Diagnose creative fatigue before you refresh the wrong thing

A falling chart has more than one cause

When cost rises or response falls, teams often ask for new ads immediately. Fresh work can help, but it can also conceal the signal. Did the same people receive repeated exposure? Did a competitor change the auction? Did the offer expire, the landing page slow down, tracking change or the audience pool narrow? The work may be fine; the delivery conditions may no longer be.

Google’s video guidance makes the distinction plainly: a rising cost per view can be a sign of fatigue, but can also reflect greater auction pressure. Its target-frequency documentation also warns that campaign reports may not include exposure from the same creative used elsewhere. Treat frequency as a clue with scope, not a universal diagnosis.

Source context: Google Ads Help — Tips for optimizing your Video campaign · Google Ads Help — About Target frequency

Read four signals together

Start with a fixed comparison window and split the view by audience, placement, geography and creative ID where the platform permits. A single account average blends too much. You are looking for a coherent change, not a dashboard score that tells a comforting story.

If frequency rises within a stable audience while view or interaction quality drops for one long-running asset, a creative refresh is plausible. If costs worsen across all assets while frequency is steady, inspect auction and inventory conditions. If only a new asset changes, allow for its learning and serving history before declaring it a loser. If clicks remain but qualified actions disappear, inspect the destination and event pipeline.

• Exposure: frequency distribution, reach and overlap with other campaigns—not only campaign average frequency.
• Response: view quality, completion, qualified action and complaint or hide signals by asset.
• Delivery: CPM, CPV or CPA, placement mix, bid and budget changes, plus major market events.
• Integrity: landing-page availability, offer validity, tracking changes and the time a new asset has had to learn.

Refresh a proposition, not just a thumbnail

When the evidence favors a refresh, do not manufacture cosmetic variation. Decide what audience tension the current work has exhausted: recognition, product proof, seasonal relevance, social proof, a demonstration or the opening beat. Make one new route answer that gap while retaining the brand codes people need to recognize. A new crop of the same weak claim is volume, not learning.

Google’s current Demand Gen guidance recommends gradual changes for evergreen campaigns and notes that high volumes of new assets can lack enough spend for reliable exploration. That fits an ordinary production truth: protect a control route, introduce a meaningful challenger, and give the system and audience enough time to reveal a difference.

Source context: Google Ads Help — Demand Gen creative assets refresh guidance

Use a reversible refresh card

Our recommended card names the suspected cause, evidence, chosen intervention, protected control, review date and rollback owner. Example: suspected cause—repeat exposure among warm visitors; evidence—frequency up and completed useful actions down for one asset only; intervention—new opening that demonstrates the product sooner; control—the current asset remains in a limited arm; review—after the agreed observation window.

That card prevents a team from replacing everything at once and losing the ability to diagnose the next shift. It also respects the possibility that the best fix is a cap, an audience change or a repaired landing page rather than another production sprint.