The Best Strategy Tool Might Be the One That Says: Here Are Your Options.
A client asks whether to increase spend in generic search. An AI assistant can produce a recommendation in seconds, complete with a tidy rationale and a confident list of next steps. It might even sound like a senior strategist. But has it seen the account’s previous tests? Does it know why the budget was cut last year, whether conversion measurement changed or which commercial constraints mattered? Without that context, a polished recommendation may be little more than a generic answer wearing the client’s name.
Good planning needs alternatives you can challenge, not instructions you’re expected to obey.
A more useful response might offer several ways forward: expand investment where measurement is credible; run a bounded test; or fix attribution problems before committing more money. Each option should explain what evidence supports it, what cuts against it and what would need to be true. Previous experience could help rule out an approach, but it shouldn’t turn an old conclusion into permanent doctrine. Different conditions can make the same strategy sensible at one moment and foolish at another. And sometimes the most valuable answer is simply that the organisation doesn’t yet know enough to choose.
That leaves the strategist in charge, which is where they belong. The purpose of connecting past decisions with present performance isn’t to manufacture automated wisdom. It’s to stop intelligent people repeatedly starting from a blank page, and to give them a clearer set of facts, uncertainties and possibilities to argue about. Agencies already sell judgement. Better access to their own evidence ought to make that judgement stronger, not replace it with something that merely sounds decisive.