LeadUp Keyword + Competitor Researcher

Purpose

Given a business, product, or page, return a structured keyword map and a short competitor analysis the LeadUp team can use to plan SEO content, service pages, and landing pages. The output is grouped by topic, scored by intent and priority, and honest about which numbers are verified vs estimated.

When to use

Use when the user wants keyword research or competitor analysis. Do not trigger when the user wants the full SEO audit (use leadup-seo-strategist), a written blog (use leadup-blog-content-writer), or a paid-ads keyword plan only (use leadup-digital-ads-planner).

Trigger phrases: "keyword finder", "competitor analysis", "find keywords", "seo keywords", "research competitors", "keyword plan", "what keywords should we target", "build a keyword map".

Inputs needed

• Business / product description in 1–2 lines.
• Audience and market (India tier-1/2/3, international, languages).
• 3–5 competitors (URLs if known; else infer from search).
• Existing pages to map keywords to (URLs or page list).
• Access to data sources: Google Search Console, Google Keyword Planner,

Ahrefs / Semrush, Ubersuggest, AnswerThePublic, Reddit, YouTube (mark which are available).

• Constraint: max keywords (default 30–50 for a small business, 100–200

for a SaaS).

Ask at most 2 clarifying questions if the business or market is unclear.

Required resources

• Official / authoritative: Google Keyword Planner, Google Search

Console (Performance + queries), Google Trends.

• Free: AnswerThePublic, AlsoAsked, Google "People also ask",

autocomplete suggestions, Reddit threads, YouTube auto-suggest.

• Paid (if user has access): Ahrefs, Semrush, Moz, Ubersuggest.
• Competitor sites: their HTML (titles, H1s, URL structure), their

blog index, their sitemap.xml.

• LeadUp memory: prior project keyword lists if relevant.

If exact volume, CPC, or difficulty is not available in a tool, estimate with a banded label (low/med/high) and mark the row estimated.

Internet research workflow

1. Extract seed terms from the business description (services, audiences,

problems, locations).

1. Expand seeds via: Google autocomplete, "People also ask", related

searches, and AnswerThePublic; record each suggestion.

1. Pull competitor URLs from a search for the top 3 seed terms; record the

top-ranking 3–5 organic results per seed.

1. For each competitor: scan sitemap.xml or /blog/ index, list URLs by

topic; record their title pattern, H1, and any obvious topic clusters.

1. If Search Console / Keyword Planner / Ahrefs is available, pull

volumes, difficulty, and CPC; record source per row.

1. Cluster keywords by intent (informational, commercial, transactional,

navigational) and by topic.

1. Identify content gaps: topics competitors rank for that the user does

not yet have a page for.

1. Note assumptions and any term where intent was ambiguous.

If a browser or SEO MCP is available, hand off the fetch step to leadup-mcp-tool-orchestrator.

Step-by-step workflow

1. Restate the brief (business, audience, market, max keywords).
2. Build the seed list from the business description.
3. Expand seeds using free sources, then paid if available.
4. Cluster by topic and intent.
5. Score each keyword for priority (high/med/low) using:

• intent fit to business goal,
• estimated volume band (low/med/high),
• difficulty band (low/med/high),
• effort to rank (page already exists vs needs new page).

1. Map each keyword to a page (existing URL or "new").
2. Competitor pass: for the top 5–10 keywords, list 3 competitor URLs

and one-line "why they rank" + content gap.

1. Output the keyword map (Markdown table or CSV) + the competitor

summary + assumptions block.

Full framework: references/keyword-research-framework.md and references/competitor-analysis-framework.md. CSV shape: assets/keyword-map.template.csv.

Required output format

One Markdown report with these sections, in this order:

1. Brief — business, audience, market, max keywords.
2. Seed keywords — 8–15 seeds the rest of the map expands from.
3. Keyword map — table with columns: keyword, cluster, intent

(info/comm/trans/nav), page type, target page, priority (high/med/low), volume band, difficulty band, data source, confidence (verified / estimated / requires verification). Provide also a CSV-ready version using assets/keyword-map.template.csv.

1. Clusters — list of topic clusters; for each, the 5–10 keywords

inside and the recommended page type (pillar / spoke / service / landing / blog).

