How to Read Your ChatGPT Ads Report (and What to Do Next)
A metric-by-metric guide to reading ChatGPT ads reports, with diagnostic patterns, worked math and a weekly review checklist.
A ChatGPT ads report is only useful if it changes what you do next week. Most advertisers open the dashboard, see a CTR and a cost per click, and close it without a decision. The numbers are there; the interpretation is missing.
This guide walks through every metric you will typically see in a ChatGPT ads report, explains what healthy and unhealthy patterns look like, and maps each pattern to a specific action. It works whether you read reports in OpenAI's Ads Manager yourself or receive them from a partner.
The metric stack, from top to bottom
Think of your report as a funnel. Each metric depends on the one above it, and problems usually live at one specific layer.
| Layer | Metric | What it tells you |
|---|---|---|
| Delivery | Impressions, spend, CPM | Whether your ad is finding conversations to appear in |
| Attention | Clicks, CTR, CPC | Whether the ad is relevant to the people who see it |
| Arrival | Landing page views, bounce, time on page | Whether clicks reach a page that matches the promise |
| Action | Add to cart, sign up, lead, purchase | Whether the page and offer convert |
| Economics | CPA, ROAS, revenue | Whether the whole thing makes money |
Read the report top to bottom. If delivery is broken, nothing below it matters yet.
Delivery metrics
Impressions and spend
Impressions count how often your chat card was shown. If spend is far below your daily budget, your ad is not finding enough matching conversations. The usual causes are context hints that are too narrow, bids below what the auction needs, or ad review issues.
CPM
CPM is cost per thousand impressions. Third-party observers have reported ChatGPT CPMs roughly in the $25–45 range, which is higher than commonly cited Meta ranges. A high CPM is not automatically bad. ChatGPT inventory sits next to a person actively asking for help, so higher prices can be justified if downstream metrics hold. Check your own numbers with the CPM calculator.
What to do if delivery is low: broaden or add context hints, raise your bid modestly (10–20%), and confirm every ad is approved.
Attention metrics
CTR
Click-through rate is clicks divided by impressions. Similarweb has reported an average ChatGPT ads CTR around 0.68%, with top brands near 1.57%. Use these as rough reference points, not targets; your category and message will shift them.
A low CTR usually means one of two things: the ad is appearing in the wrong conversations, or the message does not connect with the people seeing it. To tell which, compare CTR across campaigns with different hints. If one hint set has double the CTR of another with the same ad, the problem is placement, not copy.
CPC
Cost per click is spend divided by clicks. OpenAI's recommended starting bids work out to roughly $3–5 per click. If you bid on CPM, your effective CPC equals CPM divided by (CTR × 1,000). So a $35 CPM at 0.7% CTR gives an effective CPC of $5.00. At 1.0% CTR, it drops to $3.50. Improving CTR is often the fastest way to cut CPC.
What to do if CTR is low: test a new message before rewording the headline, tighten context hints toward higher-intent situations, and check that the image supports the headline. Our guide on writing ChatGPT ad headlines has examples.
Arrival metrics
These usually come from your own analytics (GA4, Shopify, or similar), connected via UTM parameters on your landing URL.
Landing page views vs clicks
If Ads Manager reports 200 clicks but your analytics shows 120 landing page sessions from ChatGPT, something is leaking. Common causes are slow page loads, redirects that strip UTMs, or consent banners that block tracking. A gap of 10–20% is common across ad platforms; a gap over 30% deserves investigation.
Bounce and engagement
High bounce from ChatGPT traffic often signals a message mismatch. The ad promised one thing and the page leads with another. Someone who clicked "A three-step routine for sensitive skin" and lands on a homepage hero about a summer sale will leave. The message match checker can flag those gaps.
What to do if arrival is weak: fix tracking first, then align the landing page headline with the ad, then speed up the page.
Action metrics
These come from the OpenAI pixel and Conversions API, which report events such as page view, add to cart, sign up, lead and purchase back to Ads Manager. Automatic advanced matching rolled out in August 2026, improving how conversions are matched to ad clicks.
Conversion rate
Conversion rate is conversions divided by clicks. Criteo reported in early 2026 that traffic from ChatGPT converted around 1.5x better than other channels, which suggests the intent behind these clicks can be strong. Your rate depends on price, offer and page quality.
Look at each step, not just purchases. If add-to-cart rate is healthy but purchases are low, the problem is checkout, shipping cost surprises, or payment friction. If add-to-cart is low, the product page or offer is not convincing.
