The longest job on a new grant account is the first structure, and it is the one most worth not doing by hand. Give the agent the brief and it reads the account, then writes the whole thing as a proposal: campaigns, ad groups, keywords with match types, negatives at the level that belongs, and the ads.
The brief it needs
- The organisation, and what it does.
- Who you want to reach, and where.
- What the website asks a visitor to do.
- What counts as a conversion, and whether it is tracked yet.
That last line is not a formality. An account that cannot measure a conversion is the account that loses the grant — not the one that spends badly — so the agent is told to say so before anything else and to propose fixing it first.
What it builds to
The programme's rules, rather than correcting into them afterwards: no single-word or generic keywords, phrase and exact preferred over broad, the $2 ceiling kept unless the campaign bids smart, geography and language set rather than left at everywhere, and each ad group pointed at the page that answers its keywords instead of at the home page.
Nothing is applied until a member of the room applies it. On a grant account that order is the safeguard: a change that wastes money is a judgement call, and a change that breaks a policy is an account that stops serving.
Questions people ask
Does it read the account first?
Yes, and it is told to reuse what is there — conversion actions, naming, existing campaigns — rather than building a second set beside them.
Can we change it before applying?
That is what the proposal is for. Read the summary and the operations, change what you disagree with, apply what is left.
Will it tell us what it expects?
It is asked to say what click-through rate it expects and which ad group it expects to be weakest, because that is the one worth watching from the first week.