ChatGPT Ads Conversion Rate Still Trails Google, Six Thousand Brands In

ChatGPT now handles roughly a fifth of the search queries Google does, yet its ad business is worth considerably less. Digiday has pulled together figures from several research firms and a working paper to check whether those ads pay off. The short version is that more brands are buying, but the clicks still don't convert as well as clicks from Google.

OpenAI only started selling ads this year, with a pilot that launched in February. The ad formats are still taking shape, and the company has been careful about how many ads it puts in front of users. Usage has run ahead of advertising, and ad buyers are already trying to find out how quickly OpenAI can close that gap. Keep that mismatch in mind as you read the numbers below.

The Digiday ChatGPT ads report in numbers

Start with reach, where data from Similarweb and Graphite shows the UK leads on the share of ad-eligible ChatGPT users who see ads. The US and Australia sit close behind, yet even in these established markets fewer than half of eligible users currently see ads. Mexico trails well behind those three, and Digiday notes the gap could reflect how far each market's rollout has gone.

The number of brands buying ChatGPT ads is rising fast. Adthena tracked 1,082 distinct brands running ads on ChatGPT across the US, Australia, the UK and the EU in June. That rose to 3,741 in July and 6,333 in August. Adthena's September count only runs to the 20th, but its projection points to a higher total again for the full month.

That's where the six thousand brands figure comes from, and the pool keeps growing as OpenAI adds markets and ad features. Volume isn't value, though, and the conversion data is where the trouble starts.

ChatGPT ads vs Google Ads on conversion rate and revenue

The clearest evidence comes from a working paper under review by Maximilian Kaiser and Christian Schulze, based on data from ecommerce websites. For US advertisers, paid Google Search visits converted at a much higher rate, more than twice that of visits from ChatGPT ads. The gap in revenue per session was similar, so each ChatGPT visit earned far less.

Before you treat those findings as the rule, note what the study covers. It looks at US ecommerce sites only, so a clinic in Dubai or a B2B firm selling software may see something quite different. It's also a working paper under review, covering ads that OpenAI is still building out. Even so, Digiday's reading is that OpenAI has yet to show its ads can compete with the performance channels advertisers already rely on.

The study doesn't explain the difference, so any reason you read elsewhere is a guess for now. It does tell you not to value a ChatGPT ad click like a search click until your own data agrees. In the meantime, plan any spend as a trial with a fixed end date rather than an always-on line in the media plan.

The two platforms also reach people at different moments. Search ads respond to a keyword, which often shows purchase intent, and they sit among the search results for that query. ChatGPT ads appear inside a conversation instead. That conversational context is new to most advertising teams, and habits built on traditional keyword ads may not all carry over.

Shopping's share of ChatGPT ad spend is shrinking

Spending on ChatGPT is widening beyond retail, according to Sensor Tower figures on US ad spend. Shopping's share fell sharply between April and September. Financial services grew its share many times over in the same months, from a small starting point. Software gained share as well, and travel and tourism nearly doubled its slice.

Sensor Tower only started counting spend in April, but it estimates that shopping brands placed the largest share of ads at launch. Jobs and education came next, followed by food and dining. Digiday reads the shift as ChatGPT drawing demand from a wider range of advertisers, rather than leaning on shopping moments alone. If you sell financial products, software or travel, companies in your category are already learning how ChatGPT ads behave in other markets.

Cost-per-click bidding and what a click is worth

OpenAI recently said cost-per-click (CPC) and outcome-optimised bidding account for most of its ad campaigns. That share is higher than New Street Research had assumed in its own model of the business. Pay-per-click ads usually earn more per ad event than impression-based pricing, so on paper it could lift OpenAI's revenue. New Street also cautions that monetisation may be weaker than it first modelled, and that its own CPC estimate could be too high.

For an advertiser, paying by the click shifts the question from what an impression costs to what a click is worth. A click that converts less often has to cost less to deliver the same acquisition cost. That's why the average CPC on its own tells you little about whether a campaign is working. The number to watch is what ChatGPT ads cost you per sale or qualified lead, measured in your own systems.

New Street models commercial chats on ChatGPT growing from 45.9 billion this year to nearly 380 billion by the end of the decade. Over the same period, it expects a larger share of total chats to carry ads. Those are forecasts, and more ad slots only help you if the clicks behind them turn into customers.

Outcome-optimised bidding, on any platform, depends on accurate conversion data to learn which clicks lead to results. If that feed is patchy, the bidding chases the wrong signals and the results it shows get harder to trust. So the problems covered next aren't a side note, because they shape the numbers you'll see.

Measurement and attribution are still unsettled

Performance is only one problem, according to Farhad Divecha, group CEO of Accuracast. He told Digiday that the latest integrations OpenAI has added, with Magnite and Pubmatic, haven't necessarily fixed the underlying tracking issues. "We continue to see discrepancies there," he said. He blamed some of it on teething problems with a new system advertisers must set up, and some on faulty conversion attribution.

