8.3 KiB

Raw Blame History

Deploy-Batch Analysis for test_shaft_machining.json

Quick Answers to Your Questions

1. What should the anchor text be at each level?

Tier 1 Articles (5 articles):

Money Site Links: Uses main_keyword variations from project
- "shaft machining"
- "learn about shaft machining"
- "shaft machining guide"
- "best shaft machining"
- "shaft machining tips"
- System tries to find these phrases in content; picks first one that matches
Home Link: Now in navigation menu (not injected into content)
See Also Links: Uses article titles as anchor text

Tier 2 Articles (20 articles):

Lower Tier Links: Uses related_searches from CORA data
- Depends on what related searches were in the shaft_machining.xlsx file
- If no related searches exist, falls back to main_keyword variations
Home Link: Now in navigation menu (not injected into content)
See Also Links: Uses article titles as anchor text

Configuration:

Anchor text rules come from master.config.json → interlinking.tier_anchor_text_rules
Can be overridden in job config with anchor_text_config

2. How many links should be in each article?

Tier 1 Articles:

1 link to money site (https://fzemanufacturing.com/capabilities/shaft-machining-services)
4 "See Also" links (to the other 4 tier1 articles)
Total: 5 links per tier1 article (plus Home in nav menu)

Tier 2 Articles:

2-4 links to tier1 articles (random selection, count is interlinking.links_per_article_min to max)
19 "See Also" links (to the other 19 tier2 articles)
Total: 21-23 links per tier2 article (plus Home in nav menu)

Your JSON Configuration:

"interlinking": {
    "links_per_article_min": 2,
    "links_per_article_max": 4
}

This controls the tiered links (tier2 → tier1). Each tier2 article will get between 2-4 random tier1 articles to link to.

3. Should "Home" be a link?

YES - Home is a link in the navigation menu at the top of every page.

How it works:

The HTML template (basic.html) includes a <nav> menu with Home link
Template line 113: <li><a href="/index.html">Home</a></li>
This is part of the template wrapper, not injected into article content

Old behavior (now removed):

Previously, system searched article content for "Home" and tried to link it
This was redundant since Home is already in the nav menu
Code has been updated to remove this injection

Step-by-Step: What Happens During deploy-batch

Step 1: Load Articles from Database

- Project 1 has generated content already
- Tier 1: 5 articles
- Tier 2: 20 articles
- Each article has: title, content (HTML), site_deployment_id

Step 2: URL Generation (already done during generate-batch)

Tier 1 URLs (round-robin between getcnc.info and textbullseye.com):
- Article 0: https://getcnc.info/{slug}.html
- Article 1: https://www.textbullseye.com/{slug}.html
- Article 2: https://getcnc.info/{slug}.html
- Article 3: https://www.textbullseye.com/{slug}.html
- Article 4: https://getcnc.info/{slug}.html

Tier 2 URLs (round-robin):
- Articles 0-19 distributed across both domains

Step 3: Tiered Links (already injected during generate-batch)

For Tier 1:

Target: Money site URL from project database
Anchor text: main_keyword variations
Links already in generated_content.content HTML

For Tier 2:

Target: Random selection of tier1 URLs (2-4 per article)
Anchor text: related_searches from project
Links already in HTML

Step 4: Homepage Links

Home link is in the navigation menu (template)
No longer injected into article content

Step 5: See Also Section (already injected)

HTML section with links to other articles in same tier

Step 6: Template Application (already done)

HTML wrapped in template from src/templating/templates/basic.html
Navigation menu added
Stored in generated_content.formatted_html

Step 7: Upload to Bunny.net

For each article:
  1. Get site deployment credentials
  2. Upload formatted_html to storage zone
  3. File path: /{slug}.html
  4. Log URL to deployment_logs/
  5. Update database: deployed_url, status='deployed'

For each site's boilerplate pages:
  1. Upload index.html (if exists)
  2. Upload about.html
  3. Upload contact.html
  4. Upload privacy.html

Database Link Tracking

All links are tracked in article_links table:

