Google gives you fifteen headlines and four descriptions for a search ad. Most of the accounts we audit are running four or five headlines and two descriptions, and the person who built them thinks the ad is finished.
It is not finished. It is a form with most of the fields left blank, and the blank fields are not neutral. They are the difference between an ad Google can optimize and an ad Google can only serve.
An RSA Is a Pool, Not an Ad
The format catches people out because it looks like the old one. You are filling in headlines and descriptions, so it feels like you are writing an ad.
You are not. You are stocking a pool.
A responsive search ad holds up to fifteen headlines at 30 characters each and four descriptions at 90 characters each. Google assembles them into combinations at auction time, typically showing three headlines and two descriptions, and works out over time which arrangements perform. Every ad served is a different draw from your pool.
Which means an ad with five headlines is not a smaller version of an ad with fifteen. It is a machine with far fewer moves available to it. You have not written a leaner ad, you have handed the system almost nothing to test and then judged it on the result.
The arithmetic is unforgiving. Three headline slots drawn from fifteen assets is a large number of possible arrangements. Three drawn from five is a handful. When people say an RSA "did not work," what usually happened is that it was never given enough distinct material to find a winner, and it settled quickly on the least-bad option available because that was the only option available.
The corollary matters just as much. The pool is only useful if the things in it are genuinely different from each other. Fifteen restatements of the same sentence is technically a full pool and practically the same five-headline ad with extra steps.
The descriptions deserve the same attention and rarely get it. Four slots at 90 characters, and most accounts run two, usually written as one long sentence about the company. Two of those slots exist to carry the things a headline is too short to say: the qualifying detail, the proof, the specific promise about what happens after the click.
The Job Nobody Assigns to Ad Copy
Here is the part that changes how you write, and it is the idea in this chapter most worth carrying away.
Ad copy has two jobs. The obvious one is to attract the right click. The one almost nobody uses it for is to repel the wrong one.
We audited a manufacturer selling clinical rehabilitation equipment to physical therapy practices. Their search terms were full of consumer queries, people looking for exercises and home recovery advice, and every one of those clicks cost money and could never become a sale.
Negative keywords deal with that after the fact. Ad copy can deal with it before the click, because the searcher reads the ad and decides. Copy that says "clinical equipment for PT practices" is doing filtering work that no keyword setting can do, because it speaks to the human being rather than to the matching system.
That is a genuinely different way to think about a headline. A price point, a minimum order, an industry, a license requirement, a service area, a "for businesses" or "for clinics": each one is a small gate. Some people read it and do not click, and you keep that money.
The consequence is worth stating plainly, because it contradicts most advice: a higher click-through rate is not automatically better. An ad that attracts everybody has a wonderful CTR and a terrible cost per acquisition. What you want is a high click-through rate among people who could actually buy, which sometimes means writing copy that deliberately loses the rest.
The Two Fields Almost Every Account Leaves Empty
Under the headlines sit two small boxes called the display path, fifteen characters each, appended to your visible URL.
They cost nothing, they take ten seconds, and in most accounts we audit they are blank. A blank path shows the bare domain. A filled one lets the visible URL read as your-site.com/commercial/repairs rather than just your-site.com.
Two things happen when you fill them. The ad reads as more specifically relevant to what was searched, which is the same lever as ad relevance in your quality score. And the searcher gets a small signal about where the click leads, which is reassurance at exactly the moment they are deciding.
They do not have to match your real URL structure. They are display text. Which means the right words go there, not whatever your CMS happens to call the page.
Pinning Is a Trade, Not a Setting
Pinning fixes a headline to a specific position. Position one, two or three, guaranteed to appear there.
There are real reasons to use it. A regulatory disclosure, a legally required phrase, a brand name that must lead. Those are legitimate and you should pin them.
The problem is that pinning is not free. Every pin removes combinations from the pool. Pin one headline to position one and no other headline can ever be tested there. Pin several and you have quietly rebuilt the old fixed-format ad inside a format designed to be flexible, while keeping none of the advantages of either.
Google's own guidance is to pin two or three different headlines to a position rather than one, which preserves the testing while still constraining what can appear. That is the version to use when you have a genuine constraint.
The audit finding we see most often is an account where somebody pinned things for control, not for a requirement, and then concluded the format does not work.
Ad Strength Is Feedback, Not a Grade
Google shows an Ad Strength rating: Poor, Average, Good, Excellent. It is useful and it is widely misunderstood.
Ad Strength measures things like quantity, variety and keyword relevance in your assets. It does not measure whether your copy is persuasive, and it has no idea whether your offer is any good. It is closer to a completeness check than a quality judgment.
Google's own data says advertisers moving from Poor to Excellent see meaningfully more clicks and conversions on average, which is a real effect and worth having. But the mechanism matters: the gain comes from having more genuinely varied assets to test, not from the rating itself. Adding four near-duplicate headlines to push the meter up produces the label without the effect.
So treat it as a checklist telling you the ad is unfinished, and then finish it properly rather than to the meter.
What Actually Goes in the Fifteen
The hardest part is not the character limit. It is coming up with fifteen genuinely different things worth saying.
The way through it is to write against categories rather than staring at fifteen empty boxes. Something like:
- What it is, matching the search closely. Two or three variations, because this is the relevance workhorse.
- Who it is for. The filter. Industry, business type, license, scale.
- Proof. Years, volume, credential, guarantee, a number you can defend.
- The differentiator. What is true of you and not of the ad above you.
- The constraint. Price band, minimum, service area. The gate that saves you money.
- The action. What happens next, and how quickly.
Six categories, two or three variations each, gets you to fifteen without repeating yourself. And because each category says something different, Google's combinations stay coherent no matter which three get drawn.
One structural note. An ad can only be this specific if the ad group it lives in is about one thing. An ad group holding forty unrelated keywords forces every headline to be vague enough to cover all of them, which is a structure problem rather than a copywriting one. Fix that first or none of this is possible.
The Twenty Minute Version
- Count your headlines. Fewer than ten and the ad is unfinished. Fifteen is the target, and they must be different from each other.
- Count your descriptions. Four is the maximum. Two is what most accounts run.
- Fill the display path. Both fields, fifteen characters each, using the words the searcher used rather than your folder names.
- Audit your pins. Anything pinned for preference rather than requirement should be unpinned. Where you must pin, pin two or three to the position.
- Write one headline that repels. A price, a minimum, an industry, a service area. Something that makes the wrong person not click.
- Read Ad Strength as a completeness check, not a score to chase.
- Check the ad group first. If it covers several unrelated ideas, no single ad can be relevant to any of them.
- Run more than one ad per ad group so there is something to compare against.
One last habit worth adopting. Run at least two responsive search ads in an ad group rather than one. A single ad has nothing to be compared against, so you learn only whether it works, never whether something else would work better. Two ads with genuinely different angles, left running long enough to produce real data, is how you find out which argument your market responds to.
Ad copy is one of the few things in a Google Ads account that costs nothing but attention and affects both what you pay per click and who clicks at all. It is also the part most likely to have been written once, at launch, by whoever was building the campaign that afternoon.
We rewrite ad copy on almost every account we take over, and the first pass is usually not clever positioning. It is filling in the boxes that were left empty. Hand it to us and we will write the fifteen. Or open one ad group tonight and count what is actually in your ad. If the answer is five headlines and a blank display path, you already know what to do first.
