ChatGPT Ads UK 2026: Costs, Targeting & How to Win

Ads inside ChatGPT went from a leaked line of code to a live, self-serve advertising platform in under a year, and as of August 2026 they’re running in the UK. Most of what’s been written about them is either hype or a press-release summary. This is the working version: what ChatGPT Ads are, what they cost in pounds, how the auction actually decides who wins, and the step-by-step playbook we use to turn AI conversations into leads and sales.

Short answer: ChatGPT Ads are sponsored placements shown below ChatGPT’s answers to Free and Go users, bought through OpenAI Ads Manager Beta on a CPM, CPC or oCPC basis. OpenAI recommends starting bids of $3–$5 per click (roughly £2–£4), the UK minimum daily budget is £15, and there is no minimum contract. You don’t bid on keywords: you describe buying situations with plain-English context hints. The advertisers winning right now are the ones whose hints, ad copy and landing page all match the exact situation the user is describing in the chat.

What are ChatGPT Ads? (And what they are not)

ChatGPT Ads are sponsored placements that appear directly beneath a ChatGPT answer. When someone on the Free or Go plan asks a question and the conversation is commercially relevant, OpenAI’s ad system can show one clearly labelled ad unit under the response. The answer itself is generated independently and isn’t changed by the ad, and advertisers never see the conversation: you get aggregate reporting only.

Every unit is built from the same six parts, which matters because relevance is scored across all of them:

Global PPC Sponsored

£950 fixed-fee PPC audit

Find where your ad spend is leaking. Full account review by a Google Partner agency.

globalppc.org/paid-ppc-audit

The anatomy of a ChatGPT ad: advertiser name, favicon, title, copy, image and landing page. The system reads all six (plus your context hints) when deciding whether your ad fits a conversation.

Who sees ChatGPT ads

• Free and Go plan users who are logged in and over 18. In the UK, ads have been appearing for these users since the June 2026 pilot, with the full launch confirmed by OpenAI on 11 August 2026.
• Not Plus, Pro, Business, Enterprise or Education subscribers. That matters for B2B: the professionals most likely to pay for ChatGPT are the least likely to see your ad.
• Not anyone OpenAI identifies or predicts to be under 18, and not in sensitive contexts. OpenAI keeps ads away from conversations about personal health, mental health, financial distress and politics.

The ad formats available in 2026

• Standard conversational ad. The single sponsored unit shown above, matched to the topic and intent of the live conversation. This is the workhorse for lead generation and services.
• Product feed ads. Launched June 2026 for retailers. You upload a catalogue (up to two million products via SFTP) and OpenAI generates product-level ads from it. OpenAI has described feed ads as among the strongest-performing ads in the programme so far.
• Multi-product carousel. Several products from one advertiser in a single unit. In testing as of August 2026, not general release, but it’s where ecommerce is heading.

Why advertise on ChatGPT in 2026?

The case for ChatGPT advertising rests on two numbers: the size of the audience and the quality of its intent. The audience part is easy. OpenAI reported 900 million weekly active users in February 2026, with the 1 billion mark reported in August, and users send around 2.5 billion prompts a day. In the UK, the IAB counted 24 million people using AI tools in January 2026, with roughly three-quarters of that time spent on ChatGPT, the large majority on the free tier that now carries ads.

The intent part is what makes this channel different from a display network with a big audience. People don’t open ChatGPT to scroll. They type a specific problem (“which boiler cover is worth it for a 12-year-old combi?”, “compare cloud accounting software for a VAT-registered sole trader”), and they research, compare and often decide inside a single session. An ad that appears under that answer is landing at the exact moment of evaluation, which is the moment Google Search has monetised for twenty years.

Three early signals from the US pilot back that up. The US ad business was reported to have passed $100 million in annualised revenue within six weeks of launch, with 600+ advertisers in the first pilot phase. Ad load was deliberately light, reportedly around one in five free-tier queries, so inventory is scarce rather than saturated. And early advertisers consistently describe traffic that is new to their site and further along in its decision than social traffic. Early click-through rates sit below Google Search, but the users who click are problem-aware and solution-ready, and that changes what a click is worth.

