Turn last month's winners and tiring ads into a brief that tells designers which themes, formats and hooks to develop, plus what they should stop making. A creative brief from ad performance data should connect every request to evidence, state what must remain fixed and define the question the next asset is meant to answer.
The job is not to hand the creative team a dashboard. It is to turn account signals into useful direction without pretending the numbers explain more than they do.
A high-performing ad might owe its result to the hook, format, offer, audience or placement. A tired ad may still contain a strong idea. The brief should separate those possibilities and request creative that helps the buyer learn what is actually working.
Read the account as a creative system
Creative reporting often stops at a list of winners and losers. That is useful for deciding what stays live, but it is not enough to tell a designer what to make next.
The designer needs to know which parts of the work deserve another test. The media buyer needs new assets that answer a clear account question. Both need a shared language for moving from performance to production.
Group assets before judging them
Start by grouping ads according to the decisions the creative team can act on:
- Theme or message, such as convenience, control, savings or proof
- Hook, meaning the opening idea, line or visual
- Format, such as a static image, short video, carousel or testimonial
- Offer and callout
- Visual treatment
- Audience angle
- Landing page or destination
- Platform and placement
Do not assume that ads with different names are different creative ideas. The same asset may be running in several campaigns or on several platforms. Judge the asset as a whole, then examine where its performance changes.
This prevents duplicated ads from looking like independent winners. It also helps the team distinguish a durable creative idea from an isolated campaign result.
Compare like with like
A static prospecting ad and a retargeting video may have different jobs. Their raw CPA figures do not tell you which creative concept is stronger.
Before drawing conclusions, check whether the assets served comparable audiences, objectives, offers and placements. Note material differences directly in the brief.
The point is not to remove every variable. That is rarely possible in a working account. The point is to avoid giving a designer false certainty.
Separate winners, fatigue and weak ideas
Use three practical labels:
- Extend: The asset or idea is producing a useful signal and deserves controlled variations.
- Refresh: The core idea may still be sound, but delivery or efficiency is deteriorating.
- Stop: The concept has had a fair chance and does not merit more production without a new reason.
Fatigue needs context. Rising frequency paired with rising CPM or CPC and worsening CPA is more useful than frequency alone. The response is usually a refresh of the delivery, hook or visual treatment, not an immediate rejection of the whole message.
A weak idea is different. If several executions of the same concept fail under comparable conditions, the brief should tell the team to stop producing minor variations of it.
Turn performance signals into creative instructions
A useful creative brief from ad performance data does not say, “Make more like the winner.” It says what appears to be working, what evidence supports that view and what the next asset should test.
Use this sequence for each request.
State the observation
Describe what happened in plain language.
For example:
Example: The direct product demonstration held its CPA closer to the campaign target than the lifestyle concepts. Its CPC also stayed steadier as delivery increased.
This is an observation, not a verdict. It does not claim that the demonstration caused the difference.
Name the working theory
Explain what you think may account for the result.
Example: The demonstration may reduce the amount of explanation needed before the click. The opening frame shows the product in use and makes the offer easier to understand.
A theory gives the designer something concrete to work with. It also makes uncertainty visible.
Turn the theory into a test
Ask for a variation that isolates the idea.
Example: Create a new demonstration video that keeps the same offer and landing page. Change the opening hook and visual setting, but show the product in use immediately.
Now the assignment has a purpose. The team is not merely cloning the current winner.
Define what must not change
Every brief needs guardrails. If too many parts move at once, the buyer may get another asset but no useful answer.
State the fixed elements:
- Offer
- Landing page
- Audience
- Core proof point
- Required disclaimer
- Brand requirement
- Intended placement
- Any message the account must avoid
Then state where the designer has room to explore. Good briefs constrain the experiment without prescribing every frame.
For a broader testing process, the creative testing framework for agencies explains how to carry one method across multiple accounts.
Build the one-page brief
A one-page creative brief from ad performance data should be readable in a few minutes. Supporting screenshots and account exports can sit behind it. The main page should contain the decision.
Use the following template.
Creative request
Account or property:
Name the account or account group.
Business goal:
State what the campaign is trying to produce.
Campaign role:
Clarify whether the creative is intended for prospecting, retargeting, offer support or another defined job.
What the account is showing:
Summarize the relevant performance signal. Name the asset, period, campaign context and material differences between the compared ads.
Working theory:
State why the winning or tiring asset may be behaving that way. Mark this as a theory.
What to extend:
Identify the message, hook, proof device, visual pattern or format worth carrying forward.
What to change:
Name the variable the new creative should test.
What to stop making:
List concepts or variations that should not return unless the strategy changes.
Required format and placement:
Specify the intended asset type and where it needs to run.
Offer and landing page:
Give the exact current offer and destination. Include an expiration date when the offer has one.
Fixed elements:
List the elements that must remain stable for the test.
Creative freedom:
State what the designer may interpret or explore.
Success measure:
Name the campaign target the buyer will use to evaluate the test.
Review note:
Explain when the media buyer expects to have enough information to review the result. If the account cannot produce a clean answer, say that in advance.
This format forces the buyer to do the interpretation before production begins. It also gives the designer the context needed to solve the problem, rather than simply reproducing an existing ad.
Worked example: From a tiring winner to the next request
Consider an illustrative account where a campaign has a $50 target CPA. A short demonstration video has been the strongest asset, but frequency is rising, CPC is moving up and CPA is drifting above the target. A static image using the same offer has produced weaker traffic from the start.
The wrong brief would be:
Example: We need more videos like the winning ad. Make a few new versions and try some new hooks.
That instruction hides the useful evidence. It also gives the designer no boundary between the proven idea and the parts that need to change.
A stronger brief would read like this:
Example brief
Business goal:
Generate qualified conversions near the campaign's $50 target CPA.
What the account is showing:
The product demonstration video has been the most useful creative direction in this campaign. Its frequency is rising, CPC has increased and CPA is getting worse. The static image using the same offer did not produce the same quality of traffic.
Working theory:
The demonstration remains useful because it explains the product quickly. The specific opening and visual sequence may be tiring after repeated exposure.
What to extend:
Keep the immediate product demonstration and the current offer.
What to change:
Test a new opening hook, presenter and setting. Show the product in use at the start.
What to stop making:
Do not request more static variations of the existing layout for this round. They would not answer the current question.
Fixed elements:
Keep the audience, offer, destination and core product proof stable.
Creative freedom:
The designer may change the visual setting, pace, on-screen text and presentation style.
Success measure:
The media buyer will judge the new asset against the campaign target and the original demonstration under comparable delivery conditions.
The new request protects the useful idea while replacing the elements most likely to be tired. It also tells the designer why another static version is not the current priority.
Keep production tied to account questions
Each creative batch should have a reason to exist. One may test whether the winning message transfers into a new format. Another may test whether a new hook can refresh a proven format.
Do not change the message, offer, format, hook and landing page at the same time unless the goal is broad exploration. If the goal is learning, keep enough stable to read the result.
The number of variations also matters because each asset needs enough delivery to produce a useful signal. The related guide on testing new creative without starving your winners covers that production and budget tradeoff.
Close the loop with the designer
When results are available, return to the brief and add a short outcome note:
- What happened
- Whether the result supported the working theory
- What remains uncertain
- What should be repeated
- What should be dropped
- What the next test should isolate
“Inconclusive” is a valid result. It means the account did not produce enough evidence to support a call, or the test changed too many variables to identify the cause. That is more useful than forcing a winner from weak evidence.
Frequently asked questions
What should an ad creative brief include?
Include the business goal, audience, offer, performance evidence, message, hook, format, required elements, constraints and success measure. Separate fixed requirements from areas where the designer can explore.
How do you use performance data in a creative brief?
Translate account signals into creative instructions instead of pasting a dashboard into the document. Explain which asset, theme, hook or format produced the signal, what may be causing it and what the next version needs to test.
How often should you brief new creative?
Brief new creative when the account has enough evidence to support a useful decision, or when current assets show clear fatigue. Use a regular review rhythm, but do not manufacture a new request when the data has not changed.
How do you tell a designer what is working?
Show the designer the winning asset and name the element you believe matters, such as the opening hook, proof device, offer framing or format. Also explain what remains uncertain so the next version tests the idea instead of merely copying the ad.
What creative formats should you test next?
Choose formats that test a specific performance question and fit the platform placement. Adapt a proven message into a new format when you want to test delivery, or keep the format stable when you want to isolate a new hook or offer.
How Ads Arsenal supports the brief today
Ads Arsenal finds the same creative running across campaigns and platforms, then scores it as one asset. It ranks winners and flags tiring creative when signals such as frequency, CPM or CPC rise while CPA gets worse, with a recommendation to swap in creative you have staged. Nothing changes in an account until someone approves it.
It also reads landing pages and offers, so copy suggestions match what the business sells, and it remembers when an offer expires. Ads Arsenal is in a free design-partner beta for agencies and in-house teams running real spend on Google, Meta and TikTok: Become a design partner.