In short: "the ads already work" is the most expensive belief in Google Ads, and it hides two errors: "it works" only proves it clears the minimum (without a rival you can't know if it's good), and even if it were the best today, five forces wear it down — competition, fatigue, language, searches and seasonality. There is no such thing as a good ad: there is the reigning champion, defending the title.
The line is heard in almost every account: "the ads already work, don't touch them". It is the most expensive belief in Google Ads, and it is so because of two separate errors worth pulling apart.
First error · Does "it works" mean it's the best possible ad?
The fact that an ad converts only proves that it clears the minimum bar: it says nothing about how much money it leaves on the table. Without a rival to measure against, "it's doing fine" is an empty phrase: fine compared to what? The only way to know whether your ad is good is to pit it against alternatives, again and again. Every month without a test is a month assuming, with no evidence, that you wrote the best possible ad on your first try.
In twenty years of real accounts we have never once seen the first ad of a campaign still be unbeatable two years later; but we have seen hundreds of accounts convinced of it.
Second error · Why does a good ad stop being good?
An ad does not perform in a vacuum: it performs against a market. And the market moves even when you don't:
| Force | What happens |
|---|---|
| Competitors iterate | Your rivals launch new offers, prices and messages. Your ad competes today against different rivals from a year ago: its relative performance changes without you changing a comma |
| Users wear out | Someone who searches for your product several times ends up seeing the same copy over and over. Novelty catches the eye; repetition becomes invisible |
| Language goes stale | Offers that expire, years that get old («leaders since 2019»), arguments the competition already copied and devalued, words the sector burns out by repeating them |
| Searches change | With match types getting broader and AI Max, the variety of terms that trigger your ad keeps growing: copy that matched 2023 searches matches 2026 searches worse |
| Seasonality rules | The argument that converts in January is not the one for August. The eternal ad speaks the same way to every month of the year |
What is the reigning champion?
From this comes the mental model of the module: in an ad test there is never a definitive winner. There is only the best ad so far — the reigning champion, obliged to defend its title every month against a new challenger. The day you stop putting challengers in front of it, the champion starts losing to a rival you cannot see: time.
What did RSAs change about how you test?
With classic text ads, "testing" meant pitting ad A against ad B. RSAs broke that scheme: a single ad takes up to 15 headlines and 4 descriptions, and Google combines them on the fly — more than 40,000 possible combinations per ad. That turns each headline and each description into an asset with its own performance that can be measured, replaced and improved separately.
The unit of testing is no longer "the ad": it is the pieces that make it up. That is an enormous opportunity —each round can change one piece without touching the rest— and a workload that was already unmanageable before it got multiplied by fifteen. Hence the three figures that sum up the state of the market:
- 15 + 4 headlines and descriptions per RSA: each one, a measurable asset.
- +40,000 possible combinations in a single responsive ad.
- 0 active tests in most of the accounts we audit.
How do you know if your account has the eternal ad?
Three two-minute checks:
- Creation date of the ads in your five main ad groups («Created» column). More than six months without a new ad = eternal.
- Asset report for those RSAs: are there pieces labelled «Low» with thousands of impressions that nobody has replaced?
- Change history filtered by ads: when was the last copy change? And the one before that?
If all three answers make you uncomfortable, the rest of the module is for you.
💡 Ninja trick: the RSA Optimizer keeps a champion and a challenger in every ad group that matters: it spots groups with few ads, low strength or exhausted pieces, and proposes the challenger. The champion is never safe — and that is exactly the point.
What you should remember
- "It works" only proves it clears the minimum bar; with no rival there is no way to know whether it is good.
- Five forces degrade a static ad: competition, wear-out, language, searches, seasonality.
- There is only the reigning champion, defending its title every month.
- With RSAs, the unit of testing is the piece, not the ad.