OpenAI Codex's White-Collar Pivot: When 20% of Users Aren't Developers

As of June 5, 2026, OpenAI has opened ChatGPT Ads Manager to more than 1,000 advertiser accounts — the commercial launch of what may be the most consequential new advertising surface since Instagram began monetizing in 2013. Early access brands are reporting average CPMs between $25 and $60, CPC rates of $3 to $5, and conversion rates that Criteo analysis puts at roughly 2x the comparable Google Search rate for high-intent commercial queries. OpenAI has stated a $2.5 billion advertising revenue target for 2026, with longer-term projections reaching $11 billion by 2027 and $100 billion by the end of the decade — ambitions that would make its ad business larger than all but the top three platforms in global digital advertising.

The launch completes a shift OpenAI began signaling eighteen months ago: ChatGPT is no longer purely a query-response tool. It is a discovery surface with strong commercial intent, high user dwell time, and documented purchasing behavior. The activation challenge for growth teams is not whether ChatGPT Ads works — early data suggests it does, at least in specific categories. The challenge is how to structure campaigns on a platform whose engagement model, attribution logic, and bidding mechanics differ meaningfully from every other paid channel most marketers have built expertise on.

Signal's analysis of ChatGPT's retention dynamics showed that the platform's core user cohort — heavy, returning users who engage with ChatGPT for research, planning, and decision support — has substantially higher session depth than casual users who try it once. These are the users advertisers are reaching: engaged, research-oriented, and further along the consideration funnel than a cold search query implies.

The Attention Economics of Conversational AI

Paid advertising fundamentally prices attention. The question that determines whether a new ad channel justifies premium rates is whether the attention it sells is better than existing options — more qualified, more receptive, more likely to convert.

ChatGPT's attention economics are structurally different from search and social in three ways that matter for conversion rates.

Intent depth at session start. A Google search query is typically transactional: "best project management tool for small teams" or "compare HubSpot vs Salesforce." The user signals commercial intent in a one-time query, then processes results and either clicks or abandons. A ChatGPT session for the same decision-making task often involves extended dialogue — the user explains their team structure, asks follow-up questions about specific features, explores pricing scenarios, and builds a contextual understanding of their options over multiple turns. By the time the user encounters a sponsored result, they have already committed significant cognitive effort to the decision. This invested-in-the-conversation state creates a different kind of receptivity than the transactional click model.

Temporal ownership. Social media ad impressions compete with content the user actively chose to consume — a post from a friend, a video that autostarted. The ad is intrusion. In ChatGPT, the entire session is intent-driven: the user came specifically to get an answer or explore a topic. Sponsored results that are genuinely relevant to the query are experienced as helpful additions to the answer rather than interruptions. The IAB definition of contextual advertising — placement that directly relates to user intent — fits ChatGPT ad delivery more cleanly than almost any other surface.

Absence of competitive saturation. Google Search advertising is among the most competitive auctions in marketing history. For commercial B2B keywords, CPCs routinely exceed $50 on terms where first-mover competition from established brands has driven prices to levels where only large budgets can compete. ChatGPT Ads is launching into an environment where most advertisers have not yet built campaigns, category competition at auction is low, and CPM rates reflect a premium surface at pre-saturation pricing. The early mover advantage in new ad platforms — the period before competitive bidding drives CPMs to equilibrium — has historically been one of the highest-return windows in paid media.

The Rate Card: CPM, CPC, and Conversion Bidding

OpenAI is launching with three buying models, mirroring the standard options on Google Ads and Meta but calibrated to ChatGPT's different engagement pattern.

The CPM range of $25–$60 positions ChatGPT inventory as premium but not unprecedented — it overlaps with LinkedIn's B2B CPM range ($30–$80) and sits above Meta's standard social feed ($8–$15) but below high-intent Google Search rates for competitive terms ($60–$200+ CPMs for commercial queries when calculated against equivalent impression depth). Brands familiar with LinkedIn's premium pricing rationale — paying more per impression for a more qualified audience — will find the ChatGPT CPM logic familiar.

The CPC range of $3–$5 is significantly lower than comparable Google Search CPCs for commercial intent keywords, where $10–$30 per click is standard in B2B SaaS and $50+ is common in financial services and legal categories. The lower CPC reflects two factors: ChatGPT's ad auction is less mature and less competitive than Google's, and the ChatGPT click-through model differs from search — users typically click through from sponsored answers rather than clicking to initiate a search journey. The click that happens at the end of a ChatGPT research session carries different intent signals than a first-touch search click.

CPA bidding is live but early. OpenAI is offering conversion optimization against downstream events — form fills, purchases, trial signups — but the conversion signal feed relies on advertiser pixel data that many brands have not yet connected to the platform. Brands that invest in proper conversion tracking setup from day one will gain attribution data that latecomers will lack.

The 2x Conversion Claim: What the Data Actually Shows

Criteo's analysis of early ChatGPT Ads performance — drawing from a sample of its managed brand portfolio — put average conversion rates at roughly 2x comparable Google Search performance. This number has circulated widely in growth marketing channels this week, and it needs careful interpretation before it drives budget decisions.