Tier 1 Article Example (ID: 43):

| from_content_id | to_content_id | to_url | anchor_text | link_type |
|-----------------|---------------|--------|-------------|-----------|
| 43 | NULL | https://fzemanufacturing.com/... | "shaft machining" | tiered |
| 43 | 44 | NULL | "Understanding CNC..." | wheel_see_also |
| 43 | 45 | NULL | "Advanced Shaft..." | wheel_see_also |
| 43 | 46 | NULL | "Precision Machining..." | wheel_see_also |
| 43 | 47 | NULL | "Modern Shaft..." | wheel_see_also |

Tier 2 Article Example (ID: 48):

| from_content_id | to_content_id | to_url | anchor_text | link_type |
|-----------------|---------------|--------|-------------|-----------|
| 48 | NULL | https://getcnc.info/{slug1}.html | "cnc machining services" | tiered |
| 48 | NULL | https://www.textbullseye.com/{slug2}.html | "precision shaft work" | tiered |
| 48 | NULL | https://getcnc.info/{slug3}.html | "shaft turning operations" | tiered |
| 48 | 49 | NULL | "Tier 2 Article 2 Title" | wheel_see_also |
| ... | ... | ... | ... | ... |
| 48 | 67 | NULL | "Tier 2 Article 20 Title" | wheel_see_also |

Note: Home link is no longer tracked in the database since it's in the template, not injected into content.

Your Specific JSON File Analysis

{
  "jobs": [
    {
      "project_id": 1,
      "deployment_targets": [
        "getcnc.info",
        "www.textbullseye.com"
      ],
      "tiers": {
        "tier1": {
          "count": 5,
          "min_word_count": 1500,
          "max_word_count": 2000,
          "models": {
            "title": "openai/gpt-4o-mini",
            "outline": "openai/gpt-4o-mini",
            "content": "anthropic/claude-3.5-sonnet"
          }
        },
        "tier2": {
          "count": 20,
          "models": {
            "title": "openai/gpt-4o-mini",
            "outline": "openai/gpt-4o-mini",
            "content": "openai/gpt-4o-mini"
          },
          "interlinking": {
            "links_per_article_min": 2,
            "links_per_article_max": 4
          }
        }
      }
    }
  ]
}

What This Configuration Does:

Tier 1 (5 articles):
- Uses Claude Sonnet for content, GPT-4o-mini for titles/outlines
- 1500-2000 words per article
- Distributed across getcnc.info and textbullseye.com
- Each links to: money site (1) + See Also (4) = 5 total links (plus Home in nav menu)
Tier 2 (20 articles):
- Uses GPT-4o-mini for everything (cheaper)
- Default word count (1100-1500)
- Each links to: 2-4 tier1 articles + See Also (19) = 21-23 total links (plus Home in nav menu)
- Distributed across both domains
Missing Configurations (using defaults):
- tier1.interlinking: Not specified → uses defaults (but tier1 always gets 1 money site link anyway)
- anchor_text_config: Not specified → uses master.config.json rules

All JSON Fields That Affect Behavior

See MASTER_JSON.json for the complete reference. Key fields:

Top-level job fields:

project_id - Which project's data to use
deployment_targets - Which domains to deploy to
models - Which AI models to use
tiered_link_count_range - How many tiered links (job-level default)
anchor_text_config - Override anchor text generation
interlinking - Job-level interlinking defaults

Tier-level fields:

count - Number of articles
min_word_count, max_word_count - Content length
min_h2_tags, max_h2_tags, min_h3_tags, max_h3_tags - Outline structure
models - Tier-specific model overrides
interlinking - Tier-specific interlinking overrides

Fields in master.config.json:

interlinking.tier_anchor_text_rules - Defines anchor text sources per tier
interlinking.include_home_link - Global default for Home links
interlinking.wheel_links - Enable/disable See Also sections

Fields in project database:

main_keyword - Used for tier1 anchor text
related_searches - Used for tier2 anchor text
entities - Used for tier3+ anchor text
money_site_url - Destination for tier1 links

8.3 KiB Raw Blame History