# Lead Quality Auditor > A Claude skill from hayesti54-eng/ai-media-buying-skills. - **Type:** Skill - **Install:** `agentstack add skill-hayesti54-eng-ai-media-buying-skills-lead-quality-auditor` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [hayesti54-eng](https://agentstack.voostack.com/s/hayesti54-eng) - **Installs:** 0 - **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [hayesti54-eng](https://github.com/hayesti54-eng) - **Source:** https://github.com/hayesti54-eng/ai-media-buying-skills/tree/main/google-lsa/diagnostics/lead-quality-auditor ## Install ```sh agentstack add skill-hayesti54-eng-ai-media-buying-skills-lead-quality-auditor ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # LSA Lead Quality Auditor **Platform:** Google LSA **Category:** diagnostics **Tier:** pro ## Purpose Systematically evaluate the quality of every lead charged through Google Local Services Ads to separate legitimate, dispute-eligible, and genuinely chargeable leads. LSA charges per lead regardless of outcome — this skill exists to surface the cost of low-quality volume, drive dispute submissions where eligible, and reveal structural issues in job type targeting and service area configuration that are generating noise rather than revenue. ## When to Use - After any week where booked lead rate drops below 40% - When average cost-per-lead exceeds your target cost-per-acquisition threshold - When a new service area or job type is added and lead quality is unknown - Monthly, as part of a standard lead charge review cycle - When a client reports receiving leads for services they do not offer - After a budget increase that drove unexpected lead volume spikes ## Inputs Required - Lead log export from LSA dashboard (date range, lead type, lead status, charge amount) - CRM disposition data mapped to LSA lead IDs or phone timestamps - Job type configuration currently active on the LSA profile - Service area boundaries (zip codes or city/radius currently set) - Average ticket value and close rate by job type - Any dispute submissions filed in the prior 30 days and their resolution status ## What This Skill Does Performs a structured quality score on each lead bucket — answered calls, missed calls, messages — against job type alignment, geographic validity, and conversion outcome. Flags leads that are dispute-eligible based on Google's current dispute criteria: wrong service, spam/robo calls, calls under 30 seconds with no meaningful exchange, geographic mismatch outside the declared service area, and duplicate contacts. Outputs a lead quality breakdown by category, a dispute shortlist, and a structural diagnosis of what in the profile configuration is generating low-quality volume. ## Analysis Workflow 1. Pull the full lead log for the target date range — include all lead types (call, message, booking). 2. Segment leads into four buckets: Booked, Not Booked - Valid, Not Booked - Disputed, Not Booked - Unclear. 3. For each "Not Booked" lead, cross-reference CRM call notes or recording timestamps to identify call duration and content. 4. Flag leads under 30 seconds with no booked outcome as dispute candidates — log call timestamp, duration, and lead ID. 5. Identify leads where the caller requested a service not listed in the active job types — these are job type misalignment leads and may be disputable. 6. Cross-reference caller zip code or area code (when available) against service area boundaries — flag geographic outliers. 7. Check for duplicate caller numbers within the billing period — multiple charges from the same number within 30 days may qualify for dispute. 8. Calculate booked lead rate: (Booked Leads / Total Charged Leads) × 100. 9. Calculate effective cost-per-booked-lead: Total Lead Spend / Booked Leads. 10. Score each active job type by its individual booked lead rate — surface the lowest performers. 11. Compile dispute submission shortlist with evidence requirements for each flagged lead. 12. Identify the top 2–3 structural causes of quality degradation (job type, geography, time-of-day). ## Output Requirements - Lead quality scorecard: total leads, booked count, booked lead rate %, dispute-eligible count - Cost analysis: total spend, cost-per-lead, cost-per-booked-lead, estimated recoverable spend via disputes - Dispute shortlist: lead ID, date, call duration, disqualification reason, recommended dispute category - Job type quality breakdown: booked rate per job type, recommended job types to pause or refine - Service area assessment: flagged zip codes or regions generating out-of-scope leads - Actionable recommendations: profile changes, job type removals, service area tightening, call handling fixes ## Platform-Specific Best Practices - Google LSA disputes must be filed within 60 days of the lead charge — never let a dispute backlog age past 45 days. - Do not dispute leads that lasted over 2 minutes even if no booking occurred — Google will deny these. - Job type alignment is the single highest-leverage fix for lead quality: removing irrelevant job types reduces junk volume immediately. - LSA does not allow keyword-level exclusions — quality control happens entirely at the job type and service area level. - Message leads have a different dispute threshold than call leads — evaluate them separately. - A booked lead rate below 30% is a structural profile problem, not a market problem. ## Guardrails - Do not dispute leads where the conversation was substantive but the customer chose a competitor — these are valid leads. - Do not remove job types without confirming the business actually cannot service that category. - Never inflate dispute claims — Google tracks dispute abuse and can reduce profile trust signals. - Geographic flags are indicators, not automatic dispute qualifications — verify before submitting. ## Example Requests 1. "Audit our LSA leads for March — we spent $1,800 and only booked 6 jobs. Show me what's disputable and what's a profile problem." 2. "We added water heater installation as a job type last month and lead quality tanked. Run a quality breakdown and tell me if we should remove it." 3. "Pull a dispute shortlist from last week's leads — I want every call under 30 seconds and every out-of-area contact flagged with evidence notes." ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [hayesti54-eng](https://github.com/hayesti54-eng) - **Source:** [hayesti54-eng/ai-media-buying-skills](https://github.com/hayesti54-eng/ai-media-buying-skills) - **License:** MIT Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-hayesti54-eng-ai-media-buying-skills-lead-quality-auditor - Seller: https://agentstack.voostack.com/s/hayesti54-eng - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.