The 2x figure applies to high-intent commercial queries in categories where ChatGPT has strong organic engagement: software tools, consumer electronics, business services, and financial products. These are categories where users genuinely engage with ChatGPT for research and decision-making, meaning the user who sees a sponsored answer has already self-selected into a research-active posture for that category. The conversion advantage is real in these contexts.

The figure does not apply equally across all categories. Impulse-driven consumer categories where social advertising traditionally excels — apparel, food delivery, entertainment — showed smaller differentials in early data. The ChatGPT session dynamic favors considered purchases over impulse buys; the user who spends twelve minutes discussing project management software options is a different buyer profile from the user who taps "buy now" after three seconds on an Instagram post.

Attribution methodology also affects the comparison. Some of the "2x conversion" measurement compares last-click ChatGPT conversions to last-click search conversions. ChatGPT's position in the modern research journey is often middle-of-funnel: users discover options in AI search, do comparison research in ChatGPT, then convert through a paid search click or direct navigation. Last-click attribution systematically undercounts ChatGPT's role in conversions that complete elsewhere. Brands building measurement frameworks for ChatGPT Ads should build multi-touch models rather than relying on last-click.

The more defensible claim: for software, B2B services, and high-ticket consumer decisions in categories where ChatGPT has high organic engagement, early data shows conversion rates meaningfully above Google Search averages. The mechanism is plausible — session depth creates more informed consideration — and the early performance data supports it in specific contexts.

Creative Format Guide: What Works in Conversational AI

ChatGPT Ads does not display banner ads, video spots, or display placements. It surfaces three primary creative formats, each with different performance characteristics for different campaign objectives.

Sponsored Answers are the primary format. When a user asks a query relevant to a sponsored category, the response includes a clearly marked "Sponsored" answer block that directly addresses the question using the advertiser's product or service. Unlike a search ad that directs to a landing page, the sponsored answer delivers value within the ChatGPT session itself — answering the user's question with information about the advertiser's offering. The content is assembled by OpenAI's ad serving system from the advertiser brief, with model-generated elaboration appropriate to the specific query.

Creative best practices for Sponsored Answers: lead with a concrete answer to the question, not a product pitch — users are asking questions, and ads that answer first and pitch second outperform ads that pitch first. Match the technical depth of the query: a question from a developer about API rate limiting needs a different answer depth than a question from a business user about integration pricing. Include a specific, deliverable outcome claim — "reduce deployment time by 40%" outperforms "improve your workflow."

Sponsored Context (in beta) is a deeper integration where the advertiser's product or service becomes part of ChatGPT's response architecture for relevant category queries. Rather than a discrete sponsored block, the brand appears as a recommended solution in multi-turn conversation. This format requires creative assets designed for multi-turn elaboration rather than single-impression messaging.

Sponsored Links are the most familiar format — a clearly labeled recommended link appearing at the end of a relevant response. Functionally similar to a search ad but triggered by conversational context rather than query keywords. CPC-priced and the lowest-friction entry point for brands testing the platform.

The Attribution Gap Nobody Has Solved Yet

Measurement is the most underweighted challenge in ChatGPT Ads, and most brands launching this week are not yet equipped to handle it properly.

The standard pixel-and-last-click attribution model assumes a linear path: ad impression → click → conversion. ChatGPT's role in the purchase journey is rarely linear. Typical patterns in categories where ChatGPT Ads performs well include: a user who receives a sponsored answer but does not click, returns to Google the next day, searches the brand name directly, and converts through branded search — with zero attribution credit to ChatGPT. Or a user who encounters a sponsored answer, continues multi-turn research comparing options, clicks through to a competitor rather than the advertiser — but the advertiser's sponsored answer shaped the comparison framework without receiving credit.

The OpenAI ads platform provides impression and click data but does not yet integrate with the downstream CRM or conversion data that would enable full-funnel attribution. Brands that invest in setting up proper tracking — first-party data connections, server-side event tracking, and multi-touch attribution models that include AI-assisted discovery touchpoints — before they scale spend will have fundamentally better measurement than competitors who treat ChatGPT like another last-click search channel.

The attribution gap is a feature of early ad platform evolution, not a permanent flaw. Meta built its advertiser base before multi-touch attribution was standard, and the brands that figured out early how to measure Facebook's role in the consideration funnel outperformed those who waited for the platform to solve measurement for them.

The First-Mover Playbook: Six Steps to Your First ChatGPT Campaign

The brands extracting the most from early access follow a setup pattern that differs from standard paid search or paid social. For growth teams launching their first ChatGPT campaigns this month:

1. Define the intent context, not just the keyword. ChatGPT Ads targeting is query-contextual, not keyword-matching. Before building creative, map the conversational contexts where your category comes up: what problem is the user solving, what comparison are they making, what information do they need to decide? The targeting brief should describe user intent, not keyword lists.

