ChatGPT Ads Examples: What They Look Like, by Vertical (2026)

Short answer: A ChatGPT ad is a sponsored placement that appears below a relevant ChatGPT conversation, clearly labeled and visually separate from the organic answer. Per OpenAI, every ChatGPT ad is built from the same six components: advertiser name, favicon (the advertiser logo), title (headline), copy (description), landing page URL, and one image asset (creative). It looks like a single compact, labeled card — a brand favicon, a headline, a line of description, an image, and a link — that enters the conversation below the answer rather than interrupting it. OpenAI selects which ad to show by relevance-weighting it to the context and intent of the conversation.

Below we show, not just tell: a labeled anatomy of the six-component unit with a worked mock, how OpenAI describes the placement, what we can learn from OpenAI's three public reference advertisers (Best Buy, Lowe's, and VistaPrint), illustrative ad copy and title variations for DTC, SaaS, and professional services, a by-vertical comparison table, and the engagement patterns we observe at GPT Ads AI. For the foundational mechanics, see our pillar guide, What Are ChatGPT Ads?

What does a ChatGPT ad look like? The six components

Per OpenAI's documentation, every ChatGPT ad is assembled from the same six building blocks. There is no second headline, no expandable sitelink, no callout extension in the documented unit — it is a deliberately minimal, unified card designed for a user who is mid-conversation, not scrolling a feed. The six components OpenAI defines are:

1. Advertiser name — the business name shown on the card (for example, "Best Buy").
2. Favicon — the advertiser logo, displayed next to the name and the label.
3. Title (headline) — the primary line of copy that does most of the persuasive work.
4. Copy (description) — a short supporting line that adds the offer, proof, or detail.
5. Landing page URL — where the click goes. Per OpenAI, UTM and static tracking parameters are supported on landing-page URLs and persist through the ad click.
6. Image asset (creative) — a single image that gives the unit visual weight.

A labeled anatomy mock, component by component

Here is a worked mock that maps each OpenAI component onto a visible unit. Picture a user who has just asked ChatGPT, "What is the best magnesium supplement for sleep?" The model returns its organic answer. Then, below that answer and clearly separated, a single sponsored card appears. This is a GPT Ads AI illustration using a fictional brand:

• Label plus favicon (components 1-2): a label and a round brand favicon sit with the advertiser name — for our mock, "Rest Labs."
• Image (component 6): a clean product shot of a magnesium glycinate bottle on a neutral background.
• Title (component 3): "Magnesium Glycinate Built for Deeper Sleep."
• Copy (component 4): "Third-party tested. 400mg per serving. Free shipping over $40."
• URL (component 5): a visible link such as restlabs.com/sleep, carrying UTMs behind the scenes so the click lands in your analytics as a ChatGPT-sourced session.

That is the whole unit. Read it back and you can see why the format rewards clarity: there is no second headline to hide behind, so the title and copy have to earn the click on their own. If you want to plan the URL side of this, our free UTM builder generates a clean, conventioned tracking string that persists through the ad click.

Where do ChatGPT ads appear, and how are they labeled?

Per OpenAI's documentation, ads appear below the relevant conversation — never inside the answer text, never replacing the answer. OpenAI describes three trust pillars that govern the placement:

• Clearly labeled — the unit is labeled as advertising so it is unambiguous.
• Separate from answers — the ad is visually and functionally distinct from ChatGPT's organic response, and the answer is generated independently of who is advertising.
• Choice and control — users have controls over how their data is used for advertising.

The selection is relevance-weighted to the context and intent of the conversation. Per OpenAI, the inputs the system considers include the advertiser's context hints, the landing page, the ad title, and the ad copy. There is no separate, hidden ad format documented beyond the six-component unit.

Who actually sees these ad examples?

This changes which examples are even visible to your buyer. Per OpenAI, ads are shown to Free and Go users in the United States, Canada, Australia, and New Zealand. They are not shown to Plus, Pro, or any Business-plan user, and not to users OpenAI predicts to be under 18. So if you are a Plus subscriber trying to find a ChatGPT ad to screenshot, you will not see one — by design. We unpack the audience math in the pillar guide.

What real placements look like: Best Buy, Lowe's, and VistaPrint

OpenAI features exactly three advertisers publicly: Best Buy, Lowe's, and VistaPrint. These are the only OpenAI-endorsed reference examples, and each ran the same six-component card below relevant conversations.

