Scaling SEO Operations: How Automation Helped an Agency 10x Their Output

flowchart TB
    subgraph Input
        A[User pastes URLs in Input Tab] --> B[Zapier triggers workflow]
    end

    subgraph ContentAnalysis
        B --> C[Web scraping]
        C --> D[Clean HTML content]
        D --> E[Extract key elements]
        E --> F[AI Content Analysis]
    end

    subgraph KeywordAnalysis
        B --> G[SEMrush API call]
        G --> H[Sort keywords by ranking]
        G --> I[Sort keywords by traffic]
        G --> J[Sort keywords by search volume]
        H & I & J --> K[AI Keyword Selection]
    end

    subgraph Generation
        F & K --> L[Generate AI-based titles/desc]
        F & K --> M[Generate SEMrush-based titles/desc]
        L & M --> N[Generate combined titles/desc]
    end

    subgraph Output
        N --> O[Results in Output Tab]
        O --> P[Blue: AI-generated]
        O --> Q[Orange: SEMrush-based]
        O --> R[Gray: Combined refinement]
    end

1. Input Process
• Users paste URLs into the input tab of a Google Sheet
• Processing takes about 5-6 minutes per link
• Users can paste multiple links (e.g., 20) and work on other tasks while processing

2. Content Analysis Stage
• System scrapes each webpage
• Cleans up HTML by removing:
• Scripts
• Styling
• Footer
• Navigation
• Extracts key elements:
• Original title
• Meta description (if exists)
• H1 headers
• Main content
• AI analyzes content for:
• Page type (product/collection/etc.)
• Brand voice
• Unique selling propositions
• Writing style

3. Keyword Analysis Stage
• Calls SEMrush API for each URL
• Gets organic keyword data:
• Current rankings
• Search volume
• Traffic data
• Sorts keywords in multiple ways:
• By ranking position
• By traffic volume
• By search volume

4. Generation Process
   The system creates three sets of optimizations:
• Blue Output: Pure AI-generated
• Based on content analysis
• Content-aware but without ranking data
• Orange Output: SEMrush-based
• Uses existing ranking data
• Optimizes for current performing keywords
• Gray Output: Combined refinement
• Merges insights from both approaches
• Provides most relevant recommendations

5. Output Format
• Each URL gets 15 total options:
• 5 title variations per approach (15 total)
• 5 description variations per approach (15 total)
• Color-coded for easy differentiation
• All results appear in the output tab

6. Processing Time
• Average 5-6 minutes per URL
• Can handle batch processing of ~20 URLs at once
• Uses queues to maintain orderly processing