Skip to content

Instantly share code, notes, and snippets.

@arvindang
Created May 13, 2026 20:27
Show Gist options
  • Select an option

  • Save arvindang/9f1d19343f4261638e5e8beba66d8495 to your computer and use it in GitHub Desktop.

Select an option

Save arvindang/9f1d19343f4261638e5e8beba66d8495 to your computer and use it in GitHub Desktop.
SEO + Ads Launch Pathway — a phased skill & tooling template for AI-assisted founders (DataForSEO, Firecrawl, GA4, GSC). Adapted from a real rollout.

SEO + Ads Launch Pathway — A Phased Skill & Tooling Template

A reusable rollout plan for taking a B2C product from "no SEO foundation" to "paid ads with attribution." Designed for solo founders and small teams using AI-assisted development (Claude Code, Cursor, etc.).

Each phase maps to a skill (an AI agent task) and the third-party tools / APIs that skill actually needs to do useful work. Skills without their backing tools produce thin, generic output — pair them deliberately.

Skill packs referenced

Sequencing rationale

  1. Context first. Every downstream skill reads your product/audience context. Write it once and commit it to the repo so collaborators (human and AI) share the same source of truth.
  2. Research before generation. Copy, ads, and SEO content skills produce slop without real customer voice and competitor data feeding them.
  3. Pages before ads. Ad copy echoes landing pages, never the reverse. Lock copy, CRO, and (if relevant) pricing before any spend.
  4. Schema + architecture early. Structured data and URL structure are foundations both Google SEO and AI search (ChatGPT, Perplexity, AI Overviews) depend on. Most playbooks put these too late.

Required tooling (set up before you start)

Category Tool Why
Keyword + SERP data DataForSEO (MCP) Live SERPs, keyword volume, backlinks, local pack data, AI Overview detection
Web scraping Firecrawl (MCP) Competitor pages, Reddit/forum mining, review extraction, llms.txt-ready content
Search Console Google Search Console Index coverage, query data, Core Web Vitals field data
Analytics GA4 + PostHog (or Ahoy/Mixpanel) Attribution for ads, funnel analysis, event coverage
AI search testing ChatGPT, Perplexity, Google AI Mode Manual citation testing — no API substitute yet
Schema validation Schema.org validator, Rich Results Test Verify structured data before deploy
Local / Maps Google Business Profile, DataForSEO Maps endpoints Per-location signals, NAP consistency, geo-grid rank tracking

Phase 1 — Foundation

# Skill What it produces Tools / data it needs
1.1 /marketing-skills:product-marketing-context .agents/product-marketing-context.md — positioning, ICP, JTBD, voice. Commit to repo. Internal knowledge; existing pitch decks; founder interview
1.2 /marketing-skills:customer-research (Mode 1) Synthesis of existing assets — user interviews, support tickets, NPS, sales calls Transcripts, support tool exports, interview notes
1.3 /marketing-skills:customer-research (Mode 2) Digital watering hole research — Reddit threads, App Store reviews, forum quotes Firecrawl (Reddit/App Store scraping), SparkToro (optional)
1.4 /marketing-skills:competitor-profiling Per-competitor dossiers: positioning, messaging, weakness Firecrawl (competitor sites), WebFetch (press, reviews)
1.5 /marketing-skills:marketing-psychology (optional) Push/Pull/Habit/Anxiety mapped to copy levers Output of 1.2–1.4

Mode 2 of customer-research can be deferred to right before Phase 6 if you're time-boxed — it provides higher leverage on ad copy than on early SEO foundations.

Phase 2 — SEO foundation

# Skill What it produces Tools / data it needs
2.1 /keyword-research (gstack) Wedge terms, search volume, intent classification DataForSEO keywords endpoint, Google Autocomplete, Google Trends
2.2 /marketing-skills:content-strategy Topic clusters, hub-and-spoke map DataForSEO SERP analysis, Firecrawl (top-ranking pages)
2.3 /marketing-skills:site-architecture URL structure, nav, internal-link graph Sitemap of current site (Firecrawl crawl)
2.4 /seo-plan Orchestration doc — strategy before tactics Outputs of 2.1–2.3
2.5 /seo-schema (preferred) or /marketing-skills:schema-markup JSON-LD for Organization, Product/LocalBusiness/Place, FAQ, etc. Often the highest single SEO + GEO lever. Schema.org validator, Google Rich Results Test
2.6 /seo-technical Core Web Vitals, indexability, crawl-budget fixes DataForSEO on-page, GSC, CrUX, Lighthouse
2.7 /seo-sitemap Dynamic sitemap (especially for programmatic pages) App's URL generator, GSC submission
2.8 /marketing-skills:seo-audit Baseline diagnostic + benchmark DataForSEO, Firecrawl, GSC

Phase 3 — AI search / GEO / Maps

This is where most playbooks fall short. AI search (ChatGPT, Perplexity, Google AI Overviews) does not behave like classic Google.

