Search Account Build for Food, Pet and Media Brands

What the first five weeks look like: The account goes live in week five, and the first two weeks contain no keywords at all. They go to the conversion definition: which action counts, where it fires, what a single one is worth, and whether I can reproduce the daily number from raw event data without opening the interface. Automated bidding is only as good as that definition, so it is the one thing I refuse to inherit unexamined. The tracking work: Native Google Ads conversion actions rather than imported GA4 key events wherever the two disagree, because the import lags and models differently. Enhanced conversions with hashed email on lead accounts. Consent Mode v2 verified for the EEA, required since March 2024 for audience features and modelled conversions; I have opened French accounts where its absence explained most of an apparent year-on-year decline. A value on every conversion action rather than a flat 1, because a newsletter signup and a first subscription cannot share a bid. How the account is structured: One campaign per bidding decision. Not per theme, and never per keyword. Brand sits alone because it is a measurement question rather than a performance one; mixed in, it flatters every average it touches. Generic terms are grouped by what a conversion from them is worth, so each campaign holds queries that deserve the same target, whatever they happen to be about. Match types are chosen per group once search term data exists, not written into the plan beforehand. Negatives before ads: Food and drink accounts start with standing exclusions for recipe intent, nutrition lookups, product recall queries and the retailer names your distribution already covers. Pet brands start with symptom and veterinary emergency queries, adoption and shelter searches, and breed encyclopedia traffic. Publishers start with gratuit, pdf and archive-hunting variants. These lists decide whether broad match is a channel or a leak. What a translated build misses in France: Accents written correctly in copy that the keyword list will match either way. Vouvoiement held consistent across headlines Google will recombine in any order. Prices shown with TVA where the buyer expects to see it. Loi Evin limits if there is alcohol in the range, which govern what headlines are allowed to exist rather than being a compliance note appended afterwards. The part nobody enjoys: Google no longer shows every query. Real spend runs against terms held below the search terms report threshold, and no script recovers them. I build assuming I will never see them: containment at campaign level, n-gram review of what is visible, and budget ceilings that make an invisible mistake affordable rather than quarter-ending. Not included: Landing pages. Tag manager containers I did not build. Shopping feeds. Performance Max. Creative beyond responsive search ad text. Languages other than French and English. Who this is not for: Advertisers below roughly 8,000 EUR a month in media, where the build cannot repay itself inside a year. Anyone wanting percent-of-spend pricing, which pays me to spend your money. Teams who need something live next week, which is a repair of what exists and a different conversation. And any account where I am asked to optimise toward a conversion action I am not permitted to change. After launch: Two supervised weeks at reduced daily budgets while the bidding strategy collects its first conversions and stops guessing. Then a written handover covering the structure, the reasoning behind each split, the negative lists, and the three decisions I would revisit first if results move against you. Your team can run the account from that document, which is the point of writing it.