Paid Ads Bid Strategy Planner
Choose a paid-media bidding model, set a defensible starting target, and protect the campaign’s learning phase.
The skill treats exports as untrusted input, relies primarily on user-provided data, avoids requiring live credentials, and saves results only after user confirmation. It does not fully specify sensitive-data handling, rollback after memory writes, permission isolation, or source verification, so 9 points are deducted.
Inputs, decision gates, missing-data handling, target derivation, learning-phase controls, and failure branches are described coherently, and invented metrics are prohibited. However, static material does not prove executable key paths; some shared references are not supplied, and abnormal-input diagnostics and test coverage are limited, so the score is 9 under the static ceiling.
Triggers, inputs, outputs, and non-fit boundaries are clearly stated, with Chinese terminology and a Tier-1 manual-export path. However, geo-relevance is marked low, and evidence for platform-specific behavior, regional fit, and finer false-trigger boundaries is limited, so 4 points are deducted.
The skill has structured frontmatter, versioning, an Apache-2.0 license, Quick Start examples, a contract, save-path guidance, reference pointers, and next-skill routing. Maintenance ownership, update path, changelog, installation details, FAQs, and referenced-file contents are not sufficiently shown, and publisher identity is unverified, so 5 points are deducted.
The declared output includes a strategy, target derivation, portfolio map, and learning-phase plan, making the core planning result directly usable. The thresholds and formulas are explicitly estimates, and no real output or third-party execution evidence is provided; human review remains necessary, so the static maximum of 7 is applied.
The skill, matrix, repository tests, and CI materials provide some auditable primary evidence, and key numeric rules are labeled Estimated. There is no independent reproduction, third-party corroboration, or dedicated test evidence covering this skill's key paths, so the static maximum of 5 is applied.
- Do not treat rules such as 30 conversions, 15–30 conversions to exit learning, or 15% adjustment steps as guarantees from any ad platform; verify them against current platform documentation and account data.
- Before saving to memory, confirm that client, audience, and account-export sensitive data are excluded or appropriately protected, and review host access and backup controls.
- This skill plans initial bidding but does not execute platform changes; any live modification should receive separate confirmation and have a reversible plan.
- The publisher is not verified by the FollowSkills enterprise registry, so maintenance continuity and provenance should be checked manually.
What does this skill do, and when should you use it?
This skill plans bidding for new or restructured paid campaigns using tCPA, tROAS, max-conversions, or manual CPC. It derives a starting target from the user’s CPA, ROAS, and conversion history while labeling each metric as Measured, User-provided, or Estimated. It groups campaigns into portfolios that share a goal and target, then defines learning-phase entry, exit, and change restrictions. When usable conversion history is missing, it requests an export or recommends starting with max-conversions without a target.
Reads the campaign goal, CPA/ROAS history, conversion volume, current bid strategy, campaign set, budgets, and structural constraints; applies the built-in strategy-selection matrix; derives an initial tCPA or tROAS from trailing performance and labels the inputs; produces a bid-portfolio grouping map; estimates the conversions needed to exit learning, sets a do-not-touch window, and flags target, budget, or structural changes that could reset learning; emits a reusable handoff summary and, after user confirmation, can save the result under memory/ad/bid-strategy-planner.
- A paid-media operator launching a direct-response campaign needs to choose tCPA or another strategy from recent CPA and monthly conversion volume.
- A team restructuring campaigns from manual CPC to automated bidding needs a starting target that will not unnecessarily restrict delivery.
- An ads manager with several search campaigns needs to group campaigns into portfolios with shared goals and targets.
- A team with thin conversion data needs an entry and exit plan for a max-conversions learning period.
- A team preparing to scale needs to identify changes large enough to reset learning and route the risk to the ad-account auditor.
What are this skill's strengths and limitations?
- Covers tCPA, tROAS, max-conversions, and manual CPC selection.
- Requires the starting-target derivation and source label for each input metric.
- Includes portfolio grouping and learning-phase protection planning for new or restructured campaigns.
- Runs at Tier 1 with user-provided data and does not require live ad-platform access.
- Clearly separates bidding plans from budget allocation, live pacing, and account auditing.
- Without conversion history, it cannot responsibly set a tCPA or tROAS target and must request data or start without a target.
- The exact thresholds and derivation details depend on referenced repository materials whose contents were not provided here.
- It plans the initial bidding setup but does not allocate budgets, adjust live pacing, or score the account.
- Saving results requires host access to the memory directory; the supplied material does not describe persistence on other hosts.
- The supplied skill text provides no standalone test-suite or platform-specific validation evidence.
How do you install this skill?
This skill is part of the Aaron Marketing Skills collection. In Claude Code, run: /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then: /plugin install aaron-marketing@aaron. On compatible Agent Skills hosts, run: npx skills add aaron-he-zhu/aaron-marketing-skills; alternatively run: git clone https://github.com/aaron-he-zhu/aaron-marketing-skills. Use the repository folder ad/orchestrate/bid-strategy-planner/ as the skill directory. The supplied material does not document host-specific steps for manually enabling only this skill.
How do you use this skill?
Provide a campaign name, goal, recent CPA or ROAS, and conversion volume. Example: “Pick a bid strategy for my search campaign: direct-response goal, past 30 days $42 CPA and 90 conversions per month.” Another example is: “Set a starting tROAS target for this campaign—history is 3.8x ROAS and the goal is 4.5x.” If no conversion history exists, provide a last-30-day CPA/ROAS and volume export, or start with max-conversions without a target. The skill does not allocate budgets or monitor live pacing.
How does this skill compare with similar options?
Compared with budget-optimizer, this skill selects the bidding strategy and learning-phase targets rather than allocating spend across campaigns. Compared with budget-pacing-monitor, it plans pre-launch learning-phase entry rather than monitoring live delivery. Compared with ad-account-auditor, it produces the bidding plan rather than running the ROAS account gate or launch go/no-go audit.