OpenAI just turned on CPC ads in ChatGPT. Should small ecommerce brands buy in?

{“type”:”text”,”text”:{“content”:”OpenAI flipped ChatGPT ads from impressions to cost-per-click this month. The minimum spend dropped from $250,000 to $50,000. Bids are landing at $3–$5 per click.\n\nThat’s a real change. It’s also not the headline a small ecommerce brand should chase.\n\nHere’s what actually shifted, and what to do with it.\n\n## What changed, and why\n\nThe original ChatGPT ad model was CPM-based. OpenAI charged a $60 CPM at launch in February. By April, average CPMs had eroded to $25 in some categories. CPM only works when you can sell the volume — and the volume hasn’t materialized at OpenAI’s preferred price. CPC is the natural pivot when impression demand is soft but advertiser interest is real.\n\nTranslation: OpenAI is repricing because the original price didn’t hold. That’s a normal part of a new ad surface maturing — but it’s also a signal that the channel is still figuring out its own value.\n\n## Who should pay attention\n\nCategories where users actively ask ChatGPT for a recommendation:\n\n- B2B software and services\n- Supplements and specialty health\n- Travel and lodging\n- Professional services (legal, financial, agencies, accounting)\n- High-consideration durable goods\n\nThese are users in active discovery mode. A clean answer from a sponsored result has a real chance of converting.\n\n## Who should skip it for now\n\nMid-funnel impulse-buy ecommerce: apparel, home goods, beauty, food. The ChatGPT user isn’t asking \”which $40 sweater should I buy\” — they’re asking \”what’s a good neutral wardrobe for a remote worker.\” That’s a content question, not a transaction. The intent gap will eat your CAC alive.\n\nSame goes for any product that wins at the bottom of the funnel on Meta or Google’s shopping surfaces. ChatGPT isn’t a substitute for those. It’s a different funnel stage entirely.\n\n## What $50,000 actually buys\n\nAt $3–$5 per click, $50,000 is 10,000–16,500 clicks. For a B2B service averaging 2% conversion to a qualified lead, that’s 200–330 leads. At a 10% close rate, 20–33 closed deals.\n\nWhether that pencils depends on your average customer LTV. We won’t run a ChatGPT pilot for any client whose modeled LTV is below ~$3,000 — the math doesn’t work below that line, even on the high end. The minimum is still a real barrier.\n\n## The bigger pattern\n\nEvery major AI lab is becoming an ad network. OpenAI just made the most public move; Perplexity, Anthropic-adjacent search wrappers, and Google’s AI Overviews are all monetizing answer surfaces in parallel.\n\nThe brands that win this cycle won’t be the ones who buy every new surface on launch day. They’ll be the ones who:\n\n1. Lock down branded search and retargeting first, on every channel that matters\n2. Pick one new AI surface per quarter to pilot at 5% of paid budget\n3. Have an attribution model they trust enough to actually kill underperforming spend\n\nWe’re watching ChatGPT closely. We’re piloting it for one B2B client this quarter. We’re not recommending it to our ecommerce clients yet.\n\nWhen that changes, you’ll see it here first — with the actual numbers.\n\nIf you’re trying to figure out where ChatGPT ads (or any new AI surface) fits in your stack, get a free analysis at “,”link”:null},”annotations”:{“bold”:false,”italic”:false,”strikethrough”:false,”underline”:false,”code”:false,”color”:”default”},”plain_text”:”OpenAI flipped ChatGPT ads from impressions to cost-per-click this month. The minimum spend dropped from $250,000 to $50,000. Bids are landing at $3–$5 per click.\n\nThat’s a real change. It’s also not the headline a small ecommerce brand should chase.\n\nHere’s what actually shifted, and what to do with it.\n\n## What changed, and why\n\nThe original ChatGPT ad model was CPM-based. OpenAI charged a $60 CPM at launch in February. By April, average CPMs had eroded to $25 in some categories. CPM only works when you can sell the volume — and the volume hasn’t materialized at OpenAI’s preferred price. CPC is the natural pivot when impression demand is soft but advertiser interest is real.\n\nTranslation: OpenAI is repricing because the original price didn’t hold. That’s a normal part of a new ad surface maturing — but it’s also a signal that the channel is still figuring out its own value.\n\n## Who should pay attention\n\nCategories where users actively ask ChatGPT for a recommendation:\n\n- B2B software and services\n- Supplements and specialty health\n- Travel and lodging\n- Professional services (legal, financial, agencies, accounting)\n- High-consideration durable goods\n\nThese are users in active discovery mode. A clean answer from a sponsored result has a real chance of converting.\n\n## Who should skip it for now\n\nMid-funnel impulse-buy ecommerce: apparel, home goods, beauty, food. The ChatGPT user isn’t asking \”which $40 sweater should I buy\” — they’re asking \”what’s a good neutral wardrobe for a remote worker.\” That’s a content question, not a transaction. The intent gap will eat your CAC alive.\n\nSame goes for any product that wins at the bottom of the funnel on Meta or Google’s shopping surfaces. ChatGPT isn’t a substitute for those. It’s a different funnel stage entirely.\n\n## What $50,000 actually buys\n\nAt $3–$5 per click, $50,000 is 10,000–16,500 clicks. For a B2B service averaging 2% conversion to a qualified lead, that’s 200–330 leads. At a 10% close rate, 20–33 closed deals.\n\nWhether that pencils depends on your average customer LTV. We won’t run a ChatGPT pilot for any client whose modeled LTV is below ~$3,000 — the math doesn’t work below that line, even on the high end. The minimum is still a real barrier.\n\n## The bigger pattern\n\nEvery major AI lab is becoming an ad network. OpenAI just made the most public move; Perplexity, Anthropic-adjacent search wrappers, and Google’s AI Overviews are all monetizing answer surfaces in parallel.\n\nThe brands that win this cycle won’t be the ones who buy every new surface on launch day. They’ll be the ones who:\n\n1. Lock down branded search and retargeting first, on every channel that matters\n2. Pick one new AI surface per quarter to pilot at 5% of paid budget\n3. Have an attribution model they trust enough to actually kill underperforming spend\n\nWe’re watching ChatGPT closely. We’re piloting it for one B2B client this quarter. We’re not recommending it to our ecommerce clients yet.\n\nWhen that changes, you’ll see it here first — with the actual numbers.\n\nIf you’re trying to figure out where ChatGPT ads (or any new AI surface) fits in your stack, get a free analysis at “,”href”:null}, {“type”:”text”,”text”:{“content”:”camisadomarketing.com”,”link”:{“url”:”http://camisadomarketing.com/”}},”annotations”:{“bold”:false,”italic”:false,”strikethrough”:false,”underline”:false,”code”:false,”color”:”default”},”plain_text”:”camisadomarketing.com”,”href”:”http://camisadomarketing.com/”}, {“type”:”text”,”text”:{“content”:”. We’ll tell you when it’s worth your money and when it isn’t.”,”link”:null},”annotations”:{“bold”:false,”italic”:false,”strikethrough”:false,”underline”:false,”code”:false,”color”:”default”},”plain_text”:”. We’ll tell you when it’s worth your money and when it isn’t.”,”href”:null}