# Skill What it produces Tools / data it needs
3.1 /seo-maps Maps intelligence, geo-grid rank tracking, GBP audit DataForSEO Maps + GBP endpoints, Google Business Profile
3.2 /seo-geo GEO/AEO for ChatGPT, Perplexity, AI Overviews — citable answer blocks, brand mention strategy Manual prompt testing across LLMs, Firecrawl for citation source analysis
3.3 /seo-local (if you have locations) Per-city local SEO, NAP consistency, citations DataForSEO local pack, GBP, citation aggregators
3.4 /marketing-skills:ai-seo llms.txt, passage-level citability Site crawl, llms.txt generator

/seo-geo and /marketing-skills:ai-seo overlap. Pick one or layer both — /seo-geo goes deeper, ai-seo is lighter and faster.

Phase 4 — Page quality & scale

# Skill What it produces Tools / data it needs
4.1 /marketing-skills:copywriting Homepage + top landing-page copy Phase 1 research output (verbatim quotes are gold)
4.2 /marketing-skills:page-cro Conversion optimization on landing pages PostHog / GA4 / Hotjar session data, heatmaps
4.3 /marketing-skills:competitor-alternatives "[Your product] vs [Competitor]" and "[Competitor] alternative" pages Phase 1.4 competitor profiles, Firecrawl for feature parity
4.4 /seo-programmatic (preferred) or /marketing-skills:programmatic-seo Templated pages at scale — often the #1 SEO opportunity for any product with a natural [modifier] + [entity] query pattern DataForSEO for keyword templates, data source for the entities, sitemap generator
4.5 /seo-content + /seo-cluster Content briefs at scale DataForSEO SERP, Firecrawl (competing content)
4.6 /marketing-skills:free-tool-strategy (optional) Engineering-as-marketing free tools Depends on the tool — but skip if your core product already functions this way

Phase 5 — Pre-ads infrastructure (REQUIRED before spend)

Running ads without this is lighting money on fire — you won't know what worked.

# Skill What it produces Tools / data it needs
5.1 /marketing-skills:analytics-tracking Verified event coverage for ad attribution GA4, PostHog, server-side analytics (Ahoy/PostHog), Meta Pixel, Google Ads tag
5.2 /marketing-skills:pricing-strategy (if you have a paid tier) Price points, packaging, paywall placement Phase 1 research, competitor pricing
5.3 /marketing-skills:signup-flow-cro Signup conversion lift — this is the ceiling on every ad's ROAS PostHog funnels, session replay
5.4 /marketing-skills:onboarding-cro Activation lift — define your magic-moment metric Event analytics, cohort retention
5.5 /marketing-skills:form-cro (skip for most B2C) Form completion lift Form analytics
5.6 Loop back: complete customer-research Mode 2 Ad-copy language at scale Firecrawl, SparkToro

Phase 6 — Ads

# Skill What it produces Tools / data it needs
6.1 /marketing-skills:launch-strategy Sequencing, waitlist, launch-day plan All prior outputs
6.2 /marketing-skills:ad-creative Ad variants — copy, hooks, visuals Phase 1 research, copy from Phase 4
6.3 /marketing-skills:paid-ads Channel selection, account structure, budgets Meta Ads Manager, Google Ads, attribution from 5.1
6.4 /marketing-skills:ab-test-setup Landing-page tests once traffic flows GrowthBook / PostHog experiments

Skip list (with reasons)

Most playbooks try to do everything. Most products shouldn't.

Skill Skip if…
aso-audit No native mobile app
email-sequence, cold-email, lead-magnets B2C consumer app — not email-led
sales-enablement, revops No sales team / not B2B
churn-prevention No paid tier yet
community-marketing, referral-program, co-marketing Pre-PMF — defer
paywall-upgrade-cro, popup-cro Traffic isn't meaningful yet

Minimum viable path (rush mode)

If you have to ship paid ads in 3–4 weeks, this is the shortest defensible path:

product-marketing-context
  → customer-research (Mode 1 only)
  → seo-schema           ← biggest single lever, do not skip
  → seo-programmatic     ← if your product has a natural template pattern
  → seo-geo              ← AI search is where attention is moving
  → copywriting
  → page-cro
  → analytics-tracking   ← required before spending a dollar
  → signup-flow-cro
  → paid-ads

Notes on tool setup

DataForSEO MCP — easiest install is the remote HTTP server:

claude mcp add --header "Authorization: Basic <YOUR_BASE64_TOKEN>" \
  --transport http dfs-mcp https://mcp.dataforseo.com/http

Find the Base64 token in API Access → Base64 Format in your DataForSEO dashboard. Leave the IP allowlist empty — it breaks the remote MCP. Pricing is pay-as-you-go; start with $1–$5 to feel out cost per call.

Firecrawl MCP — install from npm:

claude mcp add firecrawl --env FIRECRAWL_API_KEY=<KEY> -- npx -y firecrawl-mcp

Free tier covers most early-stage research. Upgrade once you start crawling at scale (programmatic SEO research).

Search Console + GA4 — set these up before Phase 2. Most skills assume access to query and event data.


Template adapted from a real Cerca rollout. Customize the "Findings" and "Skip list" sections to your own product before using.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment