A reviewable Google Ads structure for a pet-food category — built, not uploaded
A pet-food e-commerce retailer needed a defensible starting point for a Google Ads campaign on a core product category: a coherent ad-group and keyword structure plus on-brand responsive search ad copy. AOS produced both and stopped at the deliverable, deliberately leaving the live upload paused for a human to review.
01 Ground
For a pet-supplies retailer, paid search is where most of the high-intent demand actually surfaces. Someone searching for a specific food line is closer to a purchase than almost any other signal the business can buy. But standing up a campaign for a category from scratch is the kind of task that quietly absorbs an afternoon: clustering the right keywords into sensible ad groups, deciding match types, and then writing responsive search ad assets that are both on-brand and tuned to the query — fifteen headlines and four descriptions per ad, each one fighting for relevance and ad strength.
The reflex when this work is slow is to either skip the structure and dump everything into one broad ad group, or to lift a generic template and let the platform's automation sort it out. Both produce spend without a thesis. The first wastes budget on irrelevant queries the moment broad match drags them in; the second produces ad copy that could belong to any retailer in any vertical, which is exactly the copy that loses the auction on message rather than money.
What the retailer wanted was narrower and more honest: not a launched campaign, but a reviewable artefact. A starting point that a human could open, sanity-check against their own knowledge of the category, edit, and only then decide whether to push live. The hard part of automating this is not generating text — models do that easily — it is generating text that is grounded enough to be worth reviewing rather than rewriting, and knowing where to stop before crossing into an irreversible action.
- A category-level ad-group and keyword structure with on-brand RSA headline and description variants AOS skill run cab5a8ec deliverable 90% confident
- Run completed in 3.6 minutes at $0.56 AOS skill run cab5a8ec 95% confident
02 Decide
The structural decision was to treat campaign construction and campaign deployment as two separate acts, and to perform only the first. AOS built the ad groups, organised the keywords, and wrote the responsive search ad variants — then saved the whole thing to deliverables and stopped. No upload to the live account. That boundary is the point of the run, not a limitation of it.
The reasoning is that uploading to a live Google Ads account is an outward-facing, money-spending action, and the brand voice and category nuance in the copy are exactly the things a retailer is best placed to judge. By delivering a complete, openable starting structure instead of a launched campaign, AOS keeps the reversible work automated and the irreversible decision with the human. The retailer reviews real assets rather than a blank page, which is a faster edit than authoring, but nothing goes live without their explicit sign-off.
03 Deploy
AOS read the category and brand context, then assembled the campaign skeleton: ad groups scoped to coherent intent clusters, keywords sorted into those groups rather than piled into one, and a set of responsive search ad assets — headlines and descriptions written to the category and the brand's tone, not to a generic template. The output was assembled as a deliverable a human can open and work from.
Crucially, the run treated the live upload as out of scope by design. The finished structure was saved to the client's deliverables for review, with the upload step left paused. The whole pass cost $0.56 and took 3.6 minutes — cheap enough that producing a reviewable draft is never the bottleneck, and the only remaining work is the judgement call about whether to launch.