Map each ad group to one stage. Mixing research-stage hints with a checkout landing page is the most common reason early ChatGPT campaigns look like they “don’t convert”.

The rollout timeline: how ChatGPT Ads reached the UK

Understanding the sequence matters because most articles you’ll find quote a stage of the platform that no longer exists. The $200,000 minimum, the fixed $60 CPM and the invite-only access were all real, and all gone within three months.

How ChatGPT Ads targeting and the auction actually work

This is the part most guides skip, and it’s the part that decides whether your budget converts. OpenAI’s system picks an ad for a conversation based on expected relevance and expected outcomes, using several signals at once:

• The context and intent of the live conversation. Not just the last message: the topic and direction of the whole chat.
• Your ad’s landing page, title and copy. OpenAI crawls your landing page (with its OAI-AdsBot user agent) and reads it as a targeting signal. A vague page is a vague signal.
• Your context hints. Plain-language descriptions, set at ad-group level, of what you offer, who it helps and when it’s useful. These guide matching but are not exact-match controls and don’t guarantee delivery in any specific conversation.
• Select signals from the user’s wider ChatGPT experience, such as memory and past chats, but only when ad personalisation is enabled. In the UK and EU that requires explicit opt-in, so assume UK delivery is mostly contextual for now.

What you can and can’t target

There is no demographic targeting, no interest categories, no lookalike audiences and no keyword bidding. What you do control:

• Location. Country targeting everywhere; in the US, state, DMA and ZIP level; regional, city and postcode targeting is being added market by market, so check the location picker for UK availability before you plan a local campaign.
• Platform. Choose iOS app, Android app and/or web (desktop and mobile web). Useful if your landing page or lead form performs very differently on mobile.
• Custom audiences. Your own lists can be included or excluded at campaign level, which is how you keep existing customers out of an acquisition campaign.
• Context hints and creative. In practice, these are your real targeting. The tighter the hints and the copy, the tighter the conversations you appear in.

The relevance-weighted, second-price auction

When several ads are eligible for the same conversation, OpenAI runs a relevance-weighted, second-price auction. Both halves matter. “Second-price” means the winner pays just above the next-highest eligible bid, not their own maximum, so bidding your true value doesn’t automatically mean paying it. “Relevance-weighted” means your bid is multiplied by how well your ad fits the conversation, so a sharper, better-matched ad can beat a higher bid and pay less for the placement. If you’ve run Google Ads, this is Quality Score with the volume turned up: relevance isn’t a discount on the side, it is the auction.

Objectives, bidding and campaign structure in OpenAI Ads Manager

Every account follows a three-level hierarchy that will feel familiar if you’ve used Google or Meta, with one twist at the middle layer.

1. Campaign. Sets the objective, budget, dates, countries, platforms and any custom audiences to include or exclude. Objectives can’t be changed after creation; if you need a different one, clone the campaign.
2. Ad group. One theme, intent area or buying situation, with its own context hints and its own maximum bid. This is where targeting lives.
3. Ad. The creative: advertiser name, favicon, title, copy, image and landing page URL, with UTM parameters supported per ad. Accounts can hold up to 5,000 campaigns, ad groups and ads, and campaigns can be built through the guided UI or bulk-uploaded via CSV.

The three campaign objectives

Source: OpenAI Help Center, “Create Campaigns for ChatGPT Ads”, August 2026. oCPC bills per click but bids toward a conversion goal, which is why it needs conversion data before it can help you.

