AI Search Monetization Is Splitting in Two. Brands Need a CPC Plan and a Checkout Plan.

On April 22, two different stories made the same point.

Search Engine Land reported that OpenAI has started testing cost-per-click ads inside ChatGPT, with early click pricing in the $3 to $5 range and CPMs reportedly sliding from about $60 to around $25 in some cases. A few hours earlier, the same publication reported that advertisers were already seeing a new ChatGPT Ads Manager, a sign that OpenAI is building real campaign infrastructure, not just a sponsored-response experiment.

That same day, Google Cloud and Ulta Beauty announced that Ulta products will become shoppable inside AI Mode in Search and the Gemini app, with recommendations, comparisons, and streamlined checkout happening inside Google's conversational surfaces. The release tied the rollout directly to Google's Universal Commerce Protocol, the open standard Google launched on Jan. 11, 2026 to support agentic commerce across discovery, buying, and post-purchase flows.

Most commentary will split those into separate buckets: ads news on one side, shopping news on the other. I think that misses the more important shift.

AI search is not converging on one monetization model. It is splitting into two.

One model monetizes intent through paid discovery. The other monetizes intent through in-flow transaction infrastructure.

That matters because brands now need more than a generic AI visibility strategy. They need to decide whether they are building for media efficiency, checkout capture, or both.

What changed this week

The OpenAI signal was not just that ads exist. We already knew that from the earlier ChatGPT ads rollout.

What changed on April 22 was the operating model.

Search Engine Land's coverage said OpenAI is moving beyond impression-led monetization and toward CPC pricing, with a limited Ads Manager that gives marketers more direct campaign control. That is a familiar search-business move. It turns ChatGPT from an experimental inventory source into something marketers can compare against Google Search, paid social, and retail media using performance math.

The Google signal was different.

The Ulta announcement did not pitch AI Mode as a new ad slot. It pitched Google's conversational surfaces as a place where a shopper can discover, compare, and complete checkout without leaving the AI workflow. Google's January post had already laid the groundwork with UCP, Business Agent, Direct Offers, and new Merchant Center attributes for conversational commerce. Ulta was the clearest named proof that Google is pushing that infrastructure into live retail behavior.

That is why this is bigger than two product updates.

The market is starting to separate into two monetization paths:

•CPC discovery
•Agentic checkout

The two AI search business models now taking shape

Dimension	CPC discovery model	Agentic checkout model
Current example	OpenAI's CPC ads in ChatGPT	Google's AI Mode and Gemini checkout flows powered by UCP
What gets monetized	The click	The transaction or assisted purchase path
Buyer intent moment	Evaluation and vendor discovery	Product selection and purchase completion
Primary operator	Paid media team	Commerce, product-feed, and growth teams
Core data requirement	Campaign targets, prompts, measurement, landing-page quality	Merchant feeds, product attributes, availability, offers, payment and checkout readiness
Success metric	CPC, CTR, CPA, pipeline influence	Checkout completion, revenue capture, assisted conversion, basket value
Main risk	Treating ChatGPT like just another ad network	Treating AI shopping like just another SEO content problem

That table is the useful visual, but it also exposes the deeper issue.

These models do not only change where money flows. They change which team owns the work.

A brand testing ChatGPT CPC ads needs some of the same inputs it would use in search advertising: prompt targeting, intent mapping, creative control, landing-page alignment, and post-click measurement. A retailer trying to win inside Google's agentic checkout model needs a different stack: feed quality, eligibility, product-answer coverage, offer logic, and a checkout flow that does not break when discovery happens inside AI surfaces.

That is why one generic "AI search lead" will often not be enough.

Why this matters for brands now

1. AI search budgets are about to fragment

Once OpenAI can sell AI intent on CPC terms, the budget conversation changes fast.

This stops being a speculative innovation line item and starts competing with channels that already have owners, targets, and performance benchmarks. Paid teams will want to know whether ChatGPT clicks convert better than Google Search. Finance teams will want proof that the platform is additive, not just a shiny place to burn test budget.

At the same time, Google's model will push another budget question: what happens if the monetizable moment moves closer to checkout and further away from the click?

That is not a media planning question alone. It becomes a commerce infrastructure question.

2. Merchant data is becoming part of AI visibility

We already made the broader case in our AI shopping analysis. The new wrinkle is that merchant data is no longer just a feed-management issue. It is part of how AI surfaces close the loop.

Google's January announcement said eligible retailers will be able to support checkout directly on Google's AI surfaces. It also introduced new Merchant Center attributes that go beyond classic keyword matching into product questions, accessories, substitutes, and richer conversational discovery.

That means the old separation between "SEO content," "shopping feeds," and "conversion path" is getting weaker.

If your product data is incomplete, your offers are stale, or your checkout logic is brittle, you are not just less visible. You are less transactable.

3. The measurement model is changing with it

DataForSEO's U.S. search-volume data, queried on Apr. 23, 2026, shows why this category deserves attention even before it fully settles. "google ai mode" is at 110,000 monthly searches, "gemini app" at 60,500, "chatgpt ads" at 6,600, and "agentic commerce" at 4,400.

