The decision tree
Start with the work: if your foundation is broken, indexation, technical health, intent coverage, that is finite project work, buyable from a consultant without a retainer, the checklist is in how to fix SEO.
Then capacity: if someone internal has ten hours a week and owns outcomes, in-house plus tools usually beats an agency below enterprise scale, because your team compounds category knowledge an agency relearns quarterly.
Then the honest agency case: you need sustained execution volume across content, technical and outreach, nobody internal can own it, and the budget clears the real cost floor. That is where good agencies genuinely win, and it is a narrower lane than the industry suggests.
The five questions that expose weak agencies in one call
What outcomes do you report besides rankings? Show me a client result with the work that caused it. Who exactly does my work, and how senior are they? What happens in month one versus month six? And the 2026 filter: how do you measure and win AI answers, per assistant, with citations? Vague answers to two or more, keep looking. An agency still selling rankings-only in 2026 is selling a shrinking surface, the context is in is SEO worth it in 2026.
The option that did not exist last time you decided
Part of what agencies traditionally bundled, finding the gaps, prioritizing fixes, executing them, verifying results, now exists as platforms with published pricing, especially on the answer layer where categories are being won right now. Aethon runs that loop from $199 per month: moment mapping, executed fixes, attribution, no retainer, the model compared in best AI SEO agencies. Many teams land on a hybrid: platform for the answer layer, small agency or contractor for classic execution overflow. Whatever you choose, decide from a baseline instead of a pitch: the free AI visibility audit shows what needs doing before anyone quotes you for doing it.
