not the same as profitable traffic. Track qualified leads, completed bookings, revenue and cancellations wherever the business model requires them.

OpenAI ads for travel agencies and OTAs

Travel is a useful example of why campaign structure matters. A traveller comparing flight options has different questions from someone selecting a hotel or arranging a rental car. Those products have different booking journeys, pricing rules, margins and conversion barriers. Combining them into one campaign may obscure which offer is working and why.

Build campaigns around product economics

A travel business could start with separate campaigns for flights, hotels and car rentals if each product line has sufficient budget and a distinct business goal. Within the flight campaign, ad groups might address flexible international fares, family travel, multi-city itineraries or business travel. Hotel groups could focus on destination stays, family accommodation or properties near key attractions. Car rental groups might distinguish airport pickup, longer rentals or vehicles suited to a family itinerary.

These are starting hypotheses, not guaranteed targeting categories. Use context hints to explain the actual service and use supported location settings to control where ads can be delivered.

Make the landing page and booking engine part of the test

Travel customers can abandon a booking when search results are slow, baggage costs are unclear, the fare changes unexpectedly, cancellation conditions are difficult to find or checkout requires too many steps. Sending more visitors into that experience will not automatically solve those problems.

For flights, ensure the destination supports the advertised route or search intent and makes fare conditions clear. For hotels, carry the destination and relevant stay context into the search experience where possible. For car rental, make pickup locations, vehicle terms, insurance information and additional fees easy to understand.

This is where booking engine conversion optimization becomes an important companion to media buying. Evaluate the full path from the ad click to the confirmed booking, including supplier response times, mobile performance, payment failures and cross-domain tracking.

Ring-fence a test budget alongside Google Ads

Do not automatically transfer a large share of an established Google Ads budget into a new platform before you know what it can deliver. Set a defined experimental allocation, decide the maximum acceptable cost per qualified action, and agree on what evidence would justify increasing spend.

Compare channels using consistent business outcomes, but account for differences in intent and attribution. A platform that introduces a customer early in their planning journey may not receive the same last-click credit as a search ad clicked immediately before booking.

For a broader framework, see the travel PPC marketing guide and the guidance on budget allocation and ROI optimization.

Where paid advertising sits next to GEO

Paid placement and organic citation are separate levers. An approved ChatGPT ad is a paid placement governed by advertising eligibility, campaign settings, relevance and auction mechanics. Organic citations in ChatGPT, Perplexity or Claude depend on different systems and should not be treated as a result that an ad campaign can purchase or guarantee.

That distinction is important for measurement and strategy. Paid campaigns can test an offer, introduce a brand to users and send measurable traffic to a landing page. GEO—Generative Engine Optimization—focuses on improving the chances that a brand's information is understood, trusted and cited in AI-generated responses. It involves content quality, clear answers, technical accessibility, entity consistency and credible references across the wider web.

Do not assume that buying ads automatically improves organic AI citations, or that organic citation work will make a paid ad serve. These activities can support a broader discovery strategy, but their outcomes should be measured independently. For a deeper explanation, read SEO vs AEO vs GEO and our guide to getting cited by ChatGPT, Perplexity, and Claude.

Practical insight: build the measurement system before scaling the media

A useful operating principle is to treat the first campaign as a structured experiment rather than a launch-day performance contest. The early goal is to learn which customer needs produce meaningful engagement, which messages attract the right visitors and whether the landing page converts that interest into a business outcome.

“A new advertising channel is only valuable when you can connect the message to a real customer need and then measure what happens after the click. We would rather begin with a controlled test, a clear conversion definition and a landing page that delivers on the ad's promise than scale spend before the fundamentals are in place.” — Jeffrey Mathew, Teckgeekz

For advertisers, the practical sequence is straightforward: verify eligibility, build a coherent campaign structure, create genuinely distinct ads, validate tracking, and only then use performance data to decide what deserves more budget.

How Teckgeekz helps

Teckgeekz helps businesses connect paid acquisition with the technical and conversion foundations that determine whether traffic produces results. That includes pay-per-click management, campaign structure and creative strategy, landing-page optimization, conversion tracking and travel booking-engine improvements.

For travel agencies and OTAs, the work may also involve aligning flight, hotel or car-rental campaigns with the relevant booking journey, validating tracking across supplier integrations and identifying where booking friction limits the return on media spend. The objective is not simply to generate more clicks. It is to build a measurable path from customer intent to a useful landing experience and a completed business outcome.

What changes fast: official resources to check before publishing

OpenAI Ads Manager is still evolving. Features, eligible markets, policy rules, minimum budgets, creative specifications, measurement options and integrations can change during the beta. Keep this section in the published article and review the linked documentation periodically.

• Ads Manager availability by country — check whether self-service access is available to your legal entity.

• Create campaigns for ChatGPT Ads — objectives, budgets, targeting and campaign settings.

• OpenAI Ad Policies — review category eligibility, prohibited content and placement rules.

• OpenAI advertiser crawler guidance — confirm that landing pages can be accessed for review.

• OpenAI Ads conversion tracking documentation — implementation details for the Measurement Pixel and Conversions API.

• OpenAI Ads developer documentation — current developer tools and supported API workflows.

Before each new campaign, confirm the relevant requirements in the live interface rather than relying exclusively on a previously published guide.