• Best Buy — Amy Adams, Best Buy's VP of Media, is quoted on OpenAI's materials describing the value of showing up in the moments where consumers research products and make purchase decisions. The operator lesson (a GPT Ads AI read, not an OpenAI claim): the unit fits high-consideration retail research such as "which 65-inch TV is best for a bright room?", where a brand card below a comparison answer meets the buyer at the decision point.
• Lowe's — featured by OpenAI as a reference advertiser. The operator takeaway is category fit: project and "how do I" home-improvement queries are dense with purchase intent and map cleanly onto a product or service landing page.
• VistaPrint — featured by OpenAI as a reference advertiser in small-business and print. The operator takeaway is that the channel can act as a discovery surface that reaches buyers your existing programs may not, not just a retargeting layer for people who already know you.

A discipline note: OpenAI has not published the exact creative copy or precise performance numbers for these three brands in its public docs, so do not repeat any specific CTR, ROAS, or conversion figure attributed to them — there is no OpenAI source for one. Any other brand names you may see associated with ChatGPT Ads come from third-party press reports, not OpenAI, and should be treated as unconfirmed. We only present Best Buy, Lowe's, and VistaPrint as confirmed reference advertisers.

Example ad copy and title variations by vertical

Here is the part most guides skip: organized title and copy variations by vertical, with the reasoning. Because the user is mid-research, the pattern we find wins is educational, comparison, and outcome framing — not discount-and-deadline urgency that works on a cold social feed. Everything in this section is a GPT Ads AI illustrative example, not an OpenAI template, and the brands are fictional.

DTC supplements and eCommerce

Trigger intent: product research ("best magnesium for sleep," "compare creatine versus creatine HCL").

• Title A: "Compare the Top 5 Magnesium Forms"
• Title B: "Magnesium Glycinate Built for Deeper Sleep"
• Title C: "Third-Party Tested Magnesium — See the Labels"
• Title D: "Which Magnesium Is Right for You? Take the Quiz"
• Copy: "Independently tested for purity. 400mg per serving. Free shipping over $40 and a 60-day guarantee."

Why we think it works: comparison and "see the labels" framing matches a researching buyer and routes to a buy-stage page. Urgency such as "48-hour sale" tends to read as out of place against a thoughtful answer.

B2B SaaS

Trigger intent: tool evaluation ("best month-end close software," "how to automate reconciliation").

• Title A: "Cut Month-End Close Time in Half"
• Title B: "Close Software Built for Controllers — See the Demo"
• Title C: "Compare the 4 Leading Close Tools"
• Title D: "Automate Reconciliation in 30 Days"
• Copy: "Built for finance teams. SOC 2 Type II. Book a 15-minute demo and see your close timeline mapped live."

Why we think it works: lead with the outcome (faster close), prove it (compliance, fit), and route to a demo. Pair this with the Clicks (CPC) objective when the goal is qualified pipeline.

Professional services

Trigger intent: "should I hire / how do I" prompts ("do I need a lawyer to file a trademark," "how to find a fractional CFO").

• Title A: "Fixed-Fee Trademark Filing, Attorney-Reviewed"
• Title B: "Talk to a Trademark Attorney — Free 20-Min Consult"
• Title C: "File Your Trademark Right the First Time"
• Title D: "Trademark Search and Filing, Flat $599 Plus Fees"
• Copy: "Licensed attorneys, transparent flat fees, and a conflict search before you file. Book a free consult — no obligation."

Why we think it works: services buyers are buying trust. Credentials, fixed-fee clarity, and a low-friction consult tend to outperform vague "world-class expertise" claims. Route to a consultation or intake page, not a generic home page.

Comparison table: example ChatGPT ads by vertical

This is the artifact operators want — one table mapping the triggering conversation, an example title, example copy, the creative, and a recommended objective for each vertical. These are GPT Ads AI illustrative examples, not OpenAI templates or guarantees.

GPT Ads AI observed engagement patterns (illustrative)

OpenAI has not published precise CTR, ROAS, or conversion figures for any advertiser in its public docs, and we will not invent any. What follows are GPT Ads AI managed-account observations: anonymized, aggregated, directional patterns across the three verticals we work in most. They are not OpenAI numbers, not forecasts, and not guarantees of results. For the long-form versions with challenge, approach, and results framing, see our case studies.

DTC supplements — observed pattern

Across managed DTC supplement accounts, our observed CPCs generally land in the lower-to-middle of a broad $1.50-$9 range we see across DTC, SaaS, and services (a GPT Ads AI managed-account observation, not an OpenAI rate). The biggest lever we see is matching the title to the exact research question; in our accounts, comparison and "best X" framing tends to out-click discount framing. We treat the channel as a profit-and-loss line from day one, not a six-month science project — that posture runs through every case study.

B2B SaaS — observed pattern

In SaaS accounts, we observe demo and trial requests arriving from buyers who were actively evaluating tools inside ChatGPT — a high-intent, research-mode audience. CPCs in our accounts tend to sit higher than DTC within that same observed range, which is what we would expect for a considered B2B purchase. We see cost-per-lead hold up best when the landing page and the MQL definition are tightened in parallel. These are directional observations, not promises.

Professional services — observed pattern

For high-ticket services, we observe that trust-led, fixed-fee creative tends to produce better-qualified consults than urgency-led copy. CPCs in regulated service categories — legal, financial, and health — run materially higher than the rest of our observed range, a pattern we flag in advance during planning. In our experience the win is consult quality, not raw volume: a researching buyer who clicks a credentialed, flat-fee offer tends to arrive at the call already partly qualified.

What makes a ChatGPT ad work versus fail

After running live campaigns on the channel, the pattern we see is consistent. The unit rewards ads that enter the conversation rather than interrupt it. Here is our operator heuristic — a GPT Ads AI observation, not an OpenAI rule:

• Works: educational, comparison, and resource framing that extends the answer the user just read ("Compare the Top 5...", "See the labels", "Book a fit call").
• Works: a title that mirrors the research question and a landing page that pays it off immediately.
• Fails: urgency-and-deadline copy lifted straight from a cold social feed — it clashes with a thoughtful, trusted answer.
• Fails: a generic home-page destination that forces the user to re-find what they were researching.
• Fails: treating context hints like exact-match keywords. Per OpenAI, context hints live at the ad-group level, describe relevant conversations and topics, and do not guarantee delivery in any specific conversation — the system makes the final relevance decision. See our context hints field guide.

How a ChatGPT ad differs from a Google Search ad

The units look superficially similar — text plus a link — but the buyer and the selection model differ in ways that change how you write the ad. The ChatGPT side of this table reflects OpenAI's documentation; the Google side is general industry knowledge.

The practical consequence: you brief the ChatGPT system on where you would like to appear; it makes the final relevance call. Copy should extend the conversation, not shout over it. For pricing mechanics — the Reach (CPM) and Clicks (CPC) objectives and the relevance-weighted second-price auction — see the pricing section of our pillar guide.

A note on beta and evolving formats

OpenAI is explicit that the Ads Manager Beta is a beta product that is actively evolving, so the surface area can expand over time. The unit documented today is the six-component card described above. If you see screenshots of richer formats circulating, treat them as unconfirmed unless OpenAI documents them — we do not present any format beyond the documented six-component unit as confirmed. We track changes as OpenAI ships them in the Ads Manager setup guide.

How to turn these examples into a live campaign

1. Pick the two or three research conversations your buyer actually has, and write a title that mirrors each one.
2. Draft three to four title variants and one to two copy variants per ad group, using the by-vertical patterns above.
3. Build one strong image asset per concept — a product shot for DTC, a dashboard for SaaS, a trust cue for services.
4. Lock a UTM convention before launch with our UTM builder, and brief the system with strong context hints (see the context hint generator).
5. Choose your objective — Clicks (CPC) for pipeline and sales, Reach (CPM) for awareness in a new category — and set up conversion measurement inside Ads Manager Beta.
6. Want this built and run for you? Talk to our done-for-you ChatGPT Ads agency on a discovery call. We aim to have your first live campaign up in 7-14 days.

Factual claims about OpenAI's ChatGPT Ads product are sourced from OpenAI's public documentation (Ads in ChatGPT: The Basics, Ads Manager Beta Overview, Ads Manager Beta Account Setup, Create Ad Groups for ChatGPT, and the ads.openai.com landing page) as of June 2026. Best Buy, Lowe's, and VistaPrint are the only advertisers OpenAI features publicly; any other brand names come from third-party press reports, not OpenAI endorsements. Example ad copy, fictional brands, and all engagement patterns labeled "GPT Ads AI observations" are illustrative, anonymized, directional patterns from our managed client roster — not OpenAI numbers, not forecasts, and not guarantees of results. GPT Ads AI is an independent agency and is not affiliated with, endorsed by, or owned by OpenAI; "ChatGPT" is a trademark of OpenAI, used descriptively.