How to Make a Creative Performance Report: Step by Step

Most published advice about judging ad creative in Google Ads is now out of date.

The old workflow was to read a label. Each asset carried a rating, you replaced the low ones, and that was the report.

That column is gone. Google now gives real per-asset statistics instead, and the change makes creative performance reporting both better and easier to get wrong.

This guide covers what changed and what belongs in a creative performance report. It then covers the two official warnings that decide how you present it, and how to build the report from an export.

What changed: the performance label is gone

Google's documentation on the ad-level asset report states it plainly.

The page notes that the "Performance label" column "has been deprecated as full performance statistics for each asset are now available."

It also describes what that column used to do. It "previously benchmarked an asset's performance against others of the same type," which is the Low, Good and Best language still repeated across the web.

The replacement is better raw material. You now get impressions, clicks, cost, conversions, and conversion value per asset.

One date matters more than the rest of the page. Google notes that full performance statistics is "only available for dates on or after June 5, 2025."

So a year-over-year creative comparison can silently cross a boundary where the data changes shape. Check your date range before you trust a trend.

What belongs in a creative performance report

The asset report gives you the raw columns. A report is the arrangement that makes them safe to read.

Element	Why it is there
Asset and asset type	Headlines, descriptions, images and logos are not comparable to each other
Impressions per asset	The volume behind every other number
Clicks, cost, conversions	The outcome columns, per asset
Position pinning status	A pinned asset never competed fairly
Date range with a note	Full statistics start on June 5, 2025
A stated comparison scope	Within one ad, or within one asset group

Asset type deserves the emphasis. Google lists business logo, business name, description, headline, image, sitelink and structured snippet as the kinds you may find.

Comparing a headline against an image tells you nothing. Comparing headlines against each other inside one ad is the actual question.

Position pinning is the quiet distorter. If a headline is pinned to position one, its numbers reflect that placement rather than its quality.

Why asset numbers do not add up

This is the first of two official warnings, and it invalidates the most natural thing to do with the export.

Google's guidance is explicit under non-summable metrics. Asset-level metrics "are attributed per instance of the asset served within an ad."

The worked example is the clearest part. If a single ad impression includes three different assets, "those 3 assets will each register one impression."

So three assets, one impression, three recorded impressions. Summing the column triples the truth.

The documentation follows through on the consequence. The sum of individual asset impressions, clicks or costs "may not directly match the corresponding metrics" at the asset group level.

The practical rule is short. Never total an asset column, and never present a share-of-total built from one.

Why asset ratios are directional only

The second warning is stronger than most practitioners realise, and it comes from Google rather than from us.

Ratio metrics at the asset level, "such as CTR, CPC, CPA, ROAS, should be used as directional indicators only."

The reason is given too. Those ratios "don't accurately reflect the overall performance of a single asset in isolation." They are "influenced by the combination of assets served together."

An asset never runs alone. It appears alongside other headlines and descriptions chosen by the system, and its measured rate carries their influence.

Google's own best-practice advice follows from that. It recommends evaluating performance "at the asset group level or campaign level, rather than at the individual asset level."

That does not make a creative report pointless. It changes what the report is allowed to claim.

Rank assets against each other to decide what to replace. Do not attribute a campaign's cost per acquisition to one headline.

There is a second timing problem stacked on top. Google's conversion reporting documentation notes that primary conversion columns are "calculated based on the time of the click, not the time of the conversion."

That page also warns that discrepancies "often up to 20%, are expected due to different attribution models." It adds that reporting lags "can take up to 24-48 hours."

Short-lived creative feels this hardest. An asset paused after a week may still be collecting conversions credited back to its click dates.

So a creative performance report built on the last seven days will understate whatever ran most recently. Cut the window short of the lag period, or say in the report that you did not.

How to do it manually

Option 1: One table per asset type

Download the asset report, then split it by asset type before doing anything else.

Within each type, sort by impressions and check the volume spread. Google notes an asset "might not receive any impressions if other assets are predicted to perform better." It suggests replacing assets with zero impressions after several weeks.

Rank inside the type using SUMIFS and COUNTIFS against the asset identifier.

The ceiling is that you are ranking, not measuring. That is the correct ceiling given the official guidance, and it still answers the question of what to swap out.

