ChatGPT Ads with Visual Formats: What Matters Now

ChatGPT Ads with visual formats will appear during image generation from late October 2026 onwards—initially as a test in the US. For self-employed professionals in the DACH region, the question isn't about polish, but about matching the right visual to the right intent: When is an image worth more than text, when does it just distract, and how do you measure whether it actually brings qualified enquiries?

Here's the thing: A beautiful image is worthless if it sets the wrong expectation. Someone's looking for terrace inspiration, sees your ad, clicks—and lands on a generic homepage with no price, no region, no process. The click costs you money, the enquiry never comes.

OpenAI announced a new visual ad format on 5 October 2026. The ads will appear alongside AI-generated images, clearly labelled and outside the generated image itself. The test starts initially in the US for users of the free and Go versions. A DACH launch hasn't been confirmed. Source

This article shows you a concrete test plan: visual types, offer, landing page, metrics and a small budget. It separates inspiration from purchase intent and explains when an image does more harm than good.

What's changing with ChatGPT Ads in October 2026

The previously predominantly text-based ad display is being expanded with a visual ad format. OpenAI describes possible use cases as product inspiration, demonstrating product use, and visual representation of an experience that a service enables. Source

The ad will appear next to the generated image and won't be embedded in the image. This isn't product placement within the AI image, but a separate advertising space. Source

OpenAI states that advertising shouldn't influence ChatGPT's answers. This statement is a platform commitment, but not evidence that ads are already being served in all countries or accounts. Source

According to reports, the test begins later in October 2026 in the US with a limited group of advertisers. No international launch date has been announced. Source

For self-employed professionals in the DACH region, this means: In October 2026, you can only plan a concrete campaign test seriously if OpenAI explicitly activates your country, account or partner programme. A US test isn't evidence of a bookable DACH campaign.

How visual ads work during image generation

The likely usage moment is an inspiration-oriented prompt where users seek a visual idea, such as: "Show me a modern terrace with seating area", "Create an image for a small bathroom" or "What could my shop look like after renovation?"

This isn't automatically a purchase enquiry. The user might just be orienting themselves, comparing options or gathering ideas. That's why the ad shouldn't just show a beautiful visual, but create a clear bridge to the next action.

Three visual types for local service providers

Three visual types are particularly relevant for local service providers:

• Before-after visual: Renovation, cleaning, garden, interior fit-out or room design.
• Application visual: A person using the service or seeing the result in everyday life.
• Result or trust visual: Finished project, local context, real staff or verified quality.

When an image does more harm than good

An image can be harmful when it's merely decorative and shows no concrete benefit, looks like a generic stock photo, raises expectations the offer can't fulfil, distracts from the user's actual concern, suggests an artificial or unrealistic before-after situation, or hides important information only in the image text.

The ad should therefore work visually on its own, but remain understandable without the image. Price, region, scope of service and next action belong in the ad text or on the landing page.

Which offers suit this format

Particularly suitable are offers with a visible result: painting and renovation work, garden and landscape design, interior architecture and kitchen planning, building cleaning, photography and videography, wedding and event services, beauty, hairdressing and wellness, and property-related services like home staging.

Less suitable are services whose benefit only becomes clear after lengthy consultation or that are barely visually representable, such as complex tax advice, abstract business consulting, purely technical maintenance contracts or legal advice without a clearly defined initial consultation.

For abstract services: stay concrete

For abstract services, the visual shouldn't try to explain the entire service. Better are a concrete problem, a clear first step, a visible situation or a limited entry offer.

Good entry offers for self-employed professionals include "Free 15-minute check", "On-site measurement", "Fixed-price offer for a clearly defined package", "Initial consultation with concrete outcome" or "Appointment within two working days".

A discount isn't essential. For local services, a clear scope of service can work more strongly than a blanket price reduction.

Three ad ideas for local service providers

Painting business

Visual: bright, realistically furnished room after renovation. Message: "From dark rooms to a home that fits you again." Offer: free colour and effort assessment in the service area. Landing page: regional references, service area, process, contact form. Measurement goal: qualified enquiry rather than mere image interaction.

Garden and landscape design

Visual: the same garden as a before-after representation; both states clearly labelled. Message: "Your garden. Thoughtfully planned, cleanly executed." Offer: planning appointment or on-site visit. Landing page: project examples by garden type, budget range and region. Measurement goal: enquiry with plot size and desired timeframe.

Local beauty or wellness studio

Visual: calm treatment situation or result representation without exaggerated promises. Message: "A break that fits your everyday life." Offer: initial appointment or clearly defined introductory package. Landing page: treatment, duration, price, location, availability and booking. Measurement goal: completed booking, not just visiting the appointment page.

What landing page you need for this

The landing page should confirm the ad within the first seconds: same service, same region, same promise, ideally same visual language.

What belongs on the landing page

A suitable landing page contains at least a clear headline, a concrete offer, the service area, one to three real references, price or price range (where possible), contact process, a visible call to action, phone, form or appointment booking, and privacy notice and imprint.

For local service providers, the form should contain few fields. Sensible are name, contact option, location or postcode, brief description of the project and desired timeframe.

What the landing page shouldn't be

The landing page shouldn't start with a general menu, many services and several competing calls to action. A single campaign goal makes measurement easier.

For visual ads, mobile optimisation is particularly important because the usage moment might occur in a mobile ChatGPT app.

Measurement: impressions, clicks and enquiries

Track separately at minimum: delivered ads, visible or measured impressions, clicks, click-through rate, landing page visits, form starts, submitted forms, phone clicks, booked appointments, qualified enquiries and actual conversions.

The central metric for self-employed professionals isn't the number of impressions, but the cost per qualified enquiry or cost per booked appointment.

