Direct questions · No. 30 / 31
30. When should we buy a tool, hire an agency—or buy nothing?
Last updated
Buy only when you can name the unresolved problem, the person who will use the output and the decision it will change. “We are worried about AI” is a reason to investigate, not sufficient justification for a recurring contract.
Use this proposed decision framework:
| Your situation | Recommended next purchase or action |
|---|---|
| One alarming screenshot | Small repeated diagnostic; no retainer yet |
| No validated buying questions | Prompt and customer-research work |
| Clear access or indexing problem | Scoped technical implementation |
| Recurring inaccurate product claims | Accuracy register and corrective publishing |
| Reliable baseline but inadequate collection capacity | Monitoring software trial |
| Evidence-backed content gaps with internal capacity | Brief existing content and web teams |
| Evidence-backed gaps without delivery capacity | Scoped implementation partner |
| Existing agency claims improvement without comparable evidence | Independent validation |
| No recurring commercial gap and no implementation owner | No new supplier; revisit at a defined trigger |
In the first working week, appoint the owner, collect the questions that matter, inspect existing first-party reporting and preserve the answer that triggered concern. Then approve the smallest useful test.
The buying rule is straightforward: fund evidence, implementation and validation as distinct jobs. A citation is not a conversion, a score is not a diagnosis, and a proposal is not proof.