Reading CTR and CPC When No Benchmarks Exist
By Ansh Khandelwal 9 min read
In brief
Interpret ChatGPT Ads CTR and CPC without industry benchmarks by creating a controlled first-party baseline and separating delivery from quality.
Last verified: 12 September 2026 | Version: 1.0 | Next scheduled review: 12 October 2026
There is no benchmark to read your CTR against. OpenAI publishes no cross-advertiser performance data for ChatGPT Ads, and the figures circulating on agency and tool sites are vendor content with samples that are almost never stated. The working answer is to make your own account the benchmark, which is both weaker than an industry figure and more honest than one.
Why no benchmark exists, and why that is structural
OpenAI's Ads Manager reports impressions, clicks, spend, CTR, average CPC, average CPM and conversions at campaign, ad group and ad level. It reports them for your account only. There is no industry comparison view, no vertical average, no percentile ranking, and nothing equivalent to Google's auction insights.
This is not a gap waiting to be filled in the next release. Cross-advertiser benchmarks require a platform to aggregate and publish competitive performance data, which platforms generally do once a channel is mature enough that the aggregate flatters it. ChatGPT Ads launched its self-serve manager in 2026. A benchmark published now would describe a handful of months of unstable inventory, so the absence is at least partly a decision rather than an oversight.
The practical effect is that a 0.9% CTR on this channel is not good or bad. It is a number with no denominator.
What the published figures actually are
Search for ChatGPT Ads benchmarks and you will find pages offering CTR ranges, CPC ranges and conversion rates, typically presented in a tidy table by vertical. Through 2026 these have been published by agencies and marketing-tool vendors, including pages from Tru Commerce, 360ROI, Top Growth Marketing and Adventure Media among others.
Read one of them with three questions in hand.
How many advertisers are in the sample? Usually unstated. A table of vertical-level CTR benchmarks built from fewer accounts than it has rows is a formatting exercise, not a measurement.
Over what period? Delivery on this channel changed materially during 2026. Digiday reported that fill rates climbed 30 to 50 percent from launch levels by late April, after a period in which one advertiser spent 2,500 USD of a 250,000 USD commitment across four weeks for roughly 200 impressions. A CTR average blended across that transition describes two different channels at once.
Who benefits if the number looks attractive? A vendor selling management of a channel has an interest in that channel's published benchmarks looking healthy. This is not an accusation of bad faith. It is the reason the tier label exists.
Treat every one of these as Vendor, permanently, with a date attached. That does not make them useless. It makes them evidence of what some advertisers reported, not evidence of what the channel does.
The genuinely useful published numbers are the ones attached to a named person and a named campaign. MediaPost reported on 31 August 2026 that Peter Jaffray of Choice OMG spent 415 USD across three campaigns in June 2026 and recorded 9,000 impressions, 60 clicks, a 0.6% CTR and roughly 7 USD per click. That is a sample of one and it is worth more than a table of unsourced vertical averages, because you can see exactly what it is.
Self-referential benchmarking, in practice
The method is to make comparison internal. You are not asking whether your CTR is good. You are asking whether this ad group is better than that one, and whether this fortnight is better than the last.
Step one: freeze a baseline. At the end of the first fourteen days, write down four numbers for the campaign as a whole: total spend, CTR, average CPC, and clicks that reconcile with your own analytics. Date it. This is your reference point and it does not get revised retrospectively when later results are nicer.
Step two: compare only inside the structure. Two ad groups running concurrently under one campaign share the objective, the budget pool, the locations, the platforms and the period. Differences between them are attributable to their hints, their bid and their creative. That is a real comparison. Comparing your ad group against a published industry figure is not, because you share none of those conditions with whoever produced it.
Step three: use creative variants as the controlled test. Two ads inside one ad group share the hints, the bid and the matching surface. The platform holds every upstream variable constant for you. This is the cleanest experiment available on the channel and most advertisers never run it because they are busy comparing themselves to a table.
Step four: benchmark against your other channels on outcome, not on CTR. CTR on a conversational surface and CTR on a search results page are not the same quantity, and neither is CPM against a feed. The comparison that survives across channels is cost per qualified lead, measured in your own CRM. That number means the same thing everywhere.
Step five: rebaseline on a schedule, not on a whim. Monthly. A baseline you update whenever a number looks good is not a baseline.
What a self-referential baseline cannot do
It cannot tell you whether the channel is competitive. If your cost per qualified lead is 340 USD and every advertiser in your category is achieving 120 USD, internal comparison will never reveal that. You will optimise steadily toward a local maximum and have no way of knowing where it sits.
This is the real cost of the missing benchmark and it should be stated plainly rather than managed around. Internal benchmarking tells you the direction of travel. It cannot tell you your altitude.
