Serpstat Content Strategy

Content Strategy Architect

Full content strategy in 2 minutes. This skill takes a domain, collects semantics via phrase matching, related keywords, and search questions, clusters by intent, finds missed competitor topics, and delivers a ready-to-use content plan for 3-6 months with priorities, traffic estimates, and format recommendations (article, FAQ, category, landing).

Prerequisites

The Serpstat MCP connector must be enabled in the chat. If it's not connected, use suggest_connectors to prompt the user to connect it before proceeding.

Pre-flight Check

Before executing any API calls, verify the account has sufficient credits.

1. Call get_credits_stats
2. Check data.left_lines:
   • If left_lines < 600 → stop immediately, do not proceed
   • Inform the user: "Insufficient API credits to run this analysis. Current balance: {left_lines}. Required: ~600. Please top up at serpstat.com/users/profile"
3. If left_lines >= 600 → proceed with execution

Never skip this check.

Workflow Overview

The analysis has 5 phases. Execute them in order.

Phase 1: Collect Inputs

Step 1a — Get the domain.

If the user has not provided a domain, ask for it. Extract the bare domain (no protocol, no trailing slash).

Step 1b — Auto-detect region and competitors, then confirm.

Even if the user did NOT explicitly mention a region or competitors, do NOT ask them to choose from scratch. Instead, auto-detect both and present them for confirmation. This reduces friction — the user just says "yes" or corrects.

1. Call get_domain_regions_count to find the top region:

Method: get_domain_regions_count
Params: { domain: "<domain>", size: 5 }

Pick the region with the highest keyword count as the recommended one.

2. Call get_domain_competitors using that top region:

Method: get_domain_competitors
Params: { domain: "<domain>", se: "<top_region_se>", size: 6 }

Pick the top competitor (skip the domain itself in results).

3. Present the auto-detected settings for confirmation via ask_user_input_v0. Show a summary of what was detected and ask the user to confirm or adjust:

ask_user_input_v0({
  questions: [
    {
      question: "I detected these settings for your analysis:\n\n📍 Region: <detected region, e.g. 🇺🇸 USA (g_us)>\n🏆 Competitor: <detected competitor domain>\n\nShall we proceed?",
      type: "single_select",
      options: [
        "Yes, proceed",
        "Change region",
        "Change competitor",
        "Skip competitor analysis",
        "Change both"
      ]
    }
  ]
})

4. CRITICAL: Do NOT proceed to Phase 2 until the user confirms. Wait for the user's response. If they want changes, ask for the new values and re-confirm. Only after explicit approval, move on.

If the user already provided domain + region + competitor explicitly in their opening message, skip auto-detection — use what they gave, confirm briefly, and proceed to Phase 2.

If the user provided only region OR only competitor, auto-detect the missing one and confirm both together.

Phase 2: Pull Data from Serpstat

Execute these API calls in sequence. All calls use the Serpstat MCP tools.

2a. Current keyword semantics

Pull the domain's existing keyword rankings to understand what content already exists:

Method: get_domain_keywords
Params: {
  domain: "<domain>",
  se: "<se>",
  sort: { "region_queries_count": "desc" },
  size: 100,
  withIntents: true,
  include_all_fields: true
}

This returns: keyword, position, search volume, traffic, CPC, difficulty, URL, and intent data. These are the domain's current content topics.

2b. Keyword expansion — related keywords

For the top 3-5 highest-volume keywords from step 2a, expand semantics:

Method: get_keywords
Params: {
  keyword: "<top_keyword>",
  se: "<se>",
  sort: { "region_queries_count": "desc" },
  size: 50,
  withIntents: true,
  include_all_fields: true,
  filters: { region_queries_count_from: 100 }
}

Run this for each of the top 3-5 seed keywords. This finds keywords that share ranking domains — related by competition overlap.

2c. Semantic expansion — related by meaning

For the same top 3-5 keywords, also pull semantically related keywords:

Method: get_related_keywords
Params: {
  keyword: "<top_keyword>",
  se: "<se>",
  sort: { "region_queries_count": "desc" },
  size: 50,
  withIntents: true,
  include_all_fields: true,
  filters: { region_queries_count_from: 100 }
}

These are related by meaning (not just ranking overlap) — better for topic clustering and discovering adjacent content opportunities.

