Full RSA asset sets — headlines and descriptions mapped to intent, inside character limits, with pinning guidance.
Write RSA assets for our "ai marketing agent" ad group pointing at /ai-marketing-agent.
Fifteen headlines across six angles and four descriptions, all matching the landing page promise. One pin recommended for the brand headline. The objection angle ("no autonomous spend changes") is untested in the account and worth watching.
A Google Ads copy template should produce a complete asset set, not three headlines and a shrug. This one reads the query intent and the landing page, then writes fifteen headlines and four descriptions that cover distinct angles — so Google has real variety to test instead of paraphrases.
A structure for generating a full responsive search ad asset set: headlines that each make a different argument, descriptions that expand rather than repeat, and a mapping from each asset to the query intent and landing page promise it supports.
Fill all fifteen headline slots, but make them genuinely distinct — offer, outcome, objection, proof, audience, and mechanism are separate angles. Fifteen rewordings of one benefit give the system nothing to learn from.
By anchoring each asset to the actual query group and matching the landing page language. Relevance is a three-way agreement between query, ad, and page, and the template checks all three rather than optimizing the ad in isolation.
Sparingly. Pin only what must always appear — a legal requirement or a brand name — because heavy pinning removes the combination testing that makes responsive search ads work. The template flags which assets justify a pin.
The agent reads the ad group’s keywords and search terms from Google Ads, fetches the landing page so ad and page language agree, and draws proof points, offers, and brand voice from workspace Knowledge. Assets export as a copy-ready table, or push into the account for review once you approve.
The agent pulls the keywords and reads the destination page.
Note any required legal phrasing, claims to avoid, or the promotion in market.
Headlines and descriptions arrive grouped by angle so you can see the coverage.
Export or apply, then compare against the incumbent ad after enough impressions.
Produces a full asset set with genuinely distinct angles
Checks agreement between query, ad, and landing page
Advises against over-pinning instead of encouraging it
Inherits your real offers and proof points, not invented claims
Specificity and agreement. The ad should name a concrete outcome, use the language of the query, and promise exactly what the landing page delivers. Vague benefit statements lose to specific ones almost every time.
All fifteen, provided they differ meaningfully. Google needs distinct assets to learn from — filling the slots with paraphrases is functionally the same as submitting three.
Only where a phrase must always appear, such as a legal disclaimer or brand name. Extensive pinning defeats the combination testing that makes the format work.
Yes — the same asset logic applies, and the catalog includes a Performance Max asset template with the additional image and video guidance that format needs.