You find a winner in a test campaign and duplicate it into 5 ad sets to scale. Two weeks later someone asks which creative earned the most. Ads Manager hands you 5 rows carrying the same video: 5 spends, 5 learning histories, 5 cost-per-result figures. Sort by spend and the top row is not your best creative. It is your best funded one. The asset you want a number for has no row.

Short answer: Meta Ads Manager reports on ads, and an ad is one creative bound to one ad set, so a creative that ran in 5 places produces 5 rows. Ads Reporting can group them with the Ad creative breakdown in a pivot table, but only while every copy still points at the same creative object.

The takeaways

An ad is a creative bound to one ad set. Duplicating a winner mints new objects rather than a second placement of the same asset: 5 copies, 5 ids, 5 separate histories.
Meta's native Duplicate button creates a new post ID, and Blip's writeup on post IDs is blunt about the consequence: likes, comments and shares reset to zero. Use Existing Post keeps every copy pointed at one asset.
Rates never add. Sum spend, impressions and conversions, then recompute CTR, CPA and ROAS from the summed numerators and denominators.

Why does one creative show up as 5 rows?

Because an ad is not a file. In Meta's object model an ad is a creative bound to exactly one ad set, so the same video running in 5 ad sets is 5 ad objects with 5 ids, 5 budgets and 5 delivery histories. Ads Manager reports on those objects and has no row for the asset itself.

It happens every time you scale a winner into fresh ad sets, split a campaign by country, or run the same hook at prospecting and retargeting. Each is a good reason to duplicate. All of them fragment the thing you were measuring.

The damage lands in the sorting. Rank 5 rows by result and you are largely ranking budget. One creative on 200 euro and one on 2,000 euro across 4 ad sets are not comparable rows, and nothing in the interface says so.

Can Ads Manager group a creative across ad sets?

Yes, and most accounts never open it. In Ads Reporting you can build a pivot table and check the Ad creative box, which aggregates the numbers across every campaign and ad set using that creative and puts the thumbnail and copy beside the row. Search Engine Journal's Tim Jensen documented the path in November 2025.

The limit sits in the same article: at that breakdown level you cannot see counts for custom conversions or events. So the grouped view hands you spend, impressions, clicks and CTR, then goes quiet where most accounts keep the result they care about.

Pull it anyway. Attention metrics on a real total beat conversion metrics on a fragment, and the thumbnail column kills the naming-convention chore that spreadsheet workflows exist to solve.

Why does the grouped total break after you duplicate?

Because the Duplicate button does not hand you the same creative twice. Blip's writeup on Facebook post IDs puts it plainly: Meta's native Duplicate creates a brand-new Post ID, and likes, comments and shares reset to zero. Engagement lives on the post ID rather than on the video file you uploaded.

Hold those two facts together. Grouping works on creative identity, and the usual way to scale an ad changes that identity. Scaling and reporting turn out to be one decision, made at ad-creation time, weeks before anyone goes looking for the number.

Use Existing Post, with the post id pasted in, points every copy at one asset: one engagement counter and one groupable creative from the same click. Test it on your own account before you trust a grouped total. Take a creative you know ran in several places and see whether the report returns one row or several.

How do you add up the rows without breaking the maths?

Add the counts and recompute the rates. Spend, impressions, clicks and conversions are sums. CTR, CPA and ROAS are ratios, and ratios survive neither addition nor averaging. A creative's real CTR is total clicks divided by total impressions across every row it appears in. The mean of 5 CTR figures answers no question you asked.

The conversion column needs a caveat. Inside an attribution window the same person can be credited under more than one row, so summed conversions overstate unique buyers, and the overstatement grows with audience overlap. Same family of problem as the platform's count drifting from your store's, which I went through in why Meta conversions don't match your sales.

Treat the summed total as a ceiling on what the creative contributed, then confirm anything expensive against revenue you can see yourself.

What does a creative-level total still not tell you?

It tells you what one asset earned across everything it ran in. It does not tell you the asset is better than the one below it. The rows differ in audience, budget, placement mix and the weeks they ran, so a grouped total is the output of a comparison nobody controlled.

Which is fine, for the right job. Aggregation answers how much a creative earned. Ranking creatives against each other is a test-design question, and that answer needs the other layers to hold still, which is the argument in how to test Facebook ad creative.

Thin rows make it worse. 300 impressions and 2 conversions inside a grouped total will still print a cost per result, and it will read like a fact. Look at the spread across creatives before you read the order.

So what do you do with the number once you have it?

Build the total, then hold it loosely. It is a reasonable starting point for deciding where the next 1,000 euro goes and a weak basis for crowning a winner.

Adscalr is not a reporting tool. It does not warehouse your account data, export dashboards or produce client reports. It scores creative performance: a composite from 6 metrics (hook rate, CTR, CPI, ROAS, share rate, revenue per install) with weights you set per project and funnel stage, Bayesian shrinkage with format-specific priors so a thin row gets pulled toward its format's normal instead of reading as a verdict, and fatigue detection from a multi-day CTR slope with a market-wide check. The ad-intelligence page covers how that scoring behaves on small samples.

Creative level reporting in Meta ads