A reporting breakdown, or segment, splits your ChatGPT Ads performance metrics by a single dimension such as device, country, or platform, so one blended total becomes several comparable rows. It applies to numbers you already have and reshapes how they display; it is separate from the conversion-event columns you add through the columns editor.
When you decide where to shift spend, a segment tells you whether mobile or desktop, or one country over another, is actually carrying the campaign. Skip it and you optimize against an average that hides a losing slice quietly draining budget behind a healthy-looking headline.
In ChatGPT Ads Manager, a reporting breakdown is the Segment control that divides a campaign's aggregate metrics into rows by one dimension at a time. Instead of a single impressions or spend figure for the campaign, you see that figure distributed across the segment values. The breakdown changes how the data is grouped; it does not change the underlying measurement.
A breakdown has three parts: the metric being split, the dimension you split it by, and the reporting level you apply it at. ChatGPT Ads exposes a small set of dimensions, centered on device, country, and platform, and applies them at the campaign, ad group, or ad level. The platform does not break results down by query, keyword, placement, or audience demographics, because OpenAI treats that user-level detail as private.
A segment sits next to two other reporting controls. The columns editor adds metrics and conversion events to each row, while the Segment menu slices whatever rows exist by a dimension. AirOps helps teams fold these thin platform segments into a wider view of how the same audiences find and act on their brand across AI search.
Resources: see how to structure and measure ChatGPT Ads campaigns for cleaner reporting
A breakdown runs the same way every time you open the Segment menu in Ads Manager.
Pick a level: Choose whether you are reporting at the campaign, ad group, or ad level, since the segment applies to whatever rows that view shows.
Open Segment: Select the Segment control, which is separate from the columns editor and the conversion-event settings.
Choose a dimension: Apply one dimension, such as device, country, or platform; ChatGPT Ads splits by a single dimension at a time.
Read the split rows: Each original row expands into one row per segment value, with every metric recalculated for that slice.
Export or compare: Pull the segmented table into a CSV, or line it up against your own analytics to sanity-check the numbers.
The output tells you how performance differs across a device, country, or platform cut. It does not tell you what prompt triggered the ad or who saw it, because ChatGPT Ads never reports query or audience data.
ChatGPT Ads gives you one blended number per metric unless you segment it, and that blended number is what you take into every budget conversation. Breakdowns decide whether you are reallocating spend on evidence or on a hunch, because they are the only native way to see which slice of a campaign is working.
They expose hidden losers: Without a device or country split, a campaign that looks profitable overall can be quietly subsidizing a segment that never converts, and you keep funding it.
They sharpen reallocation: Seeing spend and conversions side by side per segment lets you move budget toward the slices that pay back instead of spreading it evenly.
They set honest expectations: Because ChatGPT Ads withholds query and audience data, segments are the deepest cut you can report to a client, so they frame what the channel can and cannot prove.
SEO managers use reporting breakdowns to check whether desktop or mobile users click their ChatGPT Ads before committing more budget to either device.
Growth marketers use segments to compare conversion performance across countries and cut spend in regions that never pay back.
Demand gen leads use platform breakdowns to reconcile ChatGPT Ads reporting against their own analytics before presenting results to finance.
A dimension is the attribute a breakdown splits by, such as device, country, or platform in ChatGPT Ads, and you can apply only one at a time, so each report answers a single slicing question before you reset and run the next one from scratch.
The reporting level is the hierarchy tier the segment applies to, campaign, ad group, or ad, and the same dimension produces coarser or finer rows depending on which tier you choose to view it at inside Ads Manager.
The Segment menu reshapes existing rows by a dimension, while the columns editor adds metrics and conversion events, so knowing which control does which keeps you from hunting for a breakdown in the wrong menu entirely and wasting time.
Isolate which device or country is driving conversions in ChatGPT Ads.
Reallocate budget toward segments that pay back instead of averaging everything.
Catch underperforming slices before they drain a full reporting period of spend.
Export segmented tables to CSV for reconciliation against Google Analytics 4.
Report performance to stakeholders with the deepest cut the platform allows.
Compare ChatGPT Ads segment trends against your other paid channels when planning.
Segment one dimension at a time: Apply device, then country, then platform separately, because ChatGPT Ads will not stack two dimensions into a single table.
Match the level to the question: Run the breakdown at ad group level to test themes, and at campaign level to judge allocation.
Export before you compare: Pull the segmented CSV so you can hold the numbers steady while you reconcile them against your own analytics.
Wait for attribution to settle: Give conversions a day or two to populate before you read a segment, since early numbers understate slower-converting slices.
Tie segments to a decision: Only run a breakdown you will act on, so each report ends in a budget move and not just a saved screenshot.
Avoid treating the Segment menu as a conversion report. The most common mistake competent advertisers make is opening Segment to split conversions by event type, finding nothing there, and concluding the data is missing. Event-level detail lives in the columns editor; the Segment menu only cuts by device, country, and platform.