Calculate important metrics

For a fair visual test, ideally change only one variable at a time: visual, headline, offer, landing page, audience or region.

Image interaction isn't purchase intent

Image interactions mustn't be equated with purchase intent. A user might click an inspiring visual without planning an order.

According to OpenAI, the platform has also worked on measurement and brand safety; however, concrete performance values or DACH benchmarks aren't published in the available sources. Source

A safe test with a small budget

For the DACH region, there's no verified public ChatGPT Ads price list and no confirmed minimum budget as of the cut-off date. Cost statements like "€5 per day", "€0.50 per click" or fixed minimum budgets shouldn't therefore be published as current ChatGPT facts.

A cautious test plan

A cautious test plan for a later-activated account: initially one offer, one region or a narrow catchment area, two visual variants, one landing page, one primary conversion, limited daily budget and a pre-defined stop rule.

For small businesses, a budget framework rather than an invented platform figure makes sense:

• Test budget = maximum amount whose loss is economically acceptable.
• Target cost per enquiry = amount at which an enquiry can still be processed profitably.
• Maximum test cost = Target cost per enquiry × desired number of qualified enquiries.

Example: If an order could allow at most €80 for acquiring a qualified enquiry based on contribution margin, and you want to test five enquiries initially, the internal ceiling would be €400. This is calculation logic, not a price published by OpenAI.

Set stop rules

Stop rules: no qualified enquiries after a pre-defined spending limit, many clicks but no relevant landing page actions, enquiries outside the catchment area, wrong expectation created by the visual, or cost per enquiry above the economically viable value.

Safety and brand management

Only use your own or licensed images, no unsubstantiated healing, quality or success promises, label before-after images truthfully, no customer images without consent, ensure accessibility and sufficient readability, and don't present impressions and clicks as revenue.

Common mistakes to avoid

• Premature availability assumption: The US test is confused with a DACH launch.
• False purchase intent: Inspiration during image generation is automatically interpreted as concrete demand.
• Decorative visual: The image is beautiful but shows neither problem nor result.
• Unclear region: The ad doesn't state where the service provider operates.
• Exaggerated visualisation: The image promises a result that isn't technically, legally or budgetarily realistic.
• Weak landing page: The landing page only contains a contact form, but no references, prices or process description.
• Too many variables: Visual, offer and landing page are changed simultaneously.
• Wrong metric: Clicks are optimised although qualified enquiries are the actual goal.
• No phone measurement: For local services, calls aren't tracked.
• Invented benchmarks: Unpublished CPM, CPC or conversion values are presented as ChatGPT average values.

What existing reports cover—and what's missing

The researched results predominantly cover these points: OpenAI's announcement of the visual format, launch during image generation, initially limited test in the US, target audience Free and Go users, clear labelling of ads, separation between ad and generated image, OpenAI's statement that advertising shouldn't influence answers, possible use cases like product inspiration, product use and service experiences, and planned or expanded measurement and brand safety features. Source

For self-employed professionals in the DACH region, the results particularly lack: a confirmed Germany, Austria or Switzerland launch, a public price list, CPC, CPM or CPA benchmarks, minimum budgets, concrete targeting options, available ad formats and creative specifications, reliable conversion data, legal guidance specifically for local service providers, examples with real DACH campaigns, separation of inspiration and purchase intent, and a small, economically justified test plan.

These gaps should be explicitly named in the article. The available sources document the platform announcement, but not the economic performance of the format for German self-employed professionals.

More on ChatGPT Ads with images: What self-employed professionals can test now and Measuring ChatGPT Ads: What self-employed professionals actually see now.

Frequently asked questions

Can I book ChatGPT Ads with visual formats as a German self-employed professional in October 2026?

According to available information, OpenAI is initially testing the format later in October in the US with a limited group. A DACH rollout hasn't been confirmed. Source

Does my ad appear in the AI-generated image?

No. OpenAI describes the ad as a separate display next to or outside the generated image. Source

Does an ad influence ChatGPT's answer?

OpenAI states that advertising shouldn't influence answers. This is a statement from the provider, not an independent impact study. Source

Are users who generate images automatically ready to buy?

No. Image generation can mean inspiration, planning or entertainment. Purchase intent must be measured separately through offer, click path and qualified enquiry.

Which images work best for local services?

Before-after, application and result visuals are suitable for testing. A reliable success ranking for ChatGPT Ads hasn't been published yet.

How much budget do I need?

For the DACH region, there's no verified public minimum budget or pricing information as of the cut-off date. Budget should be derived from maximum acceptable cost per qualified enquiry and expected contribution margin.

Which metric is more important: click-through rate or enquiries?

For a local service provider, qualified enquiries, booked appointments and cost per enquiry are more important than a high click-through rate alone.

Should I test an image or just text?

Yes, a comparison test makes sense if the account allows multiple variants: same offer, same landing page, different presentation. This lets you check whether the image actually generates additional qualified demand or just more attention.

What do I do if the ad generates many clicks but no enquiries?

First check landing page, regional fit, offer clarity and form. Then test the visual against a more factual variant; an inspiring image can generate attention but set the wrong expectation.

Can I transfer the announced US results to Germany?

No. User behaviour, availability, competition, purchasing power, language and delivery can differ. As of the cut-off date, there are also no published performance data that would justify such a transfer.

The step you can take now

You don't have to wait until ChatGPT Ads are available in Germany. You can build a landing page now that clearly shows your offer: region, service, process, price, references. You can photograph or have photographed a before-after visual. You can formulate your first entry offer and test whether it brings enquiries.

When the platform is then activated, you won't be the one just starting. You'll be the one who already knows what works.

More on ChatGPT Ads and advertising in AI assistants in our comprehensive guide.

If you don't want to do it alone or want to be faster, you can start with Nelija for free: Try Nelija free.