Content
75%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-organized, practitioner-grade reference: concrete numbers, executable tool commands, real validation checkpoints, and clean one-level reference delegation. It falls short of top marks on mild redundancy (duplicated question and platform sections), template-only copy frameworks without a worked example, and an implicit rather than explicitly staged workflow.
Suggestions
Merge 'Task-Specific Questions' into the 'Before Starting' context-gathering section — the two lists overlap heavily (platform, budget, conversion value, creative, landing page, tracking) and could be one consolidated checklist.
Drop or fold the 'Tool Integrations' table into 'Platform Selection Guide' — it repeats the same platforms and 'Best For' column with little new information.
Add one worked ad copy example inline (e.g., a filled-out PAS headline/body/CTA set) so the copy frameworks are demonstrable without opening the reference file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with practitioner specifics (70/30 budget splits, frequency caps, bid progression) and assumes Claude's competence, but contains minor duplication: 'Task-Specific Questions' largely repeats 'Before Starting,' 'Tool Integrations' re-tables platforms already covered, and the marketing-context load instruction appears twice. Efficient with trimmable instances — anchor 4, not 5. | 4 / 5 |
Actionability | Copy-paste script invocations ("python3 scripts/roas_calculator.py --spend 5000 --revenue 18000 --conversions 120 --clicks 2400 --margin 0.7"), concrete naming examples ("META_Conv_Lookalike-Customers_FreeTrial_{YYYY}Q1"), and numeric windows make guidance mostly executable; the ad copy frameworks are bracketed templates ("[Problem] → [Agitate the pain]") with no worked example inline, keeping it below anchor 5. | 4 / 5 |
Workflow Clarity | A coherent flow exists (context check → platform selection → structure → pre-launch checklist → weekly review) with real checkpoints ("Conversion tracking tested with real conversion," "investigate any category that dropped") and safe-change rules (20-30% budget steps, 3-5 day waits). The sequence is implied by section order rather than explicitly staged, so it fits anchor 4 rather than 5. | 4 / 5 |
Progressive Disclosure | Four references are clearly signaled at one level deep with full paths ("See [references/ad-copy-templates.md]", "See [references/audience-targeting.md]", etc.) and detail is appropriately offloaded; however, substantial content (retargeting tables, creative best practices) remains inline, and the referenced files could not be verified as present in the bundle. Good structure with minor gaps — anchor 4. | 4 / 5 |
Total | 16 / 20 Passed |