Bid strategies and budgets

• Manual: Max bid. You set a ceiling per ad group. OpenAI’s guidance for CPC campaigns is a starting maximum of $3–$5 per click, and the default CPM cap is $60. Ads Manager shows bid-strength guidance so you can see whether a bid is likely to limit delivery.
• Maximize results. Introduced in August 2026 and now the default for eligible new ad groups. The platform sets bids automatically to get the most results for your budget, but it does not guarantee a CPA, CPC or ROAS target. If you have a hard allowable cost per lead, check this setting before you launch and switch to a manual cap.
• Daily budget is a seven-day average: a single day can spend up to twice the daily figure, and the seven-day total is capped at seven times it. Campaign-total budget is a hard ceiling with no pacing, so spend can arrive fast if your ads match a lot of conversations. OpenAI recommends new advertisers start with a daily budget, and so do we.

What ChatGPT Ads cost in the UK (2026)

As with any paid channel, “how much do ChatGPT Ads cost?” is two questions: what you pay OpenAI, and what you pay to have the account run well. Separate them when you compare quotes, and remember there is no rate card: everything is auction-priced and quoted in dollars, then billed to UK accounts in pounds on a postpay basis.

Pound conversions are indicative. OpenAI has not published UK-specific pricing, and UK clearing prices may run below US levels while the opt-in personalisation rules keep targeting largely contextual. Treat all ranges as planning baselines.

Two things to plan around. First, the auction is young and under-competed, so a £2 click on a topic where the equivalent Google keyword costs £8 is a real, temporary arbitrage. Second, that gap closes: in the US, the launch-era pricing advantage had largely gone within three months. Budget for the price you’d be willing to pay at maturity, and treat today’s rates as a bonus.

Management fees for ChatGPT Ads

The work of running a ChatGPT Ads account is different from Google: less bid-fiddling, far more writing. Context hints have to be drafted and iterated, creative volume has to be high (OpenAI explicitly favours many distinct title and copy variations per offer), and the pixel, Conversions API and reporting have to be set up properly or you’ll be judging the channel on screenshots. Most UK agencies are pricing this either as a flat monthly retainer or as an add-on to an existing paid-media retainer. We publish ours: see our PPC management pricing, and the scope of our ChatGPT Ads management service, which covers account approval, hint architecture, creative testing and full conversion measurement.

ChatGPT Ads vs Google Ads vs Meta Ads

ChatGPT Ads don’t replace either incumbent. They sit in a gap between them: higher intent than a social feed, earlier in the journey than a bottom-funnel search. Here’s how the three compare for a UK advertiser in 2026.

Google and Meta figures are typical UK 2026 ranges from our own accounts and published benchmark studies; ChatGPT figures are early advertiser reports from the US and UK pilots.

The practical model that’s working: run ChatGPT Ads to intercept people while they’re still framing the decision, retarget those visitors on Meta, and let Google Ads close the ones who go on to search for you by name. Brands that already appear organically in ChatGPT’s answers also report their paid placements performing better, because the ad reinforces something the user has just read rather than introducing a stranger.

Who can advertise on ChatGPT (and who can’t)

Eligibility is checked at three levels: your business category, each ad and landing page, and the live conversation. OpenAI opened the platform to a limited set of consumer categories first and has widened it steadily, with a policy update on 15 July 2026 clarifying health and financial services.

• Open categories. Household and consumer goods, retail and ecommerce, local and home services, travel and hospitality, entertainment, digital products, software and education were the launch categories, and most mainstream B2C and SMB-facing businesses are eligible.
• Restricted (case-by-case approval). Healthcare and medical services, and financial services, can now be approved with enhanced verification, proof of licensing where relevant, and manual review. Even when approved, these ads don’t appear beside sensitive personal conversations.
• Prohibited. Legal services (including personal injury, immigration and claims), gambling and betting, adult content and dating, alcohol and tobacco, illegal drugs, weapons, cryptocurrency, credit repair and debt settlement, political campaigning, and anything misleading or imitating the ChatGPT interface. General legal education (a law podcast, exam prep) can qualify; a personal-injury firm can’t.

Policies are moving quickly, so check the live OpenAI ad policies before you build. If you’re in a restricted category, plan for a longer approval process and build campaigns that can survive scrutiny: accurate claims, real credentials on the landing page, no pressure tactics.