Option 2: A minimum-volume filter

Add a column that flags assets below a volume threshold you set yourself, then exclude them from ranking.

Pick the threshold from your own data rather than a rule of thumb. A level that leaves you at least ten comparable assets per type is a reasonable place to start.

Then show the excluded assets separately, with their impression counts. A reader who cannot see what was filtered will assume nothing was.

The limit is that thresholds are judgement calls. Document yours in the report rather than in your head.

Option 3: A definitions tab

Record the comparison scope, the date range, the volume threshold, and which assets were pinned.

Then record what the report does not claim. Writing down that ratios are directional stops a well-meaning colleague from quoting a per-asset cost per acquisition in a board deck.

The limitation is the usual one. Documenting a rule does not apply it, and next month somebody rebuilds the filters.

The shared ceiling. All three assume you exported the same date range for every asset type. Mixed ranges across tabs is the most common silent error in this report.

Where the manual route slows down

The first creative performance report takes an afternoon. The fourth takes longer, because the creative set churned.

Assets get replaced continuously. Half of last month's headlines no longer exist, so a month-on-month table has different rows on each side.

Naming is inconsistent too. Google notes that not all assets have a name, so matching last month's rows to this month's needs identifiers rather than text.

Then the volume filter has to be reapplied by hand, along with the pinned-asset exclusions.

There is a fourth cost that only shows up under pressure. Somebody asks which creative drove last quarter's result, and the honest answer needs three caveats you now have to reconstruct.

The tooling side has its own roundup, in AI tools for ad spend analysis.

How to build it with Powerdrill Bloom

Step 1: Upload your ad exports

Upload the asset report export, or the asset and campaign exports together. Powerdrill Bloom profiles the columns on arrival, so missing asset identifiers, mixed date ranges, and blank asset types surface before any ranking is produced.

Step 2: Describe the report in natural language

State the rules rather than building them. Name the comparison scope, the asset types to separate, the volume threshold, and the pinned assets to set aside.

Then ask the questions that catch the errors. Ask which assets fall below your threshold. Ask which asset identifiers appear in only one of the two periods. Then ask for the ranking within each asset type, without summed totals.

Step 3: Export the chart, report, or deck

Take out the ranked table per asset type, a chart of impression share within a type, or slides that carry the caveats beside the ranking.

Common mistakes

Summing asset columns. One impression can register against three assets. Totals from that column are inflated by design.

Quoting a per-asset cost per acquisition. Google labels asset-level ratios as directional only, because assets never serve alone.

Comparing across asset types. A headline and an image are not competitors. Rank within a type.

Ranking assets with almost no impressions. Low-volume assets produce unstable rates. Set a threshold and show what it excluded.

Ignoring pinned positions. A pinned asset did not compete on equal terms. Flag it or exclude it.

Reading yesterday's numbers as final. Reporting lags can run 24 to 48 hours, and conversions land against earlier click dates.

Trusting advice about Low, Good and Best labels. That column is deprecated. The current report gives statistics instead.

Conclusion

Split by asset type, rank within the type, set a volume threshold, flag pinned assets, and state the date range. That produces a creative performance report someone can act on.

What the report may not do is attribute outcomes to a single asset. Google says so directly, and honouring that is what keeps the report credible.

The useful output is a replacement list, not a leaderboard. Knowing which three headlines to swap out is worth more than a ranking nobody can defend.

If rebuilding it every cycle eats a day, try Powerdrill Bloom on your ad export. See also the AI graph maker and AI report generator pages.

Frequently asked questions

Do Google Ads assets still have Low, Good and Best ratings?

No. The performance label column is deprecated, and full performance statistics for each asset are available instead.

Can I add up asset-level impressions?

No. Google states that each asset served in an impression registers its own impression, so the column does not sum to the ad's total.

Can I calculate cost per acquisition for one headline?

You can compute it, but Google labels asset-level ratios as directional indicators only. They are shaped by the other assets served alongside.

How far back does asset-level data go?

Full performance statistics are available for dates on or after June 5, 2025. Check that your comparison range does not cross that date.

What should I do with an asset that gets no impressions?

Google suggests replacing assets that receive zero impressions after several weeks, since the system may be predicting others will perform better.