CPC Arrives in ChatGPT, Google Bets the Stack on Agents, and Meta's Crossover Week

OpenAI has quietly flipped ChatGPT's ad model from CPM to CPC, with bids reported at $3–$5 per click after CPM rates collapsed from roughly $60 to as low as $25 in ten weeks. Google used Cloud Next to rename Vertex AI the Gemini Enterprise Agent Platform, fold in Agentspace, and commit a $750m partner fund. Meta reports Q1 on 29 April, with eMarketer forecasting it will overtake Google in global digital ad revenue for the first time in 2026. And Shopify's Agentic Storefronts have quietly plugged millions of merchants into ChatGPT, Copilot and Google AI Mode — reshaping where product discovery happens.

We unpack what is confirmed, what is still hype, and what a sharp operator should actually do this week: where to test, where to hold, and how to brief a client on Monday.

Transcript

SPEAKER_00 0:01

Welcome to the Daily Marketing Brief. Your daily AI news and tactics for marketers who move fast. I'm your host Jen Bryan and here's today's update. The story that ties today's news together is not a model release or a keynote headline. It is all about pricing. How attention gets priced, how agents get priced, how compute gets priced, when OpenAI quietly shifts ChatGPT ads from CPM to CPC, when Google rewires its enterprise stack around agents, when Meta is days away from overtaking Google in ad revenue. These are underneath decisions about who captures the margin when AI reshapes a workflow. That is the thread we'll pull on today. In this episode, the CPC switch inside ChatGPT and what it actually tells us about demand. Google's consolidation of its enterprise AI platform at Cloud Next, and why agencies should pay attention even if they never touch vertex. Meta's crossover moment with Google and what next week's Q1 print is really a test of. Shopify's quiet repositioning of discovery, a short watch list, and five practical things I'd actually do this week. ChatGPT goes CPC. What happened? On the 21st of April, Digiday and Search Engine Land reported that OpenAI has turned on cost per click bidding inside ChatGPT's ads pilot, with bids sitting between$3 and$5 per click, verified via screenshots from the ads manager. Skift confirmed the shift the next day. This sits alongside reporting that CPMs, which launched at around$60, have compressed to as low as$25 in the 10 weeks since the February go live. OpenAI has begun recruiting its first marketing science lead and is building a conversion tracking tool. What is confirmed? The CPC option exists in the ads manager. The$3 to$5 bid range is verified by DigiDay screenshots. The CPM slide from$60 to$25, which is huge, is reported by multiple outlets citing advertising sources. OpenAI's public revenue target for ads in 2026 is$2.5 billion, and that has been circulating since the internal plan leak. Why does it matter? For two reasons. First, CPC is the pricing model performance marketers actually understand and can plan against. CPM on an unproven surface is a brand budget experiment. CPC with conversion tracking is something a media buyer can defend in a QBR. Second, the CPM collapse is a real signal. A 58% drop in 10 weeks says demand at launch pricing was not sticky. And what that means for business is if you sell a considered purchase, SaaS, finance, travel, high AOV e-commerce, a$3 to$5 click inside a context where the user is already in research mode on paper is attractive. The question is whether ChatGPT can match the conversion signal and audience depth of Google or Meta. It almost certainly can't yet, but the cost of finding out is dropping week by week. What it means for marketers or agencies is that you can start modeling ChatGPT as a small carved out test line. Two to five percent of search or social budget, not more, with a clear hypothesis about incrementality. Pressure OpenAI or your agency rep or for any access to the conversion tracking tool as it rolls out, and resist the temptation to replatform budget based on novelty. The measurement story is still immature. My take, read the CPC switch as a concession, not a breakthrough. OpenAI needs advertisers to come back for a second insertion order and the first cohort did not at CPM. CPC lowers the psychological barrier and aligns the risk. It is a rational move. It is also evidence that the we'll reinvent advertising framing is giving way to we'll use the pricing model that works. Confidence level high on facts, medium on demand interpretation because OpenAI has not confirmed CPM figures, high confidence that performance advertisers should test within a small carve out. Google has consolidated the agent layer. What happened? At CloudNetz on the 22nd of April, Google rebranded Vertex AI as the Gemini Enterprise Agent Platform and folded agent space into a single product. It launched Agent Studio, a low code builder, a simulation environment for pre-deployment testing, a governance registry and a third-party agent marketplace that includes Adobe, Atlassian, Salesforce, ServiceNow, Workday, Oracle, Delight, and others. Alongside this, Google committed a$750 million fund to its partner ecosystem to accelerate agentic deployments. Separately, Bloomberg and TechCrunch reported Google deepened its deal with Thinking Machines Lab and rolled out a new generation of custom TPUs. What is confirmed? The rebrand, the platform consolidation, the partner list and the$750 million figure are in Google's own press release from April 22nd. The TPU launch is confirmed by Fortune and Google. So why does this matter? For three years Google's enterprise AI story has been fragmented Vertex, Duet, Gemini, Agent Space, Workspace AI, buyers complained it was hard to know what they were buying. Consolidating under Gemini's Enterprise is less a product announcement and more of a sales motion fix. It also signals where Google is placing its competitive weight against OpenAI's enterprise push and anthropics momentum is encoding. And what it means for business, if your company is already on Google's workspace or GCP, expect an aggressive push over the next two quarters to buy Gemini enterprise seats and deploy pre-built agents from the marketplace. The procurement conversation will get much harder to ignore for SaaS vendors in the Gemini enterprise marketplace, distribution is the prize. And what that means for marketers and agencies, three things. One, there is now a more defensible architecture for agent-built internal workflows, media planning, assistance, writing brief agents, QA checkers, without stitching together 10 tools. Two, Workspace Studio gives ops and finance teams a cleaner path to automate reporting, which will put pressure on agency retainers that lean on manual reporting time. Three agencies building internal IP on top of Gemini should consider listing in the marketplace while the partner push is active. My take Google's real advantage here is not the platform itself which is competent rather than extraordinary. It is distribution. So Google already sits inside most enterprise IT stacks and a unified agent platform with 750 million of partner incentives is how Google converts presence into share. OpenAI and Anthropic will have to answer stronger enterprise go to market. On the confidence level, high on facts, medium high on the strategic read, it remains to be seen whether the marketplace attracts genuine demand or becomes another underused catalog. Meta's crossover week what happened Meta reported Q1 on 29th of April. eMarketers updated the forecast has Meta overtaking Google in total digital ad revenue in 2026 at 243.5 billion versus Google's 239.5 billion. Advantage Plus reached a 60 billion annualized run rate in late 2025 tripling from 20 billion in March 2025. Instagram Reels was roughly at 50 billion annualized ad rate at the end of 2025 and Meta has guided Q1 revenue at 53.5 billion five billion with 2026 capex between 115 billion and 135 billion. What is confirmed? The eMarketer forecast, the advantage plus run rate, the reels figure and the capex range are all on the record. The crossover is a forecast not yet a reported result. So why it matters if eMarketer is right, 2026 is the first year Meta sells more advertising than Google globally. That is a generational shift in where media budget flows. Advantage Plus is the primary driver and it is essentially a campaign structure that hands more control into Meta's models. For performance marketers this is the reality of AI and advertising not chatbots but silent automation of targeting bidding and creative what this means for business if you're still splitting paid social into granular ad sets with type manual targeting, you are fighting the platform. Advertisers who have restructured into Advantage Plus last year have broadly seen CPA improvements. The upside though concentrates more of your performance inside a singular model which you can't really interrogate. But what it means for marketers and agencies is three shifts. One creative volume and variety now matter more than targeting precision. The model does the latter you feed it the former two measurement has to move up the funnel in platform attribution is increasingly meta story not yours. And three agencies pricing on optimization hours should revisit their model. The work is increasingly brief feed creative and incrementality testing not bid adjustments. My take Meta overtaking Google is notable but the CapEx question is the one that matters for the 29th of April Meta is spending more than 100 billion to justify an ad business that is already working. If Q1 revenue comes in at or above the high end and CapEx stays within guidance the stock will take it. If CapEx drifts and ad growth softens the AI ROI skepticism returns fast. Confidence level medium high on the ad revenue direction medium on the capex payment which is genuinely uncertain until we see a few more quarters. Discovery leaves the SERP okay what happened Shopify's agentix storefronts activated for eligible US merchants in late March now route products from roughly 5.6 million stores into ChatGPT, Microsoft Copilot, Gemini and Google's AI mode with no additional transaction fees and no per channel integration. In the same window, Amazon has rolled out Autoby inside Rufus which can authorize purchases on behalf of the user when a target price is hit. Shopify also released Tinker, a mobile AI creative app and extended native B2B features to all plans at no extra cost. Separately, Search Engine Journal reports ChatGPT user crawler requests are now 3.6x Googlebots and ChatGPT accounts for around 20% of search related traffic worldwide. What is confirmed? The Shopify launch, merchant count, zero additional fees and channel lists come from Shopify's own newsroom. The Rufus Auto buy feature is confirmed by Amazon. The crawler and search traffic figures come from all AI via search engine journal which is a single source treated as you know directional rather than definitive. So why does it matter? Two independent shifts are happening at once. On the supply side merchants can now list into AI surfaces without extra work. On the demand side a meaningful slice of product discovery and purchase intent is moving out of the classic SERP and into assistance. The commercial plumbing of e-commerce is being reorganized under most operators' feet. What this means for business so if you are a D2C or e-commerce operator on Shopify, your product catalog is likely already surfacing in ChatGPT whether you enabled it consciously or not. The quality of your product data title structured attributes imagery inventory accuracy is now a ranking factor