Tracking gaps cut both ways, because they can hide sales that ChatGPT drove or credit it with sales won elsewhere. Until the way advertisers measure ChatGPT results settles, the reports in any ads manager shouldn't be your only source. Check them against first-party data from your own site and sales records before you move budget. When the platform and your own figures disagree, note both rather than picking the kinder one.

The picture for businesses in the UAE

None of the numbers above come from the UAE. The reach figures cover markets such as the UK and Mexico, while the conversion study looks only at US retail websites. So treat them as an early signal from other advertising markets, not a forecast for Dubai. Adnika is already running ChatGPT Ads campaigns for UAE businesses, so we read these findings with live accounts in mind.

The news touches any UAE business whose customers use ChatGPT to compare options before they buy. That includes online shops, clinics and B2B software firms, since each of them can win or lose a customer during that research. Businesses in the categories gaining share in the US spend figures have the most reason to watch closely. For small businesses, the main risk is spending on ads before they can measure the results.

A sensible way to test ChatGPT ads from Dubai

Don't move money out of campaigns that already work to pay for a ChatGPT trial, and don't write ChatGPT off either. A better path is a small, separate trial with its own budget and a clear rule for when to stop. Before any trial, confirm with OpenAI or your agency which markets your ads can target, since the rollout differs by country. Here's how to set one up so the result means something.

Decide how you'll track conversions before the first ad goes live, because a trial with broken tracking teaches you nothing. Pick one conversion event that matters to the business, such as a completed purchase or a booked consultation. Where the platform allows it, tag each ad URL with UTM parameters so visits show up clearly in your analytics. Then capture each lead in your CRM with its source attached, so you can follow it to a sale.

Next, send each click to a landing page that answers the question people ask ChatGPT before they buy. Someone who has been comparing options in a conversation wants the product they asked about rather than your homepage. To compare two versions of that page properly, use the method in our guide to A/B testing for higher conversions.

The ad itself should suit a chat where someone is weighing up what to buy. Name the product or service plainly, say what happens after the click, and skip the slogans you'd put on a billboard. Because the ads are still taking shape, it's worth running two versions side by side to see which one sells better.

Judge the trial on the cost of each sale or lead, and compare it with Google Ads, Meta ads or TikTok. Use matching 30-day windows so a slow week or a seasonal spike doesn't skew the comparison. Count click-based results first, and be wary of view-through conversions while reporting is this unsettled. If you want help with that setup, our team offers funnel and conversion analytics.

If you sell leads rather than products, the study maps less neatly onto your business. A clinic or a property developer cares about booked appointments and viewings, and those can take weeks to turn into revenue. In that case, follow each ChatGPT lead for longer than the first month before you decide. A cheap lead that never books is worth less than a pricier one that does, so count only high-quality leads.

Some businesses should wait before they spend anything on ChatGPT ads. If you can't yet tie a lead to a sale, fix that first, because a trial would only add numbers you can't check. The same goes for a business whose current campaigns don't pay their way, because a new platform won't fix a weak offer. If the problem is the page rather than the traffic, fix the page before you buy more clicks.

Decide your stop rule before you start, while you can still be objective about it. If the trial can't reach a cost per sale you'd accept from another channel, pause it. Look again later, since the reach figures and the integrations are both still changing.

Reporting decides whether a trial teaches you anything, since newer AI advertising platforms count conversions their own way, as older ones do. The Adnika Dashboard: one live view of a client's ads, analytics and CRM data, with AI recommendations reviewed by the strategist. For a client trialling ads on ChatGPT, that means the trial's leads sit next to those from every other source. Weak sales can't hide behind a good-looking click count.

Buying ad placements isn't the only way to show up inside ChatGPT. The AI-generated answers around those ads often draw on web content, including pages from businesses like yours. Clear pages about your products and services can build AI visibility in those answers, with no cost per click attached. It's worth planning the paid and organic sides together, so your ad copy and your pages describe the business in the same terms.

Signals worth watching from here

Reach is the first thing to watch, because fewer than half of eligible users see ads even in the most established markets. If that share climbs, more ChatGPT users will meet ads in the chats where they research a purchase. The second is whether the new Magnite and Pubmatic integrations settle the discrepancies Divecha describes. The third is conversion evidence from outside the US, since the study here only covers American retail sites.

Run ChatGPT ads alongside the channels that work

ChatGPT ads are growing fast in brand count, and the spend is spreading across more categories. What the data doesn't show yet is ad traffic that converts like search, or reporting without discrepancies. For most Dubai businesses, that points to a measured trial that runs alongside their current campaigns rather than replacing them. Set the goal and the review date before the first click, then hold yourself to both.

If you'd like a team that already runs these campaigns to plan the trial, talk to our ChatGPT Ads agency in Dubai. If ChatGPT is becoming part of how your customers find you, plan the next step with the team at our AI agency in Dubai.