ChatGPT Ads and Context Engineering: The Setup That Decides Your Cost Per Lead
Quick answer
ChatGPT ads are matched to the meaning of a conversation, not to a keyword list, and the auction weighs relevance as well as price. Context engineering means giving the system clear, consistent facts about who you serve, what you sell and which questions you answer, then matching your ad, landing page and tracking to those questions. Done first, it lets a lower bid win better placements and lowers cost per lead.
On this page
- Why context beats bid in ChatGPT ads
- What context engineering means for an advertiser
- Step 1: build the fact base first
- Step 2: map buyer questions to ad groups
- Step 3: write creative for very small spaces
- Step 4: make the landing page answer the question
- Step 5: track leads, not clicks
- Common mistakes we see
- How OUNTERNET helps
- Frequently asked questions
- Sources
Why context beats bid in ChatGPT ads
On Google you buy a keyword. In ChatGPT you buy a moment in a conversation. Ads Manager uses conversational context hints instead of keyword lists, and reviews show that the auction is relevance weighted and second price. In plain English: a well matched ad can beat a higher bid that fits the conversation badly.
That changes where the money is made. Most advertisers start by tuning bids. The bigger lever sits earlier: how clearly the system can tell which conversations your business belongs in, and how well your ad and page answer the question being asked.
What context engineering means for an advertiser
Context engineering is the practice of shaping the information an AI system works with so its output is accurate and useful. For an advertiser it has three layers.
- Your public fact base. Your website, Google Business Profile, LinkedIn and directory listings should describe the same business in the same words: who you serve, where, what you sell, prices if you publish them, and proof.
- Your campaign context. The context hints, audience notes and ad groups you set in Ads Manager. Each ad group should map to one buyer question, not to a product catalogue.
- Your landing context. The page the ad opens. It must answer the exact question from the conversation in the first screen, and its facts must match the ad.
When these three agree, the system has an easy decision. When they disagree, you pay for that confusion in weaker placements and lower conversion rates.
Step 1: build the fact base first
Before you open Ads Manager, write a one page fact sheet and make your site match it.
- One sentence on what you do and for whom, used everywhere.
- Your service areas and the markets you do not serve.
- Three to five proof points with dates: case studies, certifications, years trading.
- The ten questions buyers ask before they contact you, each answered on a page.
- Organisation, Service and FAQ schema on those pages.
This is the same work that gets you named in organic AI answers. It is covered in our AEO vs GEO guide. Paid and organic draw on one set of facts, so you build it once.
Step 2: map buyer questions to ad groups
Group conversations by intent, not by product. A dental clinic might run three ad groups: "teeth whitening cost", "emergency dentist near me" and "implants vs bridges". An industrial supplier might run "how to measure hydrocarbon dew point" and "analyser for natural gas pipelines".
- QuestionA buyer asks ChatGPT a specific question
- MatchYour ad group's context fits that question
- AdA short title and line that echo the question
- PageThe first screen answers it and shows proof
- ActionBook, call or request a quote, tracked as a conversion
Keep each ad group narrow enough that one landing page can answer it completely.
Step 3: write creative for very small spaces
The ad unit is small. Reported guidance is a title of about 16 characters and a description of about 32, plus your name, favicon, an image and the link. There is no room for slogans. Use the buyer's own words.
- Title: the outcome or the service, for example "Implants in Poole".
- Line: one concrete reason to click, for example "Fixed price plan, free scan".
- Image: the real product, place or team. Stock photos look out of place under a thoughtful answer.
Write three variants per ad group and let the data pick.
Step 4: make the landing page answer the question
The person arriving from ChatGPT has just read a careful answer. A generic homepage feels like a step backwards. Send them to a page that continues the conversation: the answer in two sentences, the proof, the price range if you can share it, and one clear action.
Two technical checks matter. Landing pages must allow OpenAI's ad review crawler, OAI-AdsBot, or the ad will not be approved. And the page must load fast on a phone, because most ChatGPT use is mobile.
Step 5: track leads, not clicks
Ads Manager supports a JavaScript pixel and a Conversions API, and conversion optimised campaigns need an active conversion event. Set up both, send the same lead event you use for Google Ads, and pass a value where you can.
Reporting is aggregated. You will not see individual conversations, so your CRM is the source of truth. Compare ChatGPT leads with Google leads on booked calls and closed revenue, not on click-through rate.
Common mistakes we see
- Running one ad group for the whole business.
- Sending every click to the homepage.
- Different prices or service lists on the site, the ad and Google Business Profile.
- Judging the channel on two weeks of clicks instead of a month of leads.
- Raising bids to fix a relevance problem.
How OUNTERNET helps
We build the fact base, the ad groups, the landing pages and the tracking as one system, and report on booked calls. Our ChatGPT and AI Ads service starts from £3,800 a month. Ad spend is paid to the platform. Prices exclude VAT. Final price after we analyse your business and scope. If you want to see how assistants describe you today, start with the free AI visibility check or apply for a call.
Frequently asked questions
Do ChatGPT ads use keywords?
No. Ads are matched to the context of the conversation. You describe the situations your ad fits, and the system decides where it is relevant.
Can ads change what ChatGPT says about my business?
No. OpenAI states that ads do not influence answers. The answer depends on the information available about you, which is why context engineering matters for paid and organic.
How short does ChatGPT ad copy need to be?
Reported guidance is a title of about 16 characters and a description of about 32 characters. Write in the buyer's words and keep one idea per ad.
Which ChatGPT users see ads?
Ads are shown to logged in users on the Free and Go plans in supported countries. Plus, Pro, Business and Enterprise users do not see them.
What tracking do I need?
Install the ChatGPT ads pixel or the Conversions API and send a lead event. Conversion optimised campaigns need an active conversion event.