Web inventory is the total supply of advertising placements available across web-based surfaces, including ad slots on websites and the sponsored units now appearing inside conversational interfaces like ChatGPT. It refers to the available ad supply itself, the space waiting to be bought, separate from the creative, targeting, or bidding that fills it.
When a platform opens new web inventory, you gain fresh places to reach buyers at the moment they ask a question, and the earliest advertisers usually face the least competition for those slots. Ignore a new source of inventory and you cede that attention to competitors who claimed the space while it was cheap and uncrowded.
In advertising operations, web inventory counts how much ad space a publisher or platform can sell across its web surfaces in a given period. Each placement is defined by where it sits, what format it accepts, and whether it sells directly or programmatically. On ChatGPT, that supply takes the form of sponsored cards below an answer, labeled and kept separate from the model's response.
Inventory has two sides. Supply is everything a platform makes available: eligible slots, the audiences they reach, and the moments they trigger. Demand is the advertisers competing to fill those slots. Your addressable inventory depends on who is eligible to see ads, which is why OpenAI opening ChatGPT ads to logged-out and free users enlarged the pool marketers can buy.
Web inventory sits upstream of impressions and clicks: the slots exist first, then buyers fill them and users see them. It differs from your organic AI visibility, which you earn through citations instead of buying. AirOps tracks both the paid placements and the organic mentions across the same ChatGPT prompts, so you see how bought inventory sits beside the answers you earn.
Resources: See how AirOps tracks competitor ads and paid share of voice inside ChatGPT answers
Web inventory becomes a buyable channel through a sequence that runs the same way whether the surface is a web page or a ChatGPT answer.
Supply opens. A platform or publisher designates which slots can carry ads and who is eligible to see them. Expanding eligibility, as OpenAI did with logged-out users, grows the pool.
Slots get described. Each placement is cataloged by format, position, and environment so buyers know what they are purchasing.
Buyers bid. Advertisers enter a budget and targeting signals through an ads manager or an auction, then compete for matching slots.
Ads serve. When a user's prompt or page matches, the winning ad renders in the slot, such as a sponsored card below a ChatGPT reply.
Delivery is measured. The platform counts impressions, clicks, and fill rate, reporting how much of the available inventory actually sold.
Fill rate and impressions tell you how much of the supply moved and at what price. They do not tell you which specific prompts or conversations triggered your ad, context that platforms like ChatGPT still withhold.
Resources: Watch the AirOps webinar on testing ChatGPT ads as one paid and organic system
Deciding how much to spend on a new channel starts with knowing how fast its inventory is growing. ChatGPT ads moved from a niche test to a channel marketers now budget for in under a year, and AirOps research recorded a 113x increase in AI ad load across the industries it monitored over three months. That pace forces a planning decision: treat this inventory as a line item now, or wait and pay more later.
New reach at high intent: ChatGPT inventory appears when a user is actively asking what to buy, a moment traditional display rarely captures.
A first-mover discount: early inventory is lightly contested, so the advertisers who test now pay less for attention than those who wait until auctions fill up.
A visibility gap if you sit out: when a competitor buys the sponsored slot under a prompt you want to own, your brand is absent from that answer even if you rank well organically.
Demand gen leads use web inventory to place sponsored offers inside ChatGPT answers where buyers are comparing solutions in real time.
Growth marketers use web inventory to test a new acquisition channel with a small budget before competitors drive up auction prices.
SEO managers use web inventory data to see which prompts competitors are paying for and where paid spend overlaps with their organic gaps.
Fill rate is the share of available inventory that actually sells in a given period, and a persistently low fill rate signals either weak advertiser demand or a supply pool that has grown faster than buyers can spend against it.
An impression is a single appearance of an ad in a slot, and CPM, the price an advertiser pays for a thousand impressions, is the pricing model most ChatGPT inventory is still bought on today.
Environment describes where a slot physically lives, such as a web page, a mobile app, or a conversational AI answer, and it shapes both the format an ad must take and how users react when they see it.
Reach buyers inside ChatGPT at the moment they ask what to purchase.
Enter the channel for as little as $50,000, the minimum OpenAI set for its 2026 ChatGPT ad pilot, per eMarketer.
Test contextual targeting tied to a live prompt instead of a static audience segment.
Measure delivery through impressions, clicks, and fill rate reported by the ad platform.
Claim lightly contested slots before auctions crowd and prices climb.
Start with one high-intent prompt set so your first spend targets conversations where buyers are close to a decision.
Set a small test budget first so you can read fill rate and cost before committing to the channel.
Match creative to the conversation because ChatGPT users expect an ad that continues their question instead of a generic banner.
Track competitor placements so you know which prompts rivals are paying to own and where gaps remain.
Keep paid and organic aligned so a prompt you buy today is one you also work to earn citations on.
Recheck eligibility rules monthly because platforms change which users, plans, and countries see ads often.
Avoid treating web inventory as a set-and-forget buy. Early AI ad inventory shifts weekly as platforms open new surfaces and adjust who sees ads, so a campaign you leave unattended drifts out of the prompts and audiences you cared about. Review placements and spend on a fixed cadence.
AirOps: tracks where your ads and competitors' ads appear across your tracked ChatGPT prompts, connecting paid web inventory to the organic AI search mentions you earn.
OpenAI Ads: the advertiser platform at ads.openai.com where you buy and manage the sponsored web inventory that runs inside ChatGPT answers.
Google Ad Manager: manages and sells web ad inventory programmatically across sites, giving teams a reference point for how slots are cataloged and auctioned.
Map your highest-intent prompts. List the questions buyers ask ChatGPT right before they choose a product. You can do this in a spreadsheet this week with no budget.
Check where ads already run. Search those prompts in ChatGPT and note which competitors appear in sponsored slots, so you know the inventory is live for your topics.
Confirm eligibility. Review OpenAI's current rules for which plans, users, and countries see ads, since the eligible audience sets the ceiling on the inventory you can actually reach for your prompts.
Set a small test budget. Open the ads manager, fund a limited campaign, and tie it to your priority prompts before scaling, so you learn the channel's cost on a budget you can defend.
Measure and connect to organic. Read fill rate, impressions, and clicks, then compare paid placements against the prompts where you already earn or miss citations.
Web inventory is the supply of ad placements a platform makes available, now including sponsored slots inside ChatGPT answers.
It is measured by impressions, fill rate, and the clicks a platform reports after ads serve.
The main constraint is eligibility: only the users, plans, and countries a platform allows can see ads, which caps the reachable pool.
The main risk is ceding a high-intent slot to a competitor who buys the inventory under a prompt you want to own.
The biggest advantage is timing, since early buyers face lighter competition and pay less for attention than those who wait.
Web inventory and impressions describe different points in the same transaction. Inventory is the supply: the set of ad slots a platform makes available before anyone buys them. An impression is what happens after a buyer wins a slot and an ad actually renders for a user. Inventory is the shelf space; impressions are the products that move off it. In ChatGPT, the inventory is the pool of sponsored positions that can appear below answers, while an impression is one real appearance tied to a specific user prompt. The distinction matters for planning. You buy against inventory, estimating how many slots exist for the prompts you care about, but you are billed and judged on impressions and the clicks they drive. A platform can hold large inventory yet deliver few impressions if demand is thin or your targeting is narrow, so healthy inventory does not guarantee delivery on its own.
The cost to start buying web inventory on ChatGPT has fallen sharply since the pilot opened. Early access required large upfront commitments and ran through agency partners, which priced out most smaller advertisers. Since then, OpenAI has lowered minimums and introduced more self-serve options, so testing the channel no longer demands an enterprise budget. Your actual spend depends on three things: how competitive the prompts you target are, the pricing model the platform applies, and how many impressions you want to buy. Most ChatGPT inventory is still sold on a cost-per-thousand-impressions basis, though some placements have moved toward click-based pricing. The practical approach is to fund a small test tied to a short list of high-intent prompts, measure fill rate and cost per result, then decide whether to scale. Start small enough that the learning is affordable even if the first campaign underperforms, because early benchmarks on a new channel are rarely stable.
Available web inventory in ChatGPT varies because the supply is defined by the platform, and OpenAI keeps changing the rules. The biggest driver is eligibility. Inventory only exists where ads are allowed to show, so each time OpenAI adds a plan tier, a country, or a user group such as logged-out visitors, the pool grows. When it restricts ads on certain surfaces or for certain users, the pool shrinks. Demand also moves the number you can actually win. Even when slots exist, a surge of advertisers bidding on the same prompts leaves fewer available to you at a price you will pay. Seasonality and topic matter too, since some conversations attract far more advertiser interest than others. Finally, the channel is young, and OpenAI is still testing formats, placements, and pricing, so what counts as inventory this month may look different next month. Treat the supply as a moving target and recheck it on a regular cadence.
Partly. You cannot create web inventory, because only the platform decides how many ad slots exist and who is eligible to see them. What you can influence is how much of the existing inventory you actually reach. Start by targeting the prompts and topics where your buyers ask questions, since inventory is tied to conversations, and choosing the right ones puts you in front of slots that matter. Set clear targeting and context signals so the platform matches your ad to relevant moments. Raise your bid or budget where competition is high and the inventory is worth winning. Keep your account eligible and compliant so your ads are not restricted. You can also influence the organic side in parallel, earning citations and mentions for the same prompts, so your brand appears whether a given answer carries an ad or does not. Direct control over supply stays with the platform, but your reach within that supply is yours to shape.
A good fill rate depends on the channel's maturity, so there is no single number that applies to web inventory everywhere. Fill rate is the share of available slots that actually sell, and on established programmatic surfaces, mature publishers often expect most of their premium supply to clear. On a young channel like ChatGPT ads, the picture is different. Supply has grown faster than demand in the early phase, so a large share of inventory can go unsold, and low fill rates reflect the market finding its footing more than a problem with your campaign. For your own buying, judge fill rate against your goals instead of an industry average. A healthy result is winning enough of the slots under your priority prompts to generate meaningful impressions at a cost you can justify. Watch the trend over time, because a fill rate climbing as you refine targeting tells you more than any single snapshot.