The interest curve is already here. The measurement model is what is lagging.

A CPC program can be judged on familiar metrics. An agentic checkout program cannot. If a user discovers a product inside AI Mode, compares it there, and checks out through Google's flow, your team needs a way to connect merchant readiness, discovery prompts, offer visibility, and actual conversion outcomes.

A lot of brands do not have that stitched together yet.

The old way of organizing AI search work is starting to break

The easiest mistake right now is to treat all AI search monetization as one blended workstream.

It is not.

A good Google AI Mode optimization program may help you become citable. That does not automatically make you ready for in-flow checkout. A solid ChatGPT ads test may prove response-level demand, but it does not solve merchant data quality or transaction capture. And a sophisticated software stack on the measurement side still does not fix the operational handoff if paid, SEO, and commerce teams are all working from different assumptions.

This is also where the latest enterprise AEO infrastructure shift matters. The software market is moving toward more integrated AI visibility workflows at the exact moment the monetization side is getting more fragmented. Brands that do not clarify ownership will end up with better dashboards and messier execution.

What brands should do in the next 90 days

1. Split discovery readiness from transaction readiness

Treat them as separate audits.

Discovery readiness should include:

•prompt coverage for commercial and mid-funnel queries
•current presence on ChatGPT, AI Mode, Gemini, and other answer surfaces
•landing-page fit for paid and organic AI traffic
•performance baselines for test budgets

Transaction readiness should include:

•Merchant Center and feed completeness
•product-answer coverage for high-intent questions
•offer freshness and eligibility
•checkout and payment-path reliability
•SKU-level measurement for AI-assisted discovery

If you combine those into one vague AI search score, you will miss the actual blocker.

2. Decide which monetization model matters more by business type

Not every brand needs the same answer.

For B2B SaaS and lead-gen brands, the OpenAI path may matter first because the commercial event still happens after a click, a form fill, or a demo request. CPC discovery fits that journey more naturally.

For retail and commerce brands, Google's path may matter more because the commercial event is moving toward conversational selection plus checkout completion.

For hybrid businesses, both matter. Those teams need a shared view of prompts, products, landing pages, feeds, and conversion logic. That is a bigger operating challenge than most AI search decks admit.

3. Build reporting around decision stages, not channels alone

Most dashboards still answer a shallow question: "Did we show up?"

The better question is: where in the decision path did the AI surface influence the outcome?

That usually means splitting reporting into:

•discovery influence
•shortlist influence
•checkout influence
•post-click conversion influence

If you only measure AI search as a traffic source, you will undercount Google's model. If you only measure it as a citation or mention source, you will undercount OpenAI's paid model.

4. Prepare for overlap, not clean separation

The market will not stay tidy.

Google is still testing ads in AI Mode. OpenAI will almost certainly keep improving shopping and commercial workflows inside ChatGPT. Over time, the two models may borrow from each other.

But that does not weaken the current point. It strengthens it.

The brands that win will be the ones that can handle both paid discovery logic and transaction infrastructure logic without assuming one replaces the other.

Need to know whether your bigger AI search gap is paid discovery or transaction readiness?

Cite Solutions audits prompt-level visibility, landing-page fit, merchant data quality, and AI commerce readiness so you can invest in the model that will actually move pipeline or revenue.

Book an AI Visibility and Commerce Audit

FAQ

Is this just another ChatGPT ads story?

No. The ChatGPT ads update is one half of the evidence. The more important point is that Google's live checkout push with Ulta shows AI search monetization is no longer centered on one model. One path monetizes intent through clicks. The other monetizes intent through in-flow transaction capture.

Does agentic checkout matter if I am not an ecommerce brand?

Usually less, at least right now. If your primary conversion event is still a lead, a demo, or a sales conversation, CPC discovery models will likely matter sooner. But the broader lesson still matters: AI surfaces are moving closer to commercial action, not just content discovery.

Should paid media teams own AI search now?

Only part of it. Paid teams are the natural owners for CPC testing and campaign measurement. They are not usually the right owners for merchant-feed readiness, product-answer architecture, or checkout-path reliability. This is one of the clearest signals that AI search work now needs tighter coordination between media, SEO, and commerce operations.

What is the immediate operator takeaway?

Stop treating AI search as one generic line item. Separate discovery monetization from transaction monetization, assign clear owners, and build reporting that reflects where value is actually being created.

The bottom line

The AI search market just got easier to misunderstand.

OpenAI's April 22 CPC push makes ChatGPT look more like a performance media channel. Google's Ulta rollout makes AI Mode and Gemini look more like transaction infrastructure. Both are real. Both matter. And both ask for different operating muscles.

That is the shift brands should pay attention to.

The next stage of AI visibility will not be won by the teams with the most opinions about the future. It will be won by the teams that can tell the difference between a click business and a checkout business, then build for the one they actually need.

Framework