The short answer
For B2B marketers weighing ChatGPT Ads against Google Ads budget in 2026.
- 1Google: the user asks, and must click to get the answer. The click is the only route forward, so intent and action arrive together.
- 2ChatGPT: the user asks and already has the answer. Your ad has to persuade them to abandon a conversation that is currently working.
- 3So richer contextual intent does not automatically mean higher click or purchase intent. It can mean the opposite.
- 4Google Ads still wins for capture on high-intent commercial terms. ChatGPT Ads is closer to demand shaping inside the research moment.
- 5Treat it as an additional line item tested against CRM-recorded pipeline, not a replacement for search budget.
The argument
The pitch for ChatGPT Ads is that the intent signal beats a keyword. Someone types four sentences about their situation instead of three words, and the platform matches your offer against the lot. All true, and it does sound like it should hammer search.
It skips a step. Intent is two things, and Google kept them welded together: how relevant your offer is to the situation, and how likely someone is to act on it in the next ten seconds. On Google they came as a pair, because clicking was the only way forward. In ChatGPT they've come apart, because the conversation is already giving the person what they came for.
Richer contextual intent doesn't automatically mean higher click intent. It may mean the opposite, because the reason to click has just been satisfied.
So the question worth asking isn't whether the intent beats Google. It's what an ad has to be worth before someone abandons a conversation that's currently answering their question. That's a far harder brief than a search ad, and I've yet to see many people writing to it.
The mechanical difference
The results page is a menu, not an answer. Clicking is the only route to the thing they came for, so your ad competes against nine other routes to the same destination. Winning the click is a ranking problem.
The answer is already in front of them and the next question is one line away. Your ad competes against staying put. Winning the click is a persuasion problem.
Three things follow from that, and you can see all three in the numbers people are publishing.
- Click-through rate means something different. On search, a low CTR usually means poor relevance. Here it can mean perfect relevance and an answer that was good enough already.
- The clicks you do get are self-selected differently. Someone leaving a working conversation has a reason. That reason might be strong buying intent, or it might be idle curiosity about the shiny thing in the sidebar, and you can't tell which from the platform.
- Landing page has to continue a conversation, not start one. They arrive mid-thought, several exchanges deep into a problem. A generic homepage is a hard reset and they will bounce in seconds.
What the early evidence shows
The published results are split down the middle. I don't think one camp is lying. I think the split is telling you something.
| Source | What they reported |
|---|---|
| Search Engine Land / StubGroup | Roughly 1,500 comparable users each. Google Ads: 41 seconds average engagement, 1.13 engaged sessions per user, 3.71% conversion. ChatGPT Ads: 17 seconds, 0.88, 0.21%. |
| Choice OMG, via MediaPost | About $415 spend, 8,940 impressions, around 60 clicks, 0.6% CTR, roughly $7 CPC, no interested buyers. |
| InterTeam Marketing | 150 plus qualified leads across three B2B SaaS clients, cost per lead around 60% below Google, roughly $5 average CPC. Agency authored, so treat with the usual caution. |
| North Country Growth | Best results came from high intent B2B audiences, with cost per lead competitive against Google Search in some verticals but limited volume. |
Line those four up and a pattern shows. Where the offer is something a conversation physically can't do for you, the channel produces decent leads at a fair price and not much volume. Where the model has basically already sorted the person out, you buy cheap impressions, thin engagement and nothing at the end of it.
The side-by-side
| Dimension | Google Ads | ChatGPT Ads |
|---|---|---|
| Targeting unit | Keywords and match types | Context hints, semantic and non-binding |
| Diagnostics | Search terms report, negatives, auction insights | None on the conversational side |
| Reachable audience | Everyone using search | Free and Go users only |
| Click economics | Highly variable, competitive B2B terms are expensive | Around two to five dollars observed, plus a £15 daily minimum in the UK |
| Buying stage reached | Mostly active, in-market searching | Often problem aware, earlier than search |
| Why they click | To get the answer | To get something the answer can't provide |
| Volume available | Whatever the category supports | Limited, particularly in niche B2B |
| Maturity | Twenty five years of practice | Months, changing weekly |
Not sure whether your buyers are even reachable on ChatGPT?
Book a 30 minute call. We'll look at your organic ChatGPT referrals and your buyer's likely plan tier and tell you straight.
When ChatGPT Ads genuinely make sense
The case for this channel was never cheaper clicks. It's demand you can't otherwise buy: catching someone while they're still working out what's wrong, before they've formed the search query you'd have bid on.
- You already get organic ChatGPT referrals. The single clearest signal. People are asking about your category and being pointed at you already.
- Your offer is something a conversation can't do. Look at my account. Audit my setup. Do the work. Take responsibility for the outcome.
- High consideration, high value. The economics have to survive three to eight pound clicks and a long cycle.
- Your buyer plausibly sits on a free tier. Smaller firms, individual practitioners, people researching outside a locked down corporate account.
- Google is capped. If you're already capturing every relevant search and want incremental demand rather than more of the same, this is where it might be.
Take our own business. The buyer is a marketing lead sat there asking why their LinkedIn campaigns produce clicks and no pipeline. Plenty of them will never type "LinkedIn ads agency" into Google at all. Getting in front of the earlier question is demand search simply doesn't have for sale.
When they don't
- Your decision makers are all on paid plans. Enterprise IT, large corporate procurement, anyone whose employer has bought Business seats. Delivery will look fine and reach nobody who signs.
- Your Google account isn't maxed out. If there's unclaimed impression share on terms you know convert, buy that first. It's cheaper learning.
- Your tracking is shaky. On a channel with no search terms report, weak measurement leaves you with no information at all.
- Low order value. The click prices and the daily minimum will eat you.
- You need volume this quarter. Even the positive reports describe limited volume. This is a test line, not a pipeline plan.
How to split the budget
What we do, and what we tell clients to do: leave Google and LinkedIn doing their jobs, and carve a fixed test off the top rather than shaving a percentage out of the middle.
£750 to £1,500 over thirty days buys a reading. A rolling ten percent of budget buys a permanent unexplained line item.
You need the comparison. Cost per qualified lead on ChatGPT means nothing without the same number from your existing channels over the same window.
Not CTR, not CPC, not raw leads. A cheaper click that produces a worse lead is a more expensive customer.
Scale, extend once with a sharper hypothesis, or stop. No sunk cost theatre and no fourth month of hoping.
Read next
Common questions
Want a straight answer on whether this channel fits your business?
- A suitability read: organic ChatGPT referrals, buyer plan tier, offer fit and click price tolerance.
- A defined thirty day test built alongside your existing channels rather than out of them.
- Account, Pixel, CAPI, UTMs and CRM source capture so the comparison is like for like.
- Cost per qualified lead compared against Google and LinkedIn over the same window.
- A scale, extend or stop recommendation with the reasoning shown.
Starter £1,250/mo, Growth £1,725/mo, Scale £4,025+/mo. For current clients we cover setup and management of the first test and you fund the media. We're learning this channel in public and don't think you should pay a premium for that.