ChatGPT ads for software companies, run as a real advertising line.

Yes, on a cost-per-click basis through OpenAI's ads manager, with conversion-optimised bidding and the web pixel as the tracking layer. Digiday reports the business has expanded to 31 markets, with cost-per-click now accounting for most of the spend. What is not there yet is the machinery paid social buyers expect: no lookalike audiences, no interest targeting and no video inventory, so the leverage comes from the offer, the copy and the landing experience. We keep the ChatGPT ads overview updated as new controls ship.

The intent arrives with context. A search query is four words; a conversation carries the team size, the incumbent tool, the integration that broke and the timeline, which lets you be specific about the accounts you want instead of bidding broadly and filtering later. The economics are the comparison that matters: B2B SaaS advertisers pay about $6.81 a click on Google with an $84 blended cost per lead, and this channel has to beat that on qualified pipeline to keep its budget. We run both from one plan so the comparison is honest rather than competitive between vendors.

Enough to reach a decision, which is usually a small share of existing paid media over eight to twelve weeks rather than a separate war chest. The variable that decides the answer is not the spend but the contract value you are buying: a product selling $40-a-month seats needs a very different volume of activations than one selling $60,000 annual contracts. We size the test against average contract value and close rate, agree in advance what result would make it a keeper, and stop early when the data is clear either way.

An activated trial or a qualified demo verified in your CRM — never a download or a newsletter signup. Qualified status flows back so reporting separates a genuine evaluation from a student exploring the category, and bidding learns from the difference. Because the median free-to-paid conversion across SaaS products is 8%, sign ups on their own are a misleading target; cost per activated trial and cost per opportunity are the two numbers we hold this channel to.

Not the way LinkedIn lets you, and pretending otherwise is how budgets get wasted. There are no lookalike audiences and no interest targeting on this platform today; relevance comes from the context of the conversation plus the product data, campaign hints and conversion signals you supply. In practice that is less limiting than it sounds for software, because a fifth of over a billion weekly users show commercial intent and the conversation itself carries the qualification a filter used to approximate. Where firmographic precision is essential, we pair this with LinkedIn ads.

Most companies need two or three new pages rather than a rebuild. The mismatch is predictable: somebody asks a specific migration question and lands on a homepage listing eleven features, so the click is paid for and the visit is wasted. We build one page per use case that answers the question asked, states pricing posture instead of hiding it, puts the security and migration answers within reach of the wider buying group, and starts a trial in two minutes. Those pages usually lift the existing paid search account as much as they lift this one.

Through targeting hints rather than through settings. OpenAI asks advertisers for campaign context and product data, then decides where an advertisement is relevant to the conversation in front of the user. So the brief we write is unusually plain: who the product is for, the problem it solves, the tools it replaces, the price band, and the buyer we do not want. That content, plus your conversion signals and clean UTM tracking on every click, is what improves relevance over time. It is closer to briefing a well-read salesperson than to building an audience in a paid social tool.

Handled well they help, because the advertisement appears while somebody is comparing brands rather than scrolling past them. The risk is the same one every conversational ad placement carries: an ad that contradicts the answer beside it looks worse than no ad at all. We keep the creative honest about what the product does, avoid claims the landing content cannot back, and test messaging in small budgets before scaling. For software brands with a category story to tell, appearing consistently in these conversations is also the cheapest brand-building this year, because so few competitors are there yet.

You buy clicks, the same as on Google, and each click lands on a page with campaign and advertisement identifiers in the UTM string. The ChatGPT web pixel records the conversion, Automatic Advanced Matching improves the match rate, and your CRM returns qualified status so reporting separates curiosity from a real evaluation. Attribution deserves honesty though: conversational discovery often precedes a branded search days later, so some of the value shows up in another channel's column. That is why we read the programme monthly on blended cost per opportunity and, where volume allows, with a holdout test rather than trusting any single vendor's own numbers.

The user describes a problem in their own words — a prompt rather than a keyword — and OpenAI may place an ad alongside the answer, marked as sponsored. For a software company that means appearing while somebody is mid-decision, with the context already in the conversation: the stack they run, the team size, the integration they need and the deadline they are under. Nothing about an individual prompt is bought or targeted directly; advertisers supply campaign context and product data as hints, and OpenAI decides where an ad is relevant. Free users see the most inventory today, which is where much of the early-stage evaluation happens.

Mostly the ones that used to be a G2 tab and three demo calls. Users ask what tool fits a ten-person team, how to migrate off a legacy platform without losing history, whether two products integrate, what a category typically costs, and which alternative is best for a specific constraint. That is exactly where advertisers belong, because the person is comparing rather than browsing. The strongest campaigns we run come from the questions a company's own sales engineers answer every week — they already know the prompts, they have simply never written them down as marketing.

As one more line in the same plan, not a separate project. Most teams we work with run Google Ads, LinkedIn, review sites, content and a webinar programme; these campaigns slot in beside them and share the tracking spine, which means the same conversion definition, the same CRM statuses and one monthly report showing what each source cost per qualified opportunity. Where there is process debt — no lead routing, nobody following up inside a day — we fix that first, because new advertising only exposes it faster. The pages and creative built here usually improve the older channels too.

Website access to install the pixel, an ads account in your company's name, the CRM you already use so lead status can flow back, and half an hour with whoever knows which accounts are worth winning. From our side the process is documented: a written build plan, a change log and a monthly review. No customer records or personally identifiable data ever go into an advertising platform — the events we send are conversion signals tied to an activation, not to anybody's account contents. For teams with a security review, we complete it before launch rather than after.

Delivery and tracking signals show inside the first two weeks; a reliable read on cost per activated trial takes six to eight weeks; and revenue attribution follows your sales cycle rather than the ads. We compare like with like rather than judging a quiet month on its own, and the monthly review states plainly whether ChatGPT ads earned their budget. If the answer is no twice in a row, we move the money — an easy thing to promise and the main reason software teams stay with us.

A fixed monthly fee, quoted separately from media investment and scoped to the motions you want run. After the audit you get a plan tied to targets — cost per activated trial first — and we will name the channels we would not run yet rather than selling a bundle on day one. Where the useful scope is a measurement build and a landing page rather than a retainer, we scope to that and say so on the first call. You own the ad account, the pixel and the pages throughout.