ChatGPT Ads for lead generation: the playbook

Lead generation is where UK service businesses can win early. Trades, home services, education and training, care providers, property, recruitment, local professional services and SMB-facing B2B are all categories where people use ChatGPT heavily at the research stage, and where a well-placed, useful ad has almost no competition yet. Here is the system we run.

1. Build ad groups around buying situations, not keywords

Start with a list of the six to ten situations that lead someone to need you. For a Manchester accountancy firm that might be: “sole trader deciding whether to go limited”, “first VAT return coming up and unsure about Making Tax Digital”, “landlord with two properties working out self-assessment”, “startup founder needing first-year accounts and a business bank account”. Each situation becomes an ad group with its own context hints, its own ads and its own landing page. One ad group per situation keeps the relevance signal clean, and relevance is what you’re paying with.

2. Send the click to a page that continues the answer

The user has just read a considered answer to a specific question. If your ad drops them on a homepage or a bare “Contact us” form, the thread is broken and most of them leave. The landing pages that convert ChatGPT traffic are research-stage pages: a price guide, a comparison of the options they were just weighing, an eligibility checker, a quote calculator, a short “what to expect” walkthrough, with the lead step built into it. Early advertiser data shows high-consideration categories converting well above social benchmarks when the page matches the stage.

3. Make the first conversion proportionate to the sale

Ask for the smallest commitment that still qualifies a lead. For lower-ticket services that’s an instant quote, a call-back or a WhatsApp message. For higher-ticket services, a paid diagnostic usually beats a free consultation: our own £950 PPC audit is exactly this kind of offer. It’s a bounded, fixed-fee first purchase rather than a retainer, it filters out people who aren’t serious about fixing their account, and it gives them something valuable whether or not they go on to work with us. Whatever the step, keep the form short and offer click-to-call and WhatsApp, because a large share of ChatGPT usage is on mobile apps.

4. Bid in two phases

1. Phase one: Clicks objective, manual max bid. Start around £2.50 (roughly the $3 floor of OpenAI’s guidance), with 5–10 distinct ads per ad group. The goal is data: which situations fire, which hints attract the right clicks, which pages convert.
2. Phase two: clone to Conversions (oCPC). Once the pixel or Conversions API has recorded 30–50 of your lead event, clone the winning campaign to oCPC with that event selected. It still bills per click, but bids toward the clicks most likely to become leads. Keep a manual bid cap if your cost-per-lead ceiling is strict; the “Maximize results” default chases volume, not efficiency.

5. Close the loop on lead quality

Tag every ad with UTM parameters and OpenAI’s dynamic URL macros (campaign, ad group and ad IDs) so each lead lands in your CRM with its source intact. Then report cost per qualified lead at 60 and 90 days, not cost per form fill at day seven. ChatGPT leads tend to be earlier in their decision than search leads, so speed-to-lead and a proper nurture sequence decide whether the channel pays. Advertisers who follow up within minutes consistently report the best close rates.

ChatGPT Ads for ecommerce: the sales playbook

Retail is where the format is evolving fastest. Product feed campaigns, oCPC on purchase events and the carousel test are all aimed squarely at ecommerce, and OpenAI has said feed-driven ads are among its strongest performers. Here is how to build for sales rather than traffic.

