--- name: select-a-bidding-strategy description: >- Picks the right initial bid strategy for a campaign by crossing three inputs: the optimization objective (visibility, traffic, conversions, or value), the account's measured conversion volume for that campaign type, and campaign-type constraints, then documents the migration path from any data-gathering starting point to the destination strategy. Use it at campaign launch, when adding a new campaign type, or when a stakeholder requests a strategy reset. It measures data readiness with real GAQL pulls rather than assumptions. Not for the target math (calculate-bid-targets), the hands-on setup of the chosen strategy (set-up-conversion-based-bidding / set-up-value-based-bidding / set-up-portfolio-bid-strategies), moving off Manual CPC (migrate-from-manual-to-smart-bidding), or ongoing health checks (monitor-and-maintain-bid-strategy-health). --- # Select an Initial Bidding Strategy ## Purpose Most bid-strategy failures are selection failures: a conversion-based strategy applied to a campaign with eight conversions a month, or a value strategy with no value data. This skill forces the selection through evidence, what is the objective, how much conversion signal actually exists for this campaign type, what does the campaign type permit, and leaves behind a documented choice plus the migration trigger that upgrades it later. ## When to run - Launching a new campaign and the bidding choice is open. - Adding a campaign type the account has not run before. - Rebuilding after a structural overhaul. - A stakeholder wants the strategy re-derived from current objectives. ## When NOT to run - The strategy is chosen and needs a number -> calculate-bid-targets. - The strategy is chosen and needs configuring -> set-up-conversion-based-bidding (volume goals) or set-up-value-based-bidding (revenue/profit goals). - A live Manual CPC campaign should move to automation -> migrate-from-manual-to-smart-bidding. - Several thin campaigns share one goal -> set-up-portfolio-bid-strategies may beat any single-campaign choice. - An existing strategy is underperforming -> monitor-and-maintain-bid-strategy-health first; re-selection is rarely the fix. ## Prerequisites - Campaign goals and KPIs documented (growth + efficiency pair). - Campaign type decided (Search, Standard Shopping, PMax, Video, Demand Gen). - Conversion tracking configured and verified for conversion/value objectives. - Budget for the campaign known. ## Procedure 1. **HUMAN STEP (outside VigilDog).** Confirm the business objective with the stakeholder and map it to one optimization objective: awareness -> visibility; traffic/audience building -> clicks; leads/purchases count -> conversions; revenue/profit -> value. Cross-check that the primary KPI agrees (impression share -> visibility, CTR/clicks -> traffic, CPA -> conversions, ROAS/value -> value). A KPI/objective mismatch means the goals need rework before bidding is discussed. 2. **Measure data readiness.** Run `gads_run_gaql_query` for the last 30 days of conversions on this campaign type (and `gads_get_account_performance` for account context). Classify: under 15/month -> insufficient for conversion bidding; 15-29 -> minimum viable, volatile; 30-49 -> workable for tCPA; 50+ -> full menu including tROAS. For value objectives also verify conversion values exist and vary (`metrics.conversions_value` segmented by conversion action). 3. **Assess familiarity (judgment step).** New keywords/audiences, new geography, or an untested offer each push toward a conservative data-gathering start; proven targeting with deep niche knowledge (known CPCs, known conversion rates) supports starting directly on a targeted strategy. Record the assessment. 4. **Select via the Decision rules.** Walk objective x data-state x campaign-type. Write down the chosen strategy AND the reasoning path, the reasoning is what future sessions audit. 5. **Document the migration plan.** If the choice is a data-gathering strategy (Manual CPC, Maximize Clicks, or an untargeted Maximize Conversions/Value), record: destination strategy, trigger condition (e.g. "30+ conversions/month for 2 consecutive months"), and who watches it. A data-gathering strategy without a written exit becomes permanent by accident. 6. **Configure (write).** Consult `gads_policy_guardrail` before the session's first write. Apply with `gads_set_campaign_bidding_strategy` (or `gads_create_portfolio_bidding_strategy` when pooling was chosen). Set initial targets per Decision rules. Preview (validate_only=true default), show strategy + target + reasoning, apply only after user approval. For full setup including learning-period management, hand off to the setup skill for the chosen family. 7. **HUMAN STEP (outside VigilDog).** Brief the stakeholder: chosen strategy and why, ~two conversion cycles of volatile learning, no mid-learning changes, evaluation date, and the migration plan if starting conservative. 8. **Set the migration watch.** Schedule a recurring `gads_run_gaql_query` check of the trigger metric (30-day conversions, or value variance for tROAS candidacy). When the trigger fires, route to the appropriate setup or migration skill. ## Decision rules - **Volume thresholds (per campaign type, last 30 days):** <15 conversions -> no conversion-based strategy; use Manual CPC or Maximize Clicks to gather data, or pool campaigns into a portfolio. 15-29 -> Maximize Conversions (no target) preferred; tCPA only with a conservative target and volatility expectations set. 30-49 -> tCPA viable. 50+ -> tCPA reliable; tROAS becomes viable (tROAS wants 50+ conversions with genuinely varied values). - **Objective mapping:** visibility -> impression-share-based bidding; traffic -> Maximize Clicks (with a CPC ceiling); conversions -> Maximize Conversions / tCPA; value -> Maximize Conversion Value / tROAS. - **Value-strategy gate:** no dynamic conversion values, or all values identical -> value bidding degenerates to volume bidding; choose tCPA instead. - **Campaign-type constraints:** PMax, Video, Demand Gen, and App campaigns take campaign-level strategies only (no portfolio). PMax runs Maximize Conversions/Value with optional targets. - **Initial targets:** tCPA at or slightly above the 30-day average CPA; tROAS at or slightly below the 30-day average ROAS; Manual CPC max CPC ~= affordable CPA x expected conversion rate (50 affordable CPA x 3% CVR -> 1.50). New accounts: derive breakeven from unit economics and start 10-20% conservative of it. Never launch at an aspirational target. - **Cold start + unfamiliar targeting:** both conditions together mandate the conservative data-gathering path regardless of ambition. - **Adjustment cadence after launch:** wait ~two conversion cycles (typically 14-28 days) before the first tightening; move 10-15% per step with a cycle between steps. ## Common failure modes - **Choosing by aspiration, not data.** tROAS on a campaign with 20 conversions and flat values fails predictably; the volume gate is not negotiable. - **Aggressive launch targets.** A day-one target far better than history forces the algorithm into a corner and burns the learning period; start at reality. - **Data-gathering forever.** Without a written trigger, Manual CPC "temporarily" runs for a year. Step 5 exists for this. - **Ignoring type constraints.** Planning a portfolio strategy around PMax or Video members fails at execution time; check constraints during selection, not setup. - **Skipping the stakeholder brief.** Unbriefed stakeholders panic during learning and demand mid-learning changes, the single most damaging intervention available. ## Related skills - Before: calculate-and-validate-unit-economics + calculate-bid-targets (targets the choice will need), set-up-google-ads-conversion-tracking (signal quality). - After: set-up-conversion-based-bidding, set-up-value-based-bidding, set-up-portfolio-bid-strategies (implementation of the choice), migrate-from-manual-to-smart-bidding (when the data-gathering exit trigger fires), monitor-and-maintain-bid-strategy-health (ongoing).