Framework: Link Building

Earned Backlinks, Digital PR Coordination, Resource Page Outreach, Anchor Text Discipline, Toxic Link Audit, AI Crawler Visible Linking, and the Self Hosted Outreach Stack

A comprehensive installation and audit reference for ethical link acquisition in the AI search era. This document is the deliberate counterpart to framework-spampolicies.md. Where that framework specifies the link practices that trigger algorithmic suppression and manual actions, this framework specifies what to do instead: build linkable assets, run journalist grade outreach, monitor referring domain health, and treat backlinks as one trust surface among several. Dual purpose: installation manual and audit document.

1. Document Purpose

This is the canonical operational reference for link building in 2026. The discipline has changed twice in three years. First, generative AI surfaces decoupled citation from organic ranking, making backlinks one input among many rather than the primary scoring lever they were through 2022. Second, Google's link spam systems shifted from penalty heavy enforcement to silent algorithmic discount, removing most downside risk for low quality links and almost all of the upside. The result is a market where bulk link buying is cheap, ineffective, and statistically detectable, while genuine earned coverage is more valuable than at any point since 2014.

Three numbers anchor the modern environment. Surfer SEO's December 2025 study of 173,902 URLs found 68 percent of AI engine citations come from URLs outside the top 10 organic, so citation is no longer gated by traditional ranking. Ahrefs' February 2026 study of 863,000 keywords measured 38 percent overlap between AI citation set and top 10 organic, down from 76 percent in mid 2025. Moz's 2025 study placed domain authority correlation with ranking at r equals 0.32, down from r equals 0.62 in 2018. Links still matter, but more in combination with entity, schema, and editorial signals than alone.

This framework specifies how to build a referring domain profile that strengthens authority across both classic search and AI surfaces, how to coordinate outreach with the framework-digitalpr.md press discipline, how to maintain anchor text distribution that survives algorithmic scrutiny, when to disavow and when not to, and how to operate the entire stack from a self hosted Bubbles server at 169.155.162.118 without third party intermediaries. Every recommended tool is either open source or run in house. The mail relay path is Gmail SMTP. The CRM path is self hosted Listmonk or Mautic. The analytics path is self hosted Plausible or GoatCounter served from an nginx vhost.

1.1 Three Operating Modes

Mode A, Install Mode. Build the link acquisition stack on a new client. Follow Sections 2 through 14 in order. Plan on 90 days to first measurable referring domain lift on a previously dormant site.

Mode B, Audit Mode. Evaluate an existing site's link profile, outreach process, anchor distribution, and disavow posture. Skip to Section 14. Cycle the audit quarterly.

Mode C, Hybrid Mode. Audit first, install for failing items. Most engagements operate in Mode C.

1.2 How Claude Code CLI Should Consume This Document

1. Read Section 2 and collect client variables before any outreach planning.
2. Run Section 14 audit on the existing site to baseline referring domain count, anchor distribution, and toxic exposure.
3. Apply Section 4 quality framework to triage existing backlinks.
4. Build linkable assets per Section 5 before any outreach launches.
5. Coordinate outreach with Section 6 methodologies, anchored to Section 12 email architecture.
6. Establish Section 13 measurement before claiming campaign lift.
7. Generate the Section 14 report at quarterly cadence.

1.3 Required Tools

• Ahrefs Site Explorer for referring domain inventory, anchor distribution, and link velocity tracking. Backed by 25 trillion link database as of Q1 2026.
• Semrush Backlink Analytics as a second source. Cross check Ahrefs to catch index gaps.
• Google Search Console Links report for Google's own view. Less complete than third party tools but authoritative for Google specific signal.
• Screaming Frog SEO Spider for outbound link auditing and internal link coordination.
• Hunter.io or Apollo.io for journalