AI Ads Management for Brand and Marketing Teams: The 2026 Playbook for ChatGPT Ads, Conversational Targeting and Measurement

AI ads management is no longer a 2027 story. ChatGPT Ads Manager is live, contextual ad units are showing up under model answers, and brand and marketing teams are being asked to build playbooks for a channel that did not exist 12 months ago. The teams winning early are not the ones with the biggest budgets — they are the ones treating AI ads as a distinct discipline, with its own targeting model, creative logic, measurement stack and governance. This piece is the strategic playbook we run with brand and marketing leaders across France, the UK and the US who want to move first without burning their CFO’s patience.

By Toni Dos Santos, Co-Founder, Spicy Advisory.

What “AI Ads Management” Actually Means in 2026

AI ads management is the discipline of planning, buying, creating and measuring advertising that runs inside AI-native surfaces — conversational assistants, AI search engines, AI overviews and embedded copilots — using AI-driven targeting and AI-driven measurement. It overlaps with paid search and paid social but is not the same thing. The placements live inside an answer experience, the targeting is contextual rather than profile-based, and the creative competes with a fully satisfying response from the model itself.

For brand and marketing teams, three things matter:

• The surface is new. ChatGPT Ads Manager opens up sponsored cards under model answers for Free and Go-tier users in the US, Canada, Australia and New Zealand, with paid tiers staying ad-free.
• The targeting model is new. Instead of keyword lists or third-party cookies, advertisers write short natural-language “context hints” that describe the conversations where their offer is relevant.
• The economics are new. CPMs from the early pilot in the US$60 range have been joined by CPC bidding in the US$3–$5 starting range, with cost-per-action models in development — which makes AI ads a credible performance channel, not just a brand awareness play.

Most of what your team already knows about Google Ads and Meta Ads transfers. The campaign / ad group / ad hierarchy is the same. The reporting tables look familiar. The break is in where attention is captured, how intent is matched, and what the creative needs to do.

ChatGPT Ads Manager: The New Contextual Surface

What it is

ChatGPT Ads Manager is OpenAI’s self-serve buying platform for sponsored placements inside ChatGPT. Advertisers create campaigns, set budgets and bids, target by conversational context, and view performance in a familiar dashboard at ads.openai.com. The beta is live for US advertisers of all sizes, with inventory reaching Free and Go-tier users across the US, Canada, Australia and New Zealand. Plus, Pro, Business and Education subscribers stay ad-free.

The minimum spend has dropped from initial pilot levels of US$200–$250k down to roughly US$50k, and the introduction of CPC bidding alongside CPM has explicitly opened the channel to performance-oriented advertisers, not just brand-led media buys.

What an AI ad looks like

ChatGPT ads are card-style units — advertiser name, favicon, headline, short description, image, landing URL — rendered below the model’s answer and clearly labelled as Sponsored. They are not woven into the response. The model answers the user’s question first; the ad sits underneath as a clickable next step.

That format choice is structurally important. The ad has to earn the click against a complete, neutral, often very good answer. Generic brand creative will lose. Specific, intent-aligned creative that frames itself as the “next step to actually do this” will win.

Who sees AI ads, and who does not

• In: Adult users on ChatGPT Free and Go in approved markets.
• Out: Plus, Pro, Business and Education subscribers; users identified as under 18; conversations involving health, mental health, politics and other sensitive categories.

For regulated verticals, this is not just a brand-safety footnote — it is a structural inventory constraint. Healthcare, financial advice, political advocacy and youth-facing brands need to plan around it from day one.

How Conversational Targeting Actually Works

Context hints, not keywords

The targeting unit inside ChatGPT Ads Manager is the context hint: a short natural-language description, sitting at the ad-group level, that tells the system which kinds of conversations should trigger your ad. Examples that work in practice:

• “B2B sales and marketing teams looking to automate outbound email and LinkedIn at SMB scale.”
• “HR leaders building AI training programmes for non-technical employees.”
• “Families planning summer trips to European cities on a budget.”

The system combines context hints with your ad copy and your landing page content to decide when to surface the ad. Advertisers cannot retrieve transcripts, target a specific conversation ID, or build user-level audiences. They optimise via aggregated metrics — impressions, clicks, CTR, average CPC, average CPM, conversions — not via raw user data.

Bidding: CPM, CPC and (soon) CPA

• Reach + CPM: Buy on impressions for category presence around high-volume informational queries. Useful for brand lift and new product launches.
• Clicks + CPC: Buy on clicks, starting around US$3–$5 in a second-price auction weighted by relevance. The right default for most performance-oriented brands.
• CPA (in development): Cost-per-action bidding has been signalled as in motion. Expect this to become the default for direct-response advertisers once it ships.

Measurement: pixel, Conversions API, UTMs

The early ChatGPT ad pilot was rightly criticised for thin attribution. The May 2026 update closed the biggest gap with a Conversions API and pixel that let advertisers send back purchases, sign-ups and lead events, and read aggregated conversion metrics in the Ads Manager. Add UTM parameters on every landing URL and pipe the traffic into your existing analytics stack to compare against search and paid social.

Important: reporting is aggregated. You will not see user-level paths. Plan your measurement around channel-level efficiency (CPC, CTR, CAC, ROAS) plus periodic incrementality tests — not deterministic last-click attribution.

Why This Is a Brand and Marketing Problem, Not a Performance Marketing Problem

The instinct in most organisations will be to dump AI ads management into the paid media team and walk away. That is the wrong move. Three reasons:

1. Creative quality dominates. When the placement sits below a complete answer, the marginal click goes to whoever frames the most specific, useful next step in the user’s language. That is a brand, narrative and product marketing problem before it is a bidding problem.

2. Context hints are positioning in disguise. Writing a context hint that wins is the same exercise as writing a category positioning statement: who is this for, what job are they trying to do, what makes us the obvious answer. Performance marketers are not always the right authors for that work.

3. AI ads sit inside the same trust contract as GEO and AEO. Users who reach an AI assistant are partly there to escape the noise of traditional advertising. A misaligned ad damages trust in both the assistant and the brand. AI ads, GEO citations and AEO presence have to tell one coherent story — and that is owned by brand, not by media buying.

If you have not yet made GEO part of your operating cadence, start with our GEO playbook for brands and CMOs and our CMO playbook for AI-driven marketing operations. AI ads management plugs into both.

Where AI Ads Fit in the Brand and Marketing Funnel

AI ads are a high-intent surface that behaves more like search than social. Users have already articulated a problem in natural language — “help me choose a CRM for a 12-person sales team,” “build a 4-day Lisbon itinerary for a family of four,” “how do I run a paid LinkedIn campaign on a £2k budget.” That changes where AI ads earn their keep:

• High-consideration research: SaaS, financial products, education, healthcare, travel — categories where users compare options and want a recommendation.
• Workflow and how-to queries: Tools that help automate or simplify the exact task the user is mid-stream on.
• Niche, context-rich categories: Offerings that benefit from explanation, where a concise answer plus a targeted ad shortens discovery dramatically.

AI ads are weaker for impulse-driven, very-low-AOV products and for categories where ChatGPT’s organic answer already converges on a small set of incumbents. In those cases, GEO investment to be cited inside the answer is more efficient than buying a card underneath it.

Creative and Messaging That Wins Inside AI Conversations

The constraint is brutal: roughly 16 characters of headline, 32 characters of description, one image (logo-only is discouraged), one URL. The creative job:

• Hook off the conversation intent. Mirror the user’s outcome language, not your tagline. “Launch outbound sequences in hours” beats “The leading AI sales platform.”
• Position as the next step, not the first step. Frame the ad as “the way to actually do what you just learned” — reinforcing the answer rather than overriding it.
• Use imagery that signals the use case. Product UI, real scenarios, before/after — anything that lets the user pattern-match to their task in under a second.
• Match the landing page to the conversation. If the ad promises an outcome, the landing page must continue that outcome in the first 50 words. ChatGPT users will bounce hard from generic homepages.

Treat headlines, descriptions and context hints as testable hypotheses. Run multiple ad groups per campaign, each tied to a different conversational intent inside the same category, and let the system learn which combinations earn attention.

Measurement and the GEO/AEO Connection

AI ads, GEO (generative engine optimisation) and AEO (answer engine optimisation) share an operating logic: be the answer the model wants to surface, and be the most useful next step underneath it.

• GEO wins you citation share inside the model’s organic answer.
• AI ads win you the sponsored next step under the answer when the user is in your category.
• AEO ensures the entry-point question itself surfaces your brand.

Run them as a single programme, not three. Practically, that means:

1. Build a shared map of high-value buyer questions in your category (the same questions feed GEO content briefs, AEO FAQ schema, and AI ad context hints).
2. Standardise UTM conventions so ChatGPT-sourced traffic, AI-search-sourced traffic and traditional channels are comparable in your analytics tool.
3. Instrument the OpenAI pixel / Conversions API at the same time as your GA4 / server-side conversion stack — do not let AI ads launch without conversion plumbing.
4. For larger brands, fold AI ads into MMM or geo-incrementality tests to understand contribution against search and paid social.

Risks, Brand Safety and Governance

What can go wrong

• Adjacency risk. Context hints guide but do not control matching. Your ad can appear next to topics you did not anticipate. Monitor weekly during the first 60 days.
• Trust erosion. Users come to AI assistants partly to escape advertising noise. Aggressive or misaligned creative damages both the assistant and the brand.
• Measurement opacity. Aggregated reporting and a black-box matching model mean you cannot fully reverse-engineer performance. Plan for hypothesis-driven testing, not log-level analysis.
• Regulatory drift. Generative AI advertising rules are still being written, especially in the EU and UK. Disclosure, personalisation and data-sharing rules can shift mid-campaign.

What to put in writing before you launch

• Approved categories and topics for AI ad placement.
• A short list of context hints the brand will and will not use.
• Creative review checklist: factual accuracy, brand voice, regulated-claims compliance, alignment with on-site experience.
• Escalation path when an ad surfaces in an unintended context.
• Data-sharing posture: what conversion data you send back to the platform, what you do not.

The 90-Day AI Ads Management Rollout for Brand and Marketing Teams

Use this as a starting frame, not a finished plan — adapt to your category, geography and risk appetite.

Days 1–30: Foundation

• Map the top 20 buyer questions in your category that real customers ask AI assistants today. Pull from sales calls, support tickets, search query reports and your own ChatGPT usage.
• Stand up the OpenAI pixel and Conversions API. Define one primary on-site goal (demo, trial, lead, purchase) and one secondary goal.
• Build a UTM convention that distinguishes ChatGPT ads from AI-search organic and from traditional channels.
• Document brand-safety rules, approved categories and creative review checklist.

Days 31–60: First Campaigns

• Launch 1 campaign with 2–3 tightly scoped ad groups. Each ad group: one specific conversational intent, 2–3 context hints, 3–5 creative variants.
• Default to Clicks + CPC at the lower end of recommended bids. Reserve Reach + CPM for clear brand-lift use cases.
• Track impressions, CTR, CPC and CAC daily for the first two weeks. Cull underperforming context hints fast.
• Pair each campaign with a landing page that continues the conversational logic in its first 50 words.

Days 61–90: Scale and Integrate

• Expand to 2–3 additional intents based on which conversational territories actually convert.
• Integrate AI ads reporting into your weekly marketing dashboard alongside search and paid social.
• Run an incrementality test (geo split or staggered launch) to estimate true contribution.
• Sync the learnings back into your GEO content roadmap — the highest-converting context hints almost always indicate a content gap that is worth filling organically too.

“The marketing teams that win with AI ads are the ones treating the channel as a brand, narrative and measurement programme — not as a new line in the paid media spreadsheet.” — Toni Dos Santos, Co-Founder, Spicy Advisory

Organisational Implications: Who Owns AI Ads Management?

AI ads management forces collaboration that most marketing orgs are not yet structured for:

• Brand owns the narrative, the context-hint language, the trust posture and the “next step” framing.
• Performance owns the bidding model, the measurement stack and the daily optimisation loop.
• Product / CX owns the post-click experience, ensuring the landing page continues the conversation rather than restarting it.
• Legal / governance owns regulated claims, data-sharing posture and brand-safety escalation.

If any of these four sit outside the launch group, expect drag. The teams that move fastest are the ones with a single accountable owner — usually a senior brand or growth marketer — and a small standing group across the four functions.

What to Watch Over the Next 12–24 Months

• CPA bidding and deeper third-party measurement integrations — both will materially shift performance economics.
• Geographic expansion beyond the initial four markets, plus inventory growth as logged-out users come into scope.
• Regulatory action, especially in the EU and UK, on disclosure, personalisation and data sharing in AI ad contexts.
• Competitive AI ad surfaces from Google (AI Overviews ads), Meta’s generative ad products, and Anthropic’s evolving stance, which will fragment budgets and force a true cross-AI media plan.

The brands that invest now in understanding conversational user behaviour, that build internal context-hint libraries, and that integrate AI ads with GEO and AEO will be the ones with a real moat once the channel becomes a must-buy.