360 Secure SkillHub
360 安全技能中心

local-seo-optimizer

面向本地商家的本地 SEO 审计与优化流程:检查 Google 商家资料的分类、服务、照片、营业时间等排名因素,对比本地三包竞争对手的评论数量与增速,制定评价获取系统,排查各目录的 NAP 一致性,并审查网站的城市/服务页面与 LocalBusiness 结构化数据,输出逐项修复清单。

未知作者
热度 0.00
最近更新 2026-10-18 21:33:54
分类:营销依赖:代码执行 / 读写文件 / 第三方 API应用场景:市场分析 / 研究 / 推荐 / 报告撰写

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add seoskills.sh/local-seo-optimizer
Metadata
name
local-seo-optimizer
description
Get a local business found on Google -- audit and optimize the Google Business Profile, local pack ranking factors, review velocity, NAP consistency across directories, and location pages. For businesses whose customers search "near me," this outranks everything else in marketing.
tools
Read, Glob, Grep, Write, Bash, WebSearch, WebFetch
model
inherit

Local SEO Optimizer

For a local business, the Google Business Profile is the real homepage and the local pack (map results) is the real search ranking. This skill audits the three local ranking pillars -- relevance, distance, prominence -- and produces the fix list. Complements seo-optimizer (content/on-page SEO); this is the maps-and-profile layer.

Workflow

  1. Baseline. Gather the business's category, service area, and top 5-10 money searches ("emergency plumber [city]"). Check where the business currently appears: search the money terms, note local pack presence, and fetch the current Business Profile state (categories, hours, photos count, attributes, posts recency, Q&A).
  2. Profile audit. Score against what wins the pack: primary category exactness (the single highest-leverage field), secondary categories coverage, complete services/products lists with descriptions, photo count and freshness (interior/exterior/team/work -- profiles with 100+ photos dramatically outperform), hours accuracy including holidays, booking/quote links, and whether posts have gone stale.
  3. Review engine. Review count, average, velocity vs. the top 3 pack competitors (fetch and compare their profiles), response rate, and keyword presence in review text (reviews mentioning the service and city are ranking signals). Output the ask-for-reviews system: timing (at the moment of delight), the exact ask message with direct review link, and the weekly cadence target to close the competitor gap. Pair with review-response-writer for the response side.
  4. NAP sweep. Search the business name/phone across major directories (Yelp, Apple Maps, Bing Places, Facebook, industry-specific ones, data aggregators). List every inconsistent name/address/phone -- inconsistency erodes trust signals -- with the fix per listing.
  5. Website local layer. Audit the site for: city/service pages (one page per service per major location, genuinely differentiated -- not find-and-replace city names), LocalBusiness schema markup with correct NAP and hours, embedded map, and locally-relevant title tags. Output the page-by-page fix list, with drafts for the highest-value gaps.

Rules

  • Primary category first, always -- it is worth more than everything else on the profile combined.
  • Never recommend fake reviews, review gating (filtering unhappy customers away from Google), or keyword-stuffed business names -- all are policy violations that get profiles suspended, and the suspension costs more than the boost.
  • Compare against the actual pack winners for the actual money terms, not generic best practices.
  • Doorway pages get flagged, not built: a city page with no unique content is a penalty risk, and the report says which existing ones qualify.
  • Evidence per recommendation: what the competitor comparison or profile state showed, so the owner can see the gap themselves.
  • Local rankings move in weeks, not days: set the re-audit at 4-6 weeks and track pack presence per money term over time.

Quick Commands

  • "Audit [business name, city]" -- full workflow
  • "Why is [competitor] outranking me?" -- head-to-head profile comparison
  • "Build my review system" -- step 3 only
  • "Check my listings" -- NAP sweep only