Write and test Google Ads copy variants
copyskillsetup L2★2,106
nowork-studio/toprank ↗Causal-lift measurements
ad-copy-generation-12pp vs no-skill baselinewith-skill 84% · baseline 96%
Measured by running the task with and without this artifact, K=5, graded by deterministic checks — no LLM judging.
What it does
Write and test Google Ads RSA copy variants
Best for
Teams running Google Ads at scale who need to test copy variants systematically against baseline CTR/conversion data while respecting positioning and landing page content.
Inputs
- · Business context (services, positioning, offers)
- · Personas with search-term language
- · Current ad performance data (clicks, CTR, conversions per ad)
- · Landing page content
Outputs
- · 15 headlines (30 chars max each)
- · 4 descriptions (90 chars max each)
- · RSA pinning strategy (position 1-3 assignments)
Requires
- · Google Ads API (ad_group_ad, keyword_view, search_term_view, campaign insights)
Preconditions
- · Business context JSON file must exist and be current
- · Personas file must exist with search language examples
- · Current ads are live and have conversion data
Failure modes
- · Chases ad strength over conversion rate (inverted optimization)
- · Generic copy if business context missing
- · Overshoots character limits (rejected by Google)
- · Uses competitor names (trademark/policy violation)
Trust signals
- · RSA mechanics documented with exact limits (15 headlines @ 30 chars, 4 descriptions @ 90 chars)
- · Character counting enforcement mentioned as mandatory
- · Pinning strategy with explicit rule: pin service+location position 1, CTA position 3, leave position 2 unpinned