Responsive Search Ads: Assets, Pinning and Testing
You no longer write an ad. You supply parts and Google assembles them — which changes what "good copy" means, and makes one popular tactic quietly counterproductive.
The shift from writing ads to supplying assets caught a lot of accounts off guard.
The instinct is to write one perfect ad and break it into pieces. That produces fifteen headlines that all say roughly the same thing, and a system with nothing useful to test.
What actually works is closer to writing fifteen small, distinct arguments and letting the auction decide which lands.
How the assembly works
You provide headlines and descriptions. At auction time Google selects a subset, arranges them, and shows that combination. Different searchers see different assemblies of the same ad.
Two consequences follow, and both change how you should write.
Every headline has to stand alone. You cannot rely on headline two following headline one, because it might not. A headline that only makes sense in sequence will eventually appear somewhere it does not.
Variety is the input the system needs. If your assets are near-duplicates, every combination is effectively the same ad. You have given it fifteen ways to say one thing rather than material to test.
Write assets that are actually different
The practical test: could two of your headlines appear together without sounding repetitive? If not, they are the same asset twice.
A spread that gives the system something to work with:
| Angle | What it does | Example shape |
|---|---|---|
| The keyword | Matches the query directly. Google bolds matching terms. | "[Service] in [City]" |
| The outcome | What the buyer actually wants | "More qualified enquiries" |
| The differentiator | Why you rather than the result above | "Transparent monthly pricing" |
| The proof | Reason to believe | "Google & Meta Partner" |
| The objection | Removes the reason not to click | "No lock-in contract" |
| The action | What happens next | "Get a free audit" |
| The qualifier | Filters out the wrong clicks before you pay for them | "For brands spending ₹1L+/mo" |
That last row earns its place. A click you do not pay for is worth as much as a conversion, and stating a minimum or a price band in an asset is the cheapest qualification available. The same principle drives negative keywords on the query side.
Pinning: use it sparingly, and know what it costs
Pinning locks an asset to a specific position. It is the most misused control in the format.
Every pin removes combinations the system could have tested. Pin heavily and you have rebuilt a static ad inside a responsive one, while keeping none of the advantages.
- Pin what is legally or contractually required — a disclaimer, a regulated claim, a mandated brand mention. These are real reasons.
- Pin at most one or two positions — and if you pin a position, supply several assets to that position so it can still test within it.
- Do not pin for tidiness — "it reads better this way" is a preference, and it costs you the testing you are paying for.
- Do not pin to force a keyword in — if the keyword needs to appear every time, the ad group is probably too broad. Fix the structure instead.
- Review pins when performance stalls — an over-pinned ad is a common and invisible cause of a campaign that will not improve.
Ad Strength is not a performance metric
This is the most useful thing to understand about the format, and the most commonly misread.
Ad Strength rates asset variety and relevance. It does not measure results. It is telling you how much material you have supplied, not how well that material converts.
An "Excellent" ad can convert badly. An "Average" one can carry an account. Chasing the rating by adding weak assets to satisfy a checklist makes the ad worse while making the label better.
How many ads, and how to test
Two or three responsive search ads per ad group is the working range.
One gives you nothing to compare. Many more splits impressions so thin that no ad accumulates enough data to judge — the same dilution problem that too many ad sets creates on Meta, for the same reason.
To test properly:
- Change one thing at a time. Two ads differing in six ways teach you nothing about which difference mattered.
- Test angles, not wordings. Outcome-led against proof-led is a useful test. Two synonyms is not.
- Wait for volume, not days. A low-traffic ad group can take weeks to produce a readable result.
- Judge on conversions. Click-through rate tells you which ad was more appealing, not which found better customers.
- Check the asset report. It flags which individual assets are performing Low. Replace those rather than rebuilding the whole ad.
What ties it together
Responsive search ads sit inside a structure that has to be right first. Tight ad groups so one set of assets is genuinely relevant to every keyword, disciplined negatives so you are not writing ads for queries you never wanted, and a bid strategy the account has the data to support.
All three are covered in Google Search Ads and bidding strategies. And for the craft of writing the assets themselves — across Google and Meta both — see ad copy that earns the click.
Frequently asked questions
How many headlines should a responsive search ad have?
Use the full allowance of headlines and descriptions if you can write genuinely different ones. More assets give the system more combinations to test. But four strong, distinct headlines beat fifteen that are minor rewordings of each other, because near-duplicates give the system nothing to learn from.
Should I pin headlines in responsive search ads?
Pin only what must appear, and pin as little as possible. Legal disclaimers, a required brand mention or a regulated claim are good reasons. Pinning for tidiness is not — every pin removes combinations the system could have tested, and heavy pinning turns a responsive ad back into a static one.
Does Ad Strength affect my rankings?
No. Ad Strength measures asset variety and relevance, not performance. It is feedback on how much material you have given the system to work with. An Excellent ad can convert badly and an Average one can carry an account, so treat it as a checklist rather than a score to optimise.
How many responsive search ads per ad group?
Two or three is a sensible working range. One leaves you with nothing to compare against. Many more splits your impressions so thinly that no single ad accumulates enough data to judge. Two well-differentiated ads plus a challenger is usually the right shape.
How long before I can judge a responsive search ad?
Wait for meaningful volume rather than a fixed number of days. Judging on a few hundred impressions produces confident wrong conclusions, and low-traffic ad groups can take weeks. Change one thing at a time, and give the system room to redistribute delivery before you act.
We build search accounts that test properly
Tight ad groups, genuinely varied assets, minimal pinning, and testing that changes one thing at a time.