2. Write for the session, not the impression. Sponsored Answer creative should deliver genuine value in response to the query context. A brand sponsoring the answer to "what's the best way to manage B2B sales pipeline?" should provide a legitimately useful answer about pipeline management that positions its product — not a product pitch disguised as an answer. Users who get value from sponsored content engage; users who feel sold to abandon the session.

3. Set up first-party conversion tracking before you scale. Connect your conversion events — trial signups, form fills, purchases — to the ChatGPT Ads platform before your first campaign. The brands that build conversion tracking infrastructure now will have data quality advantages over competitors who layer tracking in after the fact.

4. Run CPM and CPC formats in parallel for the first 30 days. CPM campaigns establish impression baseline and brand awareness measurement; CPC campaigns test click-through and landing page conversion. Running both simultaneously provides enough data to understand where your category sits on the CPM-to-CPC optimization curve before committing budget to one model.

5. Build multi-touch attribution reporting before evaluating ROI. If your attribution model is last-click only, you will systematically undervalue ChatGPT's contribution to conversions that complete in other channels. Map the post-impression conversion journey using first-party data: what is the conversion rate for users exposed to a ChatGPT Ads impression in the 7 days before converting through search or direct?

6. Test three creative variants per ad group and optimize weekly. Sponsored Answer creative testing is faster-feedback than traditional display — engagement patterns are visible within 7–10 days at sufficient volume. Test on: answer depth (short versus detailed), value claim specificity (vague versus concrete), and CTA framing (explore versus try versus book).

Who Is Already Running and What They're Learning

The 1,000+ brands in early access skew toward B2B SaaS, consumer electronics, financial services, and professional services — categories where ChatGPT has the deepest organic user engagement and where the research-to-purchase journey is long enough to benefit from early-funnel insertion.

Early signals from the growth marketing community suggest B2B SaaS brands are reporting user intent quality on ChatGPT Ads that rivals high-intent Google Search for mid-funnel queries. A project management tool targeting "how to structure OKRs for a remote team" is reaching a user at a highly specific decision moment, not just a job-title demographic.

Consumer electronics brands find that ChatGPT's comparison-query contexts perform particularly well. Users asking "should I buy a MacBook or ThinkPad for software development?" are in a state of explicit active consideration that almost no other channel can identify reliably at the impression moment.

Financial services brands — insurance, fintech, investment products — are navigating OpenAI's category-specific advertising policies carefully. Not all financial categories are open for advertising in the initial launch, and regulated categories require compliance documentation. Brands in these categories should verify category access before building creative infrastructure.

Klarna's AI-first marketing transformation illustrates the broader strategic question ChatGPT Ads opens: when AI is both the discovery surface and the creative execution layer, how do brands build differentiated positioning that does not collapse to commoditized generated answers? Brands that invest in distinctive value propositions — specific claims, proprietary data, measurable product advantages — will outperform brands relying on awareness-based creative that any competitor can replicate.

The paid ad growth machine behind Temu's expansion offers a counterexample worth studying: Temu's success on performance channels relied on volume-based CPM buying, aggressive creative testing, and attribution optimization. On ChatGPT, that playbook inverts — the channel rewards relevance and answer quality over volume and creative velocity.

The Long Game: What This Means for Growth Teams in H2 2026

OpenAI's for-profit restructuring and valuation ambitions require substantial and growing revenue from commercial products. Advertising is the most proven monetization model for large user bases — and ChatGPT's 500+ million users represent an inventory pool that can support advertising revenue at Google-scale if the platform builds the advertiser tooling, measurement infrastructure, and category policies to support it.

eMarketer projects global digital advertising spend at $680 billion in 2026, with search and social commanding 72% of the market. ChatGPT Ads is unlikely to disrupt either channel's dominance in the short term — but even capturing 3% of the global digital ad market would put OpenAI at $20 billion in ad revenue, validating the platform as a major advertising business. The US programmatic market has already registered early impact: CPMs across premium inventory rose approximately 34% in the past quarter as advertiser attention shifts toward AI-native surfaces and away from commoditized banner inventory.

The strategic implication for growth teams is not "move all paid budget to ChatGPT." The smart recommendation for H2 2026 is to run it as a 10–20% budget experiment alongside existing paid search and social, with proper measurement infrastructure, treating it as a first-party data collection opportunity as much as a conversion channel. The brands that understand ChatGPT's role in their specific customers' research journey — measured rigorously with multi-touch attribution — will enter 2027 with a data advantage over brands that waited to see how the platform matures.

Takeaway: ChatGPT Ads Manager's commercial launch creates a first-mover opportunity in an auction environment that will not stay this uncrowded for long. The conversion advantage is real in high-intent commercial categories where ChatGPT has strong organic research engagement — but realizing it requires creative built for conversational context, conversion tracking established before scaling, and attribution models that capture ChatGPT's middle-funnel role rather than evaluating it on last-click alone. The brands that invest in measurement infrastructure now will have a durable data advantage over those who treat it as another performance channel.