1. Competitor analysis — for 3–5 competitors: domain, what they rank

for, content structure, one strength, one gap LeadUp can attack.

1. Content gaps — 5–10 topics the user does not yet cover that

competitors do.

1. Assumptions and data confidence — sources used, what is verified

vs estimated vs requires verification, what would unlock better data.

1. Hand-offs — pages to feed into leadup-seo-strategist,

leadup-blog-content-writer, or leadup-landing-page-cro-planner.

Keep tone simple. Defer public copy polish to leadup-human-content-editor.

Safety rules

• Never invent volumes, CPC, or difficulty. Use banded labels

(low/med/high) and mark estimated if a tool number is not available.

• Never invent competitor traffic ("they get 50K visits/month") — say

estimated band with reasoning.

• Do not copy competitor content; reference structure and patterns only.
• Respect robots.txt and paywalls; do not scrape gated content.
• No ranking guarantees.
• Flag any keyword that conflicts with brand voice, legal claims, or

regulated industries (health, finance, alcohol, gambling, real estate).

• Keep India-specific keywords (city + service) honest; do not pad lists

with cities the user does not operate in.

• Defer integrations to leadup-api-research-builder; defer browser /

search tooling to leadup-mcp-tool-orchestrator; defer public copy polish to leadup-human-content-editor.

Common mistakes

• Treating tool-free estimates as fact.
• Padding the list with high-volume but irrelevant keywords.
• Skipping intent classification — the map becomes a wish list.
• One huge cluster with 80 keywords instead of 6–10 focused clusters.
• Ignoring informational keywords because they "don't convert" — they feed

the pillar pages.

• Inventing competitors that don't actually rank for the seed terms.
• Forgetting to map each keyword to a target page (existing or new).
• Producing prose instead of a table the user can paste into a sheet.

Troubleshooting

• No paid tools, no GSC: deliver fully labelled estimated map +

list the GSC/Keyword-Planner queries that would upgrade rows to verified.

• Hindi / regional language market: cluster English and vernacular

keywords separately; note transliteration variants (e.g. "salon" vs "saloon").

• Local-only business: anchor clusters by city/locality, not

nationally.

• SaaS with global audience: split clusters by region (IN/US/EU)

because intent and competitors differ.

• Highly regulated industry: cap claims; flag every transactional

keyword for legal/compliance review.

• User wants instant answers: deliver a smaller, sharper map (20

keywords, fully labelled) instead of 200 unlabelled ones.

Test prompts

Should trigger (5)

1. "Find keywords and competitors for our dental clinic in Coimbatore."
2. "Build a keyword map for our salon booking SaaS."
3. "Research competitors for our jewellery client's website."
4. "What SEO keywords should we target for hostel-management software?"
5. "Keyword plan for a Flutter app for tutoring centres."

Should NOT trigger (3)

1. "Do a full SEO audit of our site." (→ leadup-seo-strategist)
2. "Write a blog post on dental veneers." (→ leadup-blog-content-writer)
3. "Plan a Google Ads campaign for booking searches." (→ leadup-digital-ads-planner)

Functional test cases (2)

1. Given "salon booking SaaS, target tier-2 Indian cities", with no paid

tool access, return a 30–50 row keyword map with intent, priority, volume band, difficulty band, data source per row, and every numerical cell labelled estimated / requires verification, plus 3 competitors with one-line gaps.

1. Given access to Google Search Console for the user's existing site,

return a map that explicitly marks rows pulled from GSC as verified with the GSC query name, and labels the rest as estimated; include a content-gap section listing 5+ topics competitors cover but the user does not.

Success criteria

• Keyword map has all required columns and every numeric cell is labelled

verified / estimated / requires verification.

• Keywords are clustered by topic AND classified by intent.
• Each keyword is mapped to an existing or new target page.
• Competitor analysis is honest about traffic estimates (banded, not

invented).

• Content gaps are specific (named topics, named page types).
• CSV-ready version of the map is provided via

assets/keyword-map.template.csv.

• Hand-offs to leadup-seo-strategist, leadup-blog-content-writer, or

leadup-landing-page-cro-planner are explicit.