inside AI channels. Schema hygiene is no longer just an SEO job it's an agentic commerce job. What that means for marketers or agencies, audit what your client's catalog looks like in ChatGPT and Google AI mode today. Not hypothetically actually run the searches, identify the feed issues, brand disambiguation problems, and missing attributes that keep the product from being chosen. This is a new distribution audit and few agencies are yet pricing it even as a service. My take Agentic commerce will not replace paid social or search for most of the categories this year, but the mix of discovery is shifting in a way that rewards operators with a clean data infrastructure and penalizes those relying on a single channel. The quiet story is that catalog quality, inventory feed health and returns policy clarity are now commercial motes. Unsexy, but that is usually where the money is confidence level high on the Shopify and Amazon facts, medium on the scale of near-term sales impact and we have momentum data, not yet audited sales attribution data. What's on the watch list for this week Anthropic Compute and Revenue. Anthropic disclosed a run rate of near 30 billion and signed a multi-gigawatt TPU deal with Google and Broadcom with capacity starting in 2027 per the 6th to 7th April announcements and subsequent Broadcom SEC filing referencing 3.5 gigawatts. If true this makes Anthropic one of the largest single consumers of next gen compute. Worth watching because it underwrites pricing disciplines at Anthropic and because it affects how cloud capacity is allocated in the product most agencies use daily. OpenAI, marketing science hire and conversion tracking tool a single hire may sound minor but this is the scaffolding that turns ChatGPT ads from a branded content placement into a real performance channel. Watch for the first measurement product announcement that is when a proper test is most justifiable. So what matters most from today? The pattern across today's stories is a shift from AI will change advertising to AI is quietly changing where the margin sits. OpenAI is conceding pricing to advertisers Google is consolidating its enterprise stack to convert distribution into revenue and meta is the current proof that AI and advertising can move to the top line when it sits inside an existing ad system. Shopify is moving discovery into AI channels without taking a toll anthropic is securing the compute to keep supplying the picks and the shovels. The signal is commercial infrastructure the noise is still model benchmarks and keynote choreography. If a story does not eventually translate into a line item on a PL, the ad spend, a seat license, a take rate, a compute contract, just treat it as background noise. This is what I would do running an account brand or agency this week. Carve out a small chat GPT ads test, 2-5% of monthly search budget, pick one considered purchase product with a clear offer clarity, set CPC bids in the$3 to$5 range and design it as an incrementality read, not a ROAS contest. Do not repurpose social creative write for a research mode user. Audit your catalog inside AI surfaces. Run 20 real buyer queries into ChatGPT, Gemini and Google AI mode for your top SKUs. Log where you appear how you are described and what your competitors beat you on. That document is the brief for your next feed and schema print. Prep a Q1 meta earnings read through pull your advantage plus share of spend, CPA and creative refresh rate. When Meta reports on the 29th of April you want to know whether your account mirrors or lags the platform. If it lags, investigate creative volume and feed quality before targeting. Pressure test one agency workflow with an agent. Pick one media planning brief, weekly performance report or QA checklist. Give it to Gemini Enterprise Claude or ChatGPT team. Measure time saved versus quality do not promise clients anything yet. This is an internal productivity and shouldn't yet be a billable line. Do not reallocate meaningful budget based on this week's news. The measurement infrastructure for AI channels just not there yet. Small discipline tests not replatforming. These are what I'm telling clients today. ChatGPT is not Google yet but the pricing is finally sane we should run a disciplined test and treat it as a learning not a scale. The quality of your product data inside AI surfaces is now a commercial test. Schema and feed work is overdue investment not housekeeping. And Meta's Q1 print next week is the real AI advertising benchmark. If Advantage Plus keeps delivering platform level automation is the default and we need to plan our creative operation accordingly. The cleanest way to read this week is that AI industry is maturing from what can the models do to what do the contracts say. Pricing models, platform consolidation, capex discipline and commerce taker rates are now the center of the story. And that is good news for operators because those are the parts you can actually start planning against. For marketers start testing ChatGPT ads at CPC with a small carve out and incrementality frame do not over rotate. For agencies add AI surface audits to your new business and renewal conversations. Catalog visibility in ChatGPT Gemini and AI mode is now new for the SEO audits. For e-commerce brands schema feed quality inventory accuracy and returns clarity are ranking factors in agentic commerce. Budget for the unsexy work for founders if you are building on Gemini Enterprise apply for marketplace and partner fund while the distribution push is active cheap attention windows closed. For operators Meta's 29th of April earnings is the most important single data point next week. Plan the Monday conversation in advance that's been today's episode of the Daily Marketing Brief. I'm your host Jen Bryan and see you tomorrow