Homepage or Dedicated Page for ChatGPT Ads Traffic
By Ansh Khandelwal 8 min read
In brief
Compare a homepage with a dedicated ChatGPT Ads landing page using relevance, continuity and conversion-measurement criteria.
Last verified: 12 September 2026 | Version: 1.0 | Next scheduled review: 12 December 2026
Dedicated page. The reason usually given for that answer is a conversion statistic, and the conversion statistics on this question are all produced by companies that sell the thing they are measuring. The reason that survives scrutiny is different and better: on ChatGPT Ads the landing page is not only where the click lands. It is an input to whether the click happens at all.
What OpenAI actually says
Two statements, both primary, both short.
In Create Ads for ChatGPT, OpenAI instructs advertisers to "link to the most relevant destination: use product pages, collections, or content pages rather than defaulting to a generic homepage". That is the platform telling you the answer in its own documentation, which makes the vendor statistics unnecessary for the basic decision.
In Ads in ChatGPT, describing how ads are selected, OpenAI lists the signals it considers: "the context and intent of your current conversation", "the ad's landing page, title, copy", and "advertiser-provided context hints and targeting selections". Ranking across eligible ads uses "a combination of factors, including relevance and advertiser bids".
Read those together. The page is inside the auction. It is not a downstream conversion surface that the auction ignores, it is one of the named things the system reads when deciding whether your ad appears and where it ranks.
Separately, OAI-AdsBot must be able to reach the page. OpenAI states that landing pages "must use valid links and must not block OpenAI user agents (OAI-AdsBot and OAI-SearchBot)". A page the crawler cannot read cannot contribute a relevance signal, whatever is on it.
The evidence audit
Here is the part most write-ups on this topic skip. The circulating numbers, and who produced them.
| Circulating claim | Who published it | What they sell | Tier |
|---|---|---|---|
| Dedicated pages convert at 5 to 15%, homepages at 1 to 3% | Lapis, landing page guide, updated August 2026 | ChatGPT ads landing page tooling and services | Vendor |
| Matching content depth to conversation depth lifts conversion 20 to 60% | Various agency write-ups, 2026 | Landing page and campaign services | Vendor |
| One pilot moved 2.1% to 4.8% by moving the CTA above the fold with a trust signal | Lapis, August 2026 | As above | Vendor, single unnamed account |
| ChatGPT ads converting around 2%, roughly half the rate of organic ChatGPT traffic | Q1Media, July 2026 | Media buying services | Vendor |
| ChatGPT referral traffic converted at 2.08% on high intent actions, versus 0.5% for Google Search, across 97 B2B sites and 28.9m sessions, July 2025 to June 2026 | Orbit Media Studios, 2026 | Web design and strategy, but the study measures client-agnostic GA4 data and is not a claim about its own campaigns | Reported, third party |
Only the last row is a dataset rather than a testimonial, and it measures organic referral traffic, not ads, so it cannot settle a question about ad landing pages.
There is a further problem with the top row that has nothing to do with honesty. A 5 to 15% figure for dedicated pages and 1 to 3% for homepages compares two populations that differ in more than page type. Companies that build dedicated pages are running more mature programmes, with better offers, tighter targeting and someone whose job it is to look at the numbers. The page is confounded with the competence. Nobody publishing these figures has controlled for that, and a within-account split test on this channel has not been published by anyone.
So: the direction is credible, from the platform's own instruction and from ordinary reasoning about relevance. The magnitude is not. Treat any specific multiplier you read on this topic, including the ones above, as marketing.
The argument that does not need vendor data
Relevance is weighted in the ChatGPT Ads auction, and the landing page is a named relevance input. That gives you a mechanism argument rather than a benchmark argument.
If the page is read and scored, a page whose content closely matches the ad group's hint theme plausibly earns more eligibility and a better rank at the same bid than a homepage that covers eleven product lines and commits to none. Better rank at the same bid is, in any relevance-weighted second-price system, the route to paying less for the same position. Agency write-ups through 2026 report OpenAI guiding CPC bids into the three to five dollar range, so the amount at stake per hundred clicks is real money even before conversion enters the argument.
Here is the honest limit on that reasoning. OpenAI publishes no landing page quality signal. There is no quality score, no relevance grade, no page diagnostic, no per-page eligibility view, nothing in Ads Manager that tells you whether the page helped or hurt. The mechanism is documented in a sentence. Its weight, and the size of any cost benefit, are unpublished. If someone quotes you a percentage that a better page saves on CPC, they are inventing it, because there is no instrument that would produce that number.
When a homepage is genuinely the right answer
Three cases, and they are not rare.
A single-product company whose homepage already is the product page. Building a duplicate with different furniture is work without a hypothesis.
A first two-week validation test where the question is whether the channel serves and clicks at all. At a 25 USD daily minimum, two weeks is 350 USD of spend, which is not enough click volume to read a page comparison anyway. Point it at the best existing page and spend the build effort after you know the channel produces traffic.
A brand or reach campaign on the CPM objective, where the click is not the outcome being bought.
Outside those, the build cost of one focused page is small against the cost of feeding three to five dollar clicks into a page that names eleven things and answers none of them.
The practical rule
One page per ad group, not one page per campaign. The ad group is the unit that carries the hint theme and the unit that gets reported, so it is the unit a page should correspond to. If you cannot articulate a distinct promise for a second ad group's page, you probably do not have a second ad group, you have one theme split in half.
What we cannot tell you
- How heavily the landing page weighs in ad selection. OpenAI names it as a signal and publishes no weighting.
- Whether a better-matched page lowers your CPC, and by how much. No quality score, no relevance diagnostic, no instrument in Ads Manager that would measure it.
- The real size of the homepage versus dedicated page conversion gap. Every published figure is a vendor claim, none control for account maturity, and no within-account split test has been published.
- What OpenAI extracts from the page. Undocumented. Whether it reads the headline, the full body, or structured data is not stated.
- Our own numbers. InPromptAds has no first-party campaign data and no clients, so nothing here is drawn from accounts we run.
Quick answers
Should ChatGPT Ads traffic go to my homepage? Only for a single-product company whose homepage is the product page, a first validation test too small to read, or a CPM reach campaign. Otherwise build a dedicated page. OpenAI's own Create Ads guidance says to avoid defaulting to a generic homepage.
Does the landing page affect whether my ad serves? OpenAI lists the ad's landing page, title and copy among the signals it considers when selecting ads. The direction is documented. The weight is not published.
Is there a landing page quality score in ChatGPT Ads? No. Nothing in Ads Manager reports page relevance, page quality or per-page eligibility. You infer page effects from CPC moving against a fixed bid, which is weak evidence.
How much better do dedicated pages convert? Unknown. The 5 to 15% versus 1 to 3% figures circulating in 2026 come from vendors selling landing page services and do not control for the maturity of the accounts being compared.
How many landing pages do I need? One per ad group, because the ad group carries the hint theme and is the smallest unit that gets reported.
Does OAI-AdsBot need to reach my landing page? Yes. OpenAI requires that landing pages not block OAI-AdsBot or OAI-SearchBot. A blocked page can get ads rejected and cannot contribute a relevance signal.
Sources
| Claim | Source | Tier |
|---|---|---|
| Link to the most relevant destination rather than defaulting to a generic homepage | OpenAI Help Center, Create Ads for ChatGPT, 2026 | Confirmed, primary |
| Ad selection considers the ad's landing page, title and copy alongside conversation context and hints | OpenAI Help Center, Ads in ChatGPT, 2026 | Confirmed, primary |
| Ranking uses a combination of factors including relevance and advertiser bids | OpenAI Help Center, Ads in ChatGPT, 2026 | Confirmed, primary |
| Landing pages must not block OAI-AdsBot or OAI-SearchBot | OpenAI Help Center, Create Ads for ChatGPT, 2026 | Confirmed, primary |
| No landing page quality score or page relevance diagnostic exists | OpenAI Ads Manager documentation, by absence | Absent |
| Dedicated pages 5 to 15%, homepages 1 to 3%; 2.1% to 4.8% pilot result | Lapis, ChatGPT ads landing page guide, updated August 2026 | Vendor |
| ChatGPT ads converting around 2%, about half organic ChatGPT traffic | Q1Media, July 2026 | Vendor |
| ChatGPT referral 2.08% versus Google Search 0.5% on high intent actions, 97 sites, 28.9m sessions | Orbit Media Studios, 2026 | Reported, third party |
| Three to five dollar CPC bid guidance | Multiple agency write-ups citing OpenAI guidance, 2026 | Reported |
| Confounding of page type with account maturity in published comparisons | InPromptAds analysis | Inference, ours |
Related reading
- The ChatGPT Click Arrives Mid-Conversation. Most Landing Pages Restart It
- Continuity: Making the Page Answer the Question the Ad Interrupted
- What OAI-AdsBot Is and Why It Must Not Be Blocked
- How Much Do ChatGPT Ads Cost?
- Budget Calculator
Changelog
12 September 2026, v1.0. First publication. Replaces the vendor conversion-gap argument with the documented ad selection signal, and records that OpenAI publishes no landing page quality signal of any kind.
Field kit
Tools
Site
OpenAI (primary)
Ansh works across GEO strategy, B2B research, and execution. At InPromptAds, he translates new AI advertising products into clear operating advice, tests, and measurement questions for marketing teams.