The only partial remedy is cross-channel outcome comparison. If your cost per qualified lead here is four times what Google non-brand delivers for the same offer, that is informative regardless of what any ChatGPT Ads benchmark says. Digiday reported one 2026 client tracking conversions close to Google non-brand search efficiency, which is the shape of a comparison worth making, though it is a single anonymous account.
The reckoning: we have no numbers either
InPromptAds has no first-party ChatGPT Ads benchmarks to publish. No client accounts, no campaign data, no CTR or CPC figures of our own.
That is worth saying directly on a page about benchmarks, because the standard move in this category is to publish a table anyway, built from whatever numbers are reachable, and let the presentation imply a sample that does not exist. Every figure in this article is attributed to someone else and tiered accordingly. When we have account data, it will be published with the account count, the period and the category attached, and it will still not be a benchmark. It will be a sample.
What we cannot tell you
- What a good CTR is on ChatGPT Ads. OpenAI publishes no cross-advertiser benchmarks and there is no percentile or auction-insights view.
- The sample size behind any published benchmark table. Almost universally unstated by the vendors publishing them.
- Whether reported CTRs from early 2026 are comparable to today's. Delivery changed materially during the year, so figures spanning that period blend two conditions.
- Whether OpenAI's click count reconciles with site analytics. Agencies reported mismatches in both directions in August 2026 and no reconciliation method is published.
- InPromptAds' own numbers. None exist. No client campaigns have run.
Quick answers
Does OpenAI publish ChatGPT Ads benchmarks? No. Ads Manager reports your own account's impressions, clicks, spend, CTR, average CPC, average CPM and conversions. There is no cross-advertiser comparison, no vertical average and no auction insights equivalent.
Are the ChatGPT Ads benchmark tables online reliable? Treat them as vendor content with unstated samples. Published through 2026 by agencies and tool vendors, they rarely state advertiser count, period or methodology, and delivery on this channel changed materially during the year.
What should I compare my CTR against? Your own first fourteen days, then other ad groups running concurrently in the same campaign, then creative variants inside a single ad group. Internal comparison holds the variables constant that an external benchmark cannot.
Can I compare ChatGPT Ads to Google Ads? On cost per qualified lead, yes. On CTR or CPM, no. Click-through on a conversational surface and click-through on a results page are different quantities measured against different denominators.
What is a reported ChatGPT Ads CPC? Guidance circulating in 2026 suggests three to five dollars as a starting bid. One named test reported by MediaPost in August 2026 recorded roughly 7 USD per click across 415 USD of June 2026 spend at a 0.6% CTR.
How long before my own baseline is meaningful? Fourteen days gives a usable cost-per-click baseline at the 25 USD minimum. A conversion-rate baseline takes considerably longer, because a fortnight at that budget buys roughly 70 to 110 clicks.
Sources
| Claim | Source | Tier |
|---|---|---|
| Reporting is account-scoped: impressions, clicks, spend, CTR, avg CPC, avg CPM, conversions | OpenAI Ads Manager documentation, 2026 | Confirmed, primary |
| No cross-advertiser benchmarks, no auction insights, no percentile view | OpenAI Ads Manager, by absence | Absent |
| Benchmark tables by vertical published through 2026 | Tru Commerce, 360ROI, Top Growth Marketing, Adventure Media, 2026 | Vendor |
| 415 USD spend, 9,000 impressions, 60 clicks, 0.6% CTR, roughly 7 USD per click, June 2026 | MediaPost, 31 August 2026, citing Peter Jaffray of Choice OMG | Reported, third party, single test |
| Fill rates up 30 to 50 percent from launch levels by late April 2026; 2,500 USD spent of a 250,000 USD commitment for roughly 200 impressions | Digiday, 2026, anonymous sources | Reported, third party |
| One client tracked conversions close to Google non-brand search efficiency | Digiday, 2026, anonymous source | Reported, third party, single account |
| Platform click counts diverging from site analytics in both directions | MediaPost, 31 August 2026 | Reported, third party |
| InPromptAds has no first-party benchmark data | InPromptAds | Confirmed, primary |
Related reading
- The First Fourteen Days of a ChatGPT Ads Campaign, Decision by Decision
- How to Optimise a ChatGPT Ads Campaign With No Search Terms Report
- How to Measure ChatGPT Ads, and What You Cannot Measure Yet
- Do ChatGPT Ads Work?
- Raise the Bid or Improve Relevance?
Changelog
12 September 2026, v1.0. First publication. Establishes self-referential benchmarking as the only defensible method on a channel with no cross-advertiser data, tiers the circulating benchmark tables as vendor content, and states that InPromptAds has no first-party figures.
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.