What to do if conversions are weak: compare add-to-cart and purchase rates to your site averages from other paid channels. If ChatGPT traffic underperforms at the product page, build a landing page that answers the buyer's likely question directly. See landing pages that convert ChatGPT ad traffic.
Economics metrics
CPA
Cost per acquisition is spend divided by conversions. Compare it to the maximum you can afford to pay for a customer, which depends on margin and lifetime value.
ROAS
Return on ad spend is revenue divided by spend. Before you judge ROAS, know your break-even ROAS: 1 divided by your gross margin. With a 60% margin, break-even ROAS is about 1.67. Anything above that is profitable on the first order. You can calculate yours with the break-even ROAS calculator.
A worked report, diagnosed
Here is a hypothetical two-week report for a brand selling a $55 sleep mask, with a 65% gross margin.
| Metric | Campaign A | Campaign B |
|---|---|---|
| Spend | $700 | $700 |
| Impressions | 21,000 | 19,500 |
| CPM | $33.33 | $35.90 |
| Clicks | 168 | 107 |
| CTR | 0.80% | 0.55% |
| CPC | $4.17 | $6.54 |
| Add to carts | 17 | 9 |
| Purchases | 8 | 3 |
| CPA | $87.50 | $233.33 |
| Revenue | $440 | $165 |
| ROAS | 0.63 | 0.24 |
Break-even ROAS here is 1 / 0.65, about 1.54. Neither campaign is profitable yet on first-order revenue, but they are not equally unprofitable.
Campaign A has a healthy CTR and reasonable CPC. Its add-to-cart rate is about 10% of clicks, but only 8 of 17 carts became purchases. That points to a checkout or offer issue. Next step: check shipping cost presentation and consider a bundle or free-shipping threshold to lift order value.
Campaign B has a weak CTR, which pushes CPC above $6.50. Its problem is upstream. Next step: pause B, move its budget to A, and test a new message in B's slot.
A key point: two weeks and 11 total purchases is still a small sample. You are looking for direction, not final verdicts. If you also sell repeat purchases, your real return may look better once you factor in lifetime value.
The weekly review checklist
Spend 20 minutes a week on this, in order:
- Is each campaign spending at least 80% of its budget? If not, fix delivery.
- Which campaign has the lowest CTR? Is it a hint problem or a message problem?
- Do ad clicks and landing page sessions roughly agree?
- Where is the biggest drop in the funnel: click to cart, or cart to purchase?
- What is CPA versus your target, and ROAS versus break-even?
- What is one change to make this week, and what result would confirm it worked?
Write the answer to question six down. Next week, check it first.
Reading reports from a partner
If someone else manages your campaigns, ask for the same funnel view above, not just headline ROAS. A good report shows spend, CTR, CPC, conversion steps and what was changed since the last report. Because ad spend sits in your own OpenAI ad account, you can always verify the numbers in Ads Manager directly.
How SecondWin reports on your ChatGPT ads
SecondWin runs your campaigns in your own OpenAI ad account, so every number in Ads Manager is yours to see. Depending on your plan, it refreshes creative and reports monthly, biweekly or weekly, covering what ran, what changed and what comes next. The ads themselves start from proven messages found in Meta's Ad Library and are policy-checked before launch.
Get a free analysis of your website to see the messages SecondWin would test first, and compare refresh cadences on the pricing page.
FAQ
What is a good CTR for ChatGPT ads?
There is no official benchmark yet. Similarweb has reported an average around 0.68%, with top-performing brands near 1.57%. Use that range as a loose reference. Your CTR depends heavily on how well your context hints match real buyer situations and how directly your message answers their need. Compare campaigns against each other in your own account rather than chasing an industry average.
Why do my ChatGPT ads clicks not match my analytics sessions?
Small gaps are normal on every ad platform because some people leave before the page loads or block tracking scripts. Larger gaps, above roughly 30%, usually come from slow pages, redirects that drop UTM parameters, or consent tools blocking analytics. Check that your landing URL loads quickly, keeps its UTMs through any redirects, and that the OpenAI pixel fires on page view.
How often should I check my ChatGPT ads report?
Glance daily for problems like disapprovals or spend stopping, but make decisions weekly. Daily numbers on a $50 to $100 budget are too noisy to act on. A weekly review lets you see a full cycle of weekday and weekend behavior. Make one meaningful change per week and note what you expect it to do, so the following report actually answers a question.
What ROAS should I expect from ChatGPT ads?
It depends on your margin, price and offer, so start by calculating your break-even ROAS, which is one divided by your gross margin. Early campaigns often run below break-even while you test messages and hints. Judge progress by whether CTR, conversion rate and CPA are improving week over week, and include repeat purchases if your customers buy more than once.