1. Start with the product feed. Upload your catalogue via SFTP (up to two million SKUs) and let Ads Manager generate product-level ads. Feed hygiene is everything: accurate titles, live prices and availability, clean images on plain backgrounds, and consistent product data, because an inconsistent feed will look worse, not better, once the multi-product carousel arrives.
2. Structure standard campaigns around shopping situations. Gifting (“present for a dad who’s into cycling”), replacement (“my old coffee machine died”), comparison (“is a heat-pump tumble dryer worth the extra cost?”) and budget-versus-premium. Each is an ad group with hints that describe the situation and ads that answer it.
3. Land on the collection or product, never the homepage. Show price, delivery cost and returns policy above the fold. The user was one step from a decision when they clicked; don’t make them start again.
4. Put the offer in the copy. “Free UK delivery over £40”, “30-day returns”, “in stock, dispatched today”. Benefit-led, specific and verifiable copy is what OpenAI’s relevance model and your customers both reward.
5. Optimise toward purchase. Run Clicks first to gather add-to-cart and purchase events, then clone to oCPC with the purchase event once you have volume. oCPC for feed campaigns entered beta in August 2026 and is the natural next step for any catalogue advertiser.
6. Measure like you mean it. Install the pixel and the Conversions API together, deduplicated. Shopify and DTC brands can connect through the Triple Whale, WorkMagic or Hightouch integrations, and one-day view-through conversions now appear in reporting as a supplementary column.
7. Exclude and retarget. Exclude recent purchasers with a custom audience, and retarget ChatGPT-sourced visitors on Meta and Google, where your creative can do the heavy lifting of closing.
8. Plan for seasonality. Set campaign dates around launches, sales and Q4. Because campaign-total budgets don’t pace, use daily budgets during peak periods so a burst of matching conversations doesn’t exhaust the month in a week.

Run ChatGPT product campaigns alongside Google Shopping and Performance Max rather than instead of them: the Google campaigns capture the high-intent search, ChatGPT captures the “help me choose” conversation that happens before it. If you want the whole stack managed as one system, that’s what our AI-powered ads service is built for.

Context hints: the ChatGPT Ads targeting masterclass

Context hints are the closest thing ChatGPT Ads has to keywords, and they behave nothing like keywords. They are plain-language descriptions, entered at ad-group level, that tell OpenAI’s system what your product does, who it helps and when it becomes relevant. OpenAI’s own guidance gives the formula: instead of “running shoes”, write “cushioned everyday running shoes for beginners training for their first 5K”. The difference between a hint that works and one that wastes money is almost always specificity.

The what / who / when formula

Every strong hint answers three questions. What exactly is the offer? Who is the person, described by their situation rather than a demographic label? When does the need arise, what triggers it, and what constraint are they working around? Stack those into one natural sentence and you have a hint the model can match against real conversations.

Rules that hold up in live accounts

• Write 8–15 distinct hints per ad group. Cover the different ways customers describe the same problem: symptoms, comparisons, budgets, locations, urgency. Ten specific variants outperform one broad one.
• Describe situations, not audiences. “Parents of toddlers” is a label; “parent looking for a nursery place near Leeds for a two-year-old starting in September” is a situation the model can match.
• Keep one theme per ad group. If two hints would need different ad copy or different landing pages, they belong in different ad groups.
• Make the hints, the ad and the page agree. Relevance is scored across all of them together. A brilliant hint attached to a generic ad and a homepage dilutes every signal.
• Use UK vocabulary. “Conveyancing”, “boiler cover”, “MOT”, “self-assessment”, “half-term”. The system matches meaning, not exact words, but native phrasing lands closer to how UK users actually talk to ChatGPT.
• You can paste keywords, but reframe them. A Google search-terms report is a great source of situations. Turn each converting query into a sentence about the person behind it.
• Iterate weekly. Add hints for conversations you want more of, remove hints that attract cheap clicks and no conversions, and split any ad group whose hints have drifted into two themes.

ChatGPT ad creative and landing page best practices

OpenAI’s creative guidance is unusually direct: ads with clear, specific, benefit-focused copy are more likely to be judged relevant, and a high volume of distinct title and copy variations helps the system find more conversations where your ad fits. That is the opposite of the “one perfect ad” instinct. Think coverage.

Titles and copy

• Spell out the value in the title. What it is, who it’s for or what it does. OpenAI’s bulk-upload schema expects titles of roughly 16–24 characters and copy of 32–48, so every word has to work.
• Make the copy add new information. Expand on the benefit, the use case or the proof; don’t repeat the title.
• Write 5–10 ads per ad group, each from a different angle: price, speed, guarantee, specificity, social proof, a question the user is likely asking.
• Be accurate and plain. Ads are reviewed by classifiers and humans; exaggerated claims, unfounded comparisons and clickbait get rejected, and anything that imitates the look or voice of ChatGPT is removed.

Illustrative ads for our own services. Notice each title carries a different reason to click, and each copy line adds something the title didn’t say.

Images

Keep them simple, literal and matched to the message: the product on a clean background, the result of the service, a screenshot of the tool. Abstract stock imagery and cluttered lifestyle shots interpret badly at small sizes and add nothing to relevance.

Landing pages

• Deep-link. Product pages, collections, guides or dedicated landing pages, never a generic homepage. The page is a targeting signal as well as a conversion surface.
• Don’t block OpenAI’s crawlers. Landing pages must be reachable by OAI-AdsBot and OAI-SearchBot. Aggressive bot rules or a robots.txt disallow will quietly fail your ad review.
• Match the message. Headline on the page should echo the ad. Mismatch costs you both relevance and conversion rate.
• Mobile first and fast. A large share of Free and Go usage is in the iOS and Android apps. Test the page on a phone, with the form, before launch.
• Add UTMs per ad. Static parameters persist on every click; add the dynamic macros for campaign, ad group and ad ID so your analytics can join back to Ads Manager.

ChatGPT Ads measurement, conversion tracking and attribution

The August 2026 platform update was almost entirely about measurement, and it’s the point where advertisers who instrumented properly started pulling away from those judging the channel by CTR screenshots. Set this up before the first pound is spent.

• Native reporting. Ads Manager reports impressions, clicks, spend, CTR, average CPC, average CPM and conversions at campaign, ad-group and ad level, plus a supplementary one-day view-through column (VTA 1d) that doesn’t affect billing or bidding.
• Pixel and Conversions API. Install both, deduplicated, and define the events you care about: lead, quote, call, add-to-cart, purchase. Automatic advanced matching is now on by default for web pixels (it hashes supported form fields in the browser to improve match rates), so run it past whoever owns your privacy policy.
• UTMs and dynamic macros. Static UTM parameters persist on every click, and the {campaign_id}, {ad_group_id} and {ad_id} macros populate at delivery, giving you a clean join between Ads Manager and GA4 or your CRM.
• Integrations. Triple Whale, Hightouch and WorkMagic can send conversion events back to OpenAI and show ChatGPT alongside your other channels; agencies also buy through Adobe, Criteo, StackAdapt, Kargo, Pacvue and Skai.
• Blended view. Watch branded search volume and direct traffic after you switch ChatGPT Ads on. Some of the channel’s value shows up as people who later search for you by name.

What “good” looks like in the first 90 days

ChatGPT Ads budget: what £500, £2,000 and £5,000 a month buys

With a £15 daily minimum and no contract, the barrier to entry is tiny. The barrier to learning anything is higher. Here’s how the tiers break down for a typical UK SME at £2–£4 per click (ad spend only, before management).

How to lower your ChatGPT Ads cost per click

1. Sharpen the context hints. Relevance multiplies your bid in the auction. Ten specific hints per ad group, one theme per group, situations not labels.
2. Raise creative volume and variety. More distinct titles and copy lines give the system more chances to find a relevant match at a lower clearing price.
3. Bid manually while you learn. The “Maximize results” default chases volume. A manual cap around £2.50 keeps you in the auction without paying for marginal clicks.
4. Make the landing page a targeting asset. The crawler reads it. A page that plainly describes one offer, for one situation, lifts relevance across every ad pointing at it.
5. Exclude people you’ve already won. Upload customer lists as a custom audience and exclude them from acquisition campaigns.
6. Cut surfaces that don’t convert. If Android app traffic clicks but never completes your form, use platform targeting to move that budget to web or iOS.
7. Graduate to oCPC. Once you have 30–50 conversions, conversion-optimised bidding buys you the clicks most likely to pay, which usually lowers cost per outcome even if CPC rises.
8. For retailers: fix the feed. Clean, complete product data is the difference between feed ads that OpenAI calls its strongest performers and feed ads that never serve.

How to launch ChatGPT Ads: your first 90 days

1. Week 0: set up properly. Register at openai.com/advertisers with a work email, complete identity and business verification, add billing and your tax ID, and confirm your category is eligible. Check robots.txt allows OAI-AdsBot and OAI-SearchBot. Install the pixel and Conversions API, define your conversion events, and agree your UTM convention.
2. Weeks 1–2: launch one Clicks campaign. Daily budget, manual max bid around £2.50, three to five ad groups built around buying situations, 8–15 hints and 5–10 ads in each, every ad deep-linked to a stage-matched page. Submit for review and watch the “Not serving” statuses for anything rejected.
3. Weeks 3–6: optimise weekly. Review delivery by ad group, prune hints that attract clicks without engagement, add hints for the conversations you want more of, and pause the weakest ads. Test a platform split. Raise budgets on winners in 20–30% steps.
4. Weeks 7–12: move to conversions. Clone winning campaigns to oCPC with your lead or purchase event, add custom-audience exclusions, and for retailers, launch a product feed campaign. Report cost per qualified outcome against your blended channel CPA and make the keep, scale or kill decision on that number.

If you’d rather skip the learning curve, our ChatGPT Ads management team has run this sequence since the UK beta opened, with transparent, published pricing.

ChatGPT Ads mistakes that burn budget

• Judging the channel on CTR against search. The user already got an answer; a 1% CTR from that context can be worth more than 5% from a results page. Measure outcomes.
• One broad ad group with keyword-style hints. “Plumber”, “boiler”, “heating” isn’t targeting, it’s a shrug. The relevance model gives you back exactly the vagueness you gave it.
• Sending clicks to the homepage. You lose the conversation thread and the landing-page relevance signal in one move.
• Blocking the crawlers. A security plugin that challenges OAI-AdsBot will fail ads at review, with no obvious error.
• Leaving “Maximize results” on with a thin margin. It optimises for volume and doesn’t promise a CPA. Cap the bid if your economics are tight.
• Using a campaign-total budget during a busy period. It doesn’t pace, so a burst of matching conversations can spend the month in days.
• Launching without the pixel and Conversions API. You will never be able to move to oCPC, and you’ll be arguing about screenshots in month three.
• Expecting enterprise B2B volume. Those buyers are on paid, ad-free plans.
• Writing slogans instead of answers. Catchy lines score badly on relevance and worse with a user who just read a considered explanation.
• Switching it off after two weeks. The learning window is 60–90 days; most early accounts that “failed” were killed before the data arrived.

So, are ChatGPT Ads worth it in 2026?

For businesses whose customers research before they buy, yes, as a properly measured test rather than a bet. Consumer services, ecommerce, education, travel, home services and SMB-facing software are the categories where the intent is strong, the inventory is under-priced and the competition is still thin. The auction is at its cheapest point, the measurement stack is finally good enough to hold the channel to the same standard as Google and Meta, and every pound spent now doubles as an education in a platform that 31 more European markets are about to join.

It is not a replacement for search or social. The best results we’re seeing come from running ChatGPT Ads as the first touch, with Google capturing the branded and high-intent searches that follow and Meta doing the retargeting. It also helps to be visible organically: brands that ChatGPT already mentions in its answers report stronger paid performance, so treat generative engine optimisation and paid placement as one project, not two. That integrated approach is what our AI-powered ads service delivers, and you can see the results in our case studies.

The question, as ever, isn’t “how much does it cost?” It’s “what is a customer worth to me, and can I acquire one profitably in the place they’re now doing their research?” For a growing share of UK buyers, that place is a ChatGPT conversation. For more UK PPC and AI advertising guides, browse the Global PPC blog.

Frequently asked questions

Hassan Tahir

Founder & Lead PPC Strategist · Global Performance Propel Company · Google Partner & Meta Business Partner