Advertising on ChatGPT in 2026: formats, prices and launch
Advertising on ChatGPT is no longer a rumor or a presentation of the future. In 2026, OpenAI launched an advertising account, pay-per-impression and pay-per-click campaigns, and conversion optimization tools. Ads are shown separately from the model's response and are selected not only by keyword, but by the context of the current dialogue and expected relevance.
For the advertiser, this is a new channel with high context of intent, but not a ready-made copy of Google Ads. Here, demand is formed differently, matching an ad with a question works differently, and the content of the landing page, the honesty of the offer, and the quality of post-click measurement are especially important.
What is officially available in ChatGPT Ads
On the page OpenAI Ads You can already create an Ads Manager account, set up a campaign, add ad groups and launch advertising. The OpenAI Help describes three procurement models:
| Model | What does the advertiser pay for? | When to use |
|---|---|---|
| CPM | Per thousand impressions | Outreach, new product launch, demand generation |
| CPC | Per valid click | Landing page traffic, offer comparison, lead generation |
| oCPC | Per clicks with conversion optimization | When a conversion event has already been set up and data has been accumulated |
OpenAI also points out that ads can be seen by users of free plans and ChatGPT Go. Plus, Pro and Business plans do not display ads. The format remains separate from the response: the advertiser cannot buy mentions within the model's text or influence the ranking of sources.
This is the fundamental difference between the two tasks:
- ChatGPT Ads buys separate advertising placement;
- AI SEO and GEO increase the likelihood that the site will be found and used as a source in a regular response.
If an organic presence is desired, use a separate material. how to get into ChatGPT replies. Advertising does not replace this work.
What does an advertisement look like?
By official format description the block may include:
- advertiser name;
- favicon or logo;
- title;
- description;
- image;
- link to landing page.
The advertisement appears under the ChatGPT response and is labeled as an advertisement. This is not a native model recommendation. The user must understand where the answer ends and paid placement begins.
How targeting works without exact key matching
In search advertising, a marketer is used to collecting semantics, match types, and negative keywords. In ChatGPT Ads, the keys do not disappear completely, but become just one of the signals.
OpenAI takes into account:
- the topic and intention of the current conversation;
- ad title and text;
- landing page content;
- context hints set at the ad group level;
- expected utility and probability of outcome;
- additional personalization signals if the user has authorized their use.
Context hints do not guarantee display for a specific phrase. They are more accurately understood as describing situations in which a sentence is actually useful.
For example, for a CRM implementation service, a weak hint sounds like CRM, продажи, автоматизация. A more useful description of the problem: компания сравнивает коробочную и кастомную CRM, планирует миграцию из таблиц или ищет интеграцию с 1С и телефонией.
What to prepare before launching a campaign
1. Landing page for one scenario
The general services section rarely provides enough context. The user has already discussed the task with ChatGPT and expects to continue rather than return to the abstract advertising of the company.
The landing page should have:
- the same scenario as stated in the ad;
- clear result and limitations;
- evidence: cases, team, process, integration;
- answer to a question about price or valuation method;
- one main action;
- correct mobile version and fast loading.
2. Measuring conversions
Do not start oCPC until it is determined which event is considered an outcome. Submitting any form does not mean a quality lead.
For B2B it is worth separating:
- opening a form;
- sending an application;
- valid contact;
- qualified lead;
- scheduled appointment;
- commercial offer;
- sale.
If Ads Manager optimizes only for the top event, the system can result in many cheap but weak hits.
3. A set of different ads
OpenAI Creative Guidelines It is advised to give the system several meaningfully different options. There is no need to make ten permutations of one heading. It's better to check different motives:
- launch speed;
- risk reduction;
- integration;
- data migration;
- industry scenario;
- transparency of assessment.
Test plan for 30 days
Week 1. Preparation
- Select one product and one measurable scenario.
- Check Ads Manager availability for your country.
- Set up analytics and transfer of quality leads.
- Prepare a separate landing page or a relevant section of an existing one.
Week 2: Launch
- Create 2-3 ad groups for different contexts.
- Add several truly different creatives.
- Limit your budget and fix your stopping criteria.
- Don't change your bids, landing page, and ads at the same time.
Week 3. Quality check
Compare not only CPC, but also:
- cost of a qualified lead;
- failure rate and session depth;
- conversion from request to appointment;
- time until the first meaningful action;
- the quality of queries and contexts that brought the user.
Week 4: Solution
Only scale groups where the business outcome is proven. If there are clicks, but the funnel is not moving, first check the compliance of the ad with the landing page and the quality of the offer, and do not increase the bid.
How ChatGPT Ads differs from Google Ads
| Parameter | ChatGPT Ads | Search advertising |
|---|---|---|
| Main signal | Dialogue Context and Intent | Search query and campaign settings |
| Demand structure | Often exploratory and comparative | From informational to transactional |
| Compliance Management | Context hints and ad content | Keys, match types, audiences |
| The role of the landing page | Relevance signal and conversation continuation | Landing after search query |
| Channel maturity | Emerging Product | Stable Ecosystem |
A new channel should not be evaluated only by the cost of a click. You need to compare the cost of the business result and the quality of the user after the transition.
Who should test advertising in ChatGPT
The most logical pilot is for products where a person first studies the options and asks clarifying questions:
- B2B services and SaaS;
- complex professional services;
- education;
- technology and electronics;
- products with long selection cycles;
- e-commerce, where the buyer compares characteristics and use cases.
The channel is less suitable for businesses that cannot quickly process leads, do not distinguish between targeted and non-targeted requests, or drive all traffic to an overloaded home page.
FAQ
Is it possible to already run ads in ChatGPT?
Yes, OpenAI is developing Ads Manager and partner purchasing methods. Availability depends on the country and the status of the advertising account, so you need to check it immediately before launching.
How much does advertising cost on ChatGPT?
There is no single price list. Campaigns can run on CPM, CPC or oCPC, and the actual cost depends on the goal, audience, competition, ad quality and landing page.
Is it possible to buy a company mention in a ChatGPT response?
No. The advertisement is separate from the model's response. To organically get into answers, you need to work with site accessibility, content, entities, sources and brand awareness.
Are keywords needed?
Keywords can be used as context, but the system does not function as an exact replica of search advertising. The topic of the conversation, the intent, the ad copy, and the content of the landing page are all important.
What to choose for the first test: CPM or CPC?
For testing traffic and leads, CPC is usually clearer. CPM is appropriate for reach, and oCPC makes sense once conversions are set up correctly and sufficient data has been accumulated.
Useful on the topic
- How to get included in ChatGPT answers and become a cited source
- Why ChatGPT does not see the site: reasons and verification
- Performance marketing: when does it pay off?
- Setting up performance marketing