2d. Long-tail expansion — autocomplete suggestions

For the same seed keywords, pull autocomplete-style suggestions:

Method: get_keyword_suggestions
Params: {
  keyword: "<top_keyword>",
  se: "<se>",
  size: 50
}

These capture question-style and long-tail queries — perfect for FAQ sections and "People Also Ask" opportunities.

2e. Competitor missed keywords (if competitor selected)

If the user selected a competitor in Phase 1, pull keywords the competitor ranks for that the user's domain does NOT:

Method: get_domain_uniq_keywords
Params: {
  se: "<se>",
  domains: ["<competitor>"],
  minusDomain: "<userDomain>",
  size: 100,
  include_all_fields: true,
  filters: { region_queries_count_from: 100 }
}

These are "missed topics" — content the competitor has that the user lacks. This is the highest-value data for the content plan.

Phase 3: Cluster, Classify, and Prioritize

This phase is performed entirely by Claude — no additional API calls needed.

3a. Deduplicate

Merge all keywords from steps 2a-2e. Remove exact duplicates. For near- duplicates (e.g., "best running shoes" vs "best running shoe"), keep the higher-volume variant.

3b. Cluster into Topics

Group keywords into content topics (clusters). A topic is a group of keywords that can be targeted by a single piece of content. Rules:

• Each topic gets a topic name (a short descriptive label, e.g., "Best Running Shoes 2025", "How to Start a Blog", "CRM Comparison")
• Group by semantic similarity and shared intent
• Aim for 30-100 topics total
• Each topic should have 3-15 keywords
• One keyword per topic is the primary keyword (highest volume)
• Remaining keywords are supporting keywords

3c. Classify Intent

For each topic, assign a search intent based on the keywords and any intent data from Serpstat:

If Serpstat returned withIntents data, use it. Otherwise infer from keyword text.

3d. Assign Content Format

Based on intent and keyword patterns, assign a recommended content format:

3e. Score and Prioritize

For each topic, calculate a priority score (0-100):

priority = (volume_score × 0.4) + (difficulty_score × 0.3) + (gap_score × 0.3)

Where:

• volume_score: normalized total monthly volume of all keywords in topic (0-100, where 100 = highest volume topic)
• difficulty_score: inverse of average keyword difficulty (100 - avg_difficulty), so easier topics score higher
• gap_score: 100 if topic comes from competitor gap data (step 2e), 50 if it's a new topic not in current rankings (step 2a), 0 if the domain already ranks well for it

Classify each topic:

• Quick Win (priority ≥ 70, difficulty < 40): easy to rank, high potential
• Strategic (priority ≥ 50, difficulty 40-70): medium effort, good payoff
• Long-term (priority < 50 OR difficulty > 70): requires sustained effort
• Existing — Optimize (already ranking but position > 10): improve existing

3f. Build Content Calendar

Distribute topics across a 3-6 month timeline:

• Month 1-2: Quick Wins + Existing Optimizations
• Month 2-4: Strategic topics
• Month 4-6: Long-term topics

Assign 4-8 topics per month (realistic publishing cadence).

Phase 4: Build the Dashboard

Create a standalone .html file and save it to /mnt/user-data/outputs/. The file must be fully self-contained — all CSS, JS, and data inline.

Read references/serpstat-brand.md for the complete brand style reference including colors, fonts, widget catalog, and spacing rules.

Dashboard Structure

Hero Section:

• Pill: "CONTENT STRATEGY · Serpstat"
• Title: "{domain} — Content Strategy"
• Subtitle: "Powered by Serpstat keyword data · {region} · {month} {year}"
• Report description (1-3 sentences below the subtitle, in normal body text style, explaining what this report contains and how to use it). Example: "This report analyzes {domain}'s current keyword footprint, expands it with related and long-tail queries, identifies content gaps vs {competitor}, and organizes everything into a prioritized content plan for the next 3-6 months. Use the tabs below to explore topics by priority, format, and timeline." Adapt the text to the actual analysis performed (e.g., omit competitor mention if none was used).
• Stats row: Total Topics, Total Keywords, Total Monthly Volume, Quick Wins count, Missed Competitor Topics count

Tab 1 — Overview

• Metric cards row: Topics by priority type (Quick Win, Strategic, Long-term, Optimize)
• Bar distribution: Topics by content format (Article, FAQ, Landing, etc.)
• Bar distribution: Topics by intent (Informational, Commercial, Transactional, Navigational)
• Coverage grid: Showing which topic clusters the domain already covers vs gaps

Tab 2 — Content Plan

• The core deliverable: a data table with ALL topics
• Columns: Priority (badge), Topic Name, Primary Keyword, Volume, KD (difficulty), Intent (badge), Format, Timeline (Month), Source (Current / Competitor Gap / Expansion)
• Sortable by any column
• Filterable by intent, format, priority type
• Each row expandable to show all supporting keywords with their individual volume and difficulty

Tab 3 — Quick Wins

• Action cards (Widget 6 from style guide) for the top 15 Quick Win topics
• Each card shows:
  • Priority badge (HIGH/MEDIUM) + type badge (QUICK WIN)
  • Topic title
  • Metrics: primary keyword volume, avg KD, keyword count
  • Expandable detail: all keywords in cluster, suggested format, competitor reference URL (if from gap data)
• Cards sorted by priority score descending

Tab 4 — Competitor Gaps (only if competitor was analyzed)

• Topics that came from competitor gap analysis (step 2e)
• Data table: Topic, Primary Keyword, Volume, Competitor Position, User Status (Missing / Weak), Suggested Format
• If no competitor was selected, show a helpful message: "No competitor was selected for this analysis. Re-run with a competitor to discover missed content opportunities."

Tab 5 — FAQ & Questions

• All question-style keywords extracted from step 2d
• Grouped by topic cluster
• These make excellent FAQ sections, "People Also Ask" targets, and featured snippet opportunities
• Data table: Question Keyword, Volume, Parent Topic, Difficulty

UX Requirements

• Every section needs a title + subtitle explaining what it shows
• All numbers in Roboto Mono
• All badges pill-shaped (border-radius: 20px)
• 48px gap between sections
• Sortable tables with click handlers
• Action cards with colored left border matching priority
• Dark footer: "Data: Serpstat API · get_domain_keywords, get_keywords, get_related_keywords, get_keyword_suggestions, get_domain_uniq_keywords · DB: {se}" + Serpstat logo SVG

File Naming

content-strategy_{domain}_{se}_{YYYY-MM-DD}.html

Phase 5: Present and Summarize

After creating the dashboard, use present_files to share it with the user.

Provide a brief text summary:

1. Total topics found and how they break down by priority
2. Top 3 Quick Win topics (name, volume, difficulty)
3. Biggest competitor gap (if applicable)
4. Recommended first month focus
5. One strategic insight about the domain's content landscape

Keep it concise — the dashboard is the deliverable.

Credit Budget

Estimated API credit usage per run:

To stay under 600 credits when budget is tight:

• Reduce seed keywords from 5 to 3
• Reduce size from 100/50 to 50/30
• Skip get_keyword_suggestions (least critical)

Error Handling

• If Serpstat MCP is not connected: use suggest_connectors
• If domain returns 0 keywords: suggest checking the domain spelling or trying a different region
• If credits are insufficient: warn and stop (see Pre-flight Check)
• If a keyword expansion returns no results: skip it, note in summary
• If competitor gap returns empty: note "No unique competitor keywords found — domains likely target very different keyword sets"

Example Invocations

Simple:

"Build me a content plan for serpstat.com"

With context:

"I run ahrefs.com. Create a content strategy for the US market, and show me what topics semrush.com covers that we don't."

Casual:

"What should I write about on my blog? My site is mybrand.com"

Localized:

"Контент-план для сайта example.com на украинском рынке"