AirOps: Brings your ChatGPT Ads segment data together with AI search visibility so you can see how the same devices and regions find your brand beyond paid placements.
ChatGPT Ads Manager: OpenAI's native beta console, where the Segment menu and columns editor live and where every breakdown is actually run.
Google Analytics 4: Lets you cross-check segmented ChatGPT Ads numbers against your own device, geography, and conversion data for reconciliation.
Open Ads Manager: Log in to your ChatGPT Ads account this week and find the table view on the Campaigns page. No new budget or approval is needed to look at what you already spend each day.
Confirm conversion tracking: Make sure the OpenAI pixel or Conversions API is live, so your segmented rows show conversions and not just clicks and spend.
Run your first split: Open the Segment menu and break results down by device, then read how clicks and conversions differ across the slices before you touch any settings.
Add country and platform: Repeat the split for country and platform one at a time, and note which cut reveals the sharpest performance gap worth acting on.
Turn reads into moves: Reallocate budget toward the segments that convert, then schedule a repeat read to confirm the shift worked and did not just move the problem elsewhere.
A reporting breakdown splits ChatGPT Ads metrics by a single dimension so one blended total becomes comparable rows.
You run a breakdown through the Segment menu, choosing a level first and then one dimension such as device, country, or platform.
The main constraint is depth: the platform never segments by query, keyword, placement, or audience demographics.
The main risk is reading only the blended average and missing a slice that loses money every day.
The leverage sits in acting on each cut, moving budget toward segments that convert and away from ones that do not.
A reporting breakdown and the columns editor solve two different problems, and mixing them up is the most common reporting mistake on the platform. The columns editor controls which metrics and conversion events appear as columns in your table, so it is where you add a specific conversion event, separate click-through and view-through conversions, or change an attribution window. The Segment menu, which is where breakdowns live, leaves the columns alone and instead splits whatever rows you already have into one row per segment value. If you want to see conversions broken out by individual event, you reach for the columns editor. If you want to see every metric split by device or country, you reach for the Segment menu. Advertisers routinely open Segment expecting event-level conversion detail, find only device, country, and platform, and wrongly conclude the data is broken. Knowing which control owns which job saves that confusion and a support ticket.
Check reporting breakdowns on a cadence that matches your decisions, which for most advertisers means weekly, with a monthly reconciliation against your own analytics. Daily checks rarely help, because ChatGPT Ads needs a day or two for attributed conversions to populate, so a segment you read the morning after launch will understate slower-converting slices. A weekly read gives each slice enough volume to be meaningful while still letting you react before a losing segment burns through much budget. Once a month, pull the segmented export and line it up against your conversion data to confirm the platform and your analytics agree. If a campaign is small or spending slowly, stretch the cadence out, since a segment with only a handful of clicks per row tells you almost nothing. Let the volume in each row, and the size of the decision riding on it, set how often you look.
Reporting breakdown numbers vary from your analytics because the two systems count differently, and neither is strictly wrong. ChatGPT Ads attributes conversions on a last-touch basis within the windows you set, and it applies those windows to the segment rows, so a conversion credited to a device in Ads Manager may land in a different bucket in your own tool. Analytics platforms also lose some ChatGPT ad clicks to direct or unattributed traffic when referrer data is stripped, which pulls a segment's numbers in your tool below what Ads Manager reports. Timing adds another gap, since the platform needs a day or two to finalize attributed conversions while your analytics may log them instantly. The fix is to treat Ads Manager as the system of record for spend and platform-side conversions, use your analytics for downstream behavior, and reconcile monthly instead of expecting the two to match row for row.
Not directly; you cannot add dimensions or create custom segments, so the honest answer is that you influence a breakdown only through the campaign choices feeding it. ChatGPT Ads fixes the dimensions available, centered on device, country, and platform, and gives no setting to split by anything else. What you can control is upstream. If you want a cleaner country breakdown, structure campaigns or ad groups by region so each segment maps to a deliberate target. If device performance matters, make sure your landing pages work well on mobile and desktop before you read the split, because the segment will reflect that experience. You also control conversion tracking, and a breakdown with no conversion data is just clicks and spend. So treat the breakdown as a mirror of decisions you already made, and shape those decisions instead of hoping for a reporting control that does not exist.
A good reporting breakdown is one that changes a decision, and that is the only benchmark worth holding it to. There is no industry-standard conversion rate or cost-per-click you should expect from a device or country segment, because performance depends on your offer, your landing pages, and the markets you serve. A useful starting point is to compare each segment against your own blended campaign average: a slice spending well above its share of conversions is a candidate to cut, and a slice beating the average is a candidate for more budget. The segment also needs enough volume to trust, so a row with a handful of clicks is noise and carries no reliable signal, no matter how good or bad it looks. In practice, a strong breakdown has stable rows over a couple of weeks, enough spend per row to be meaningful, and a clear gap between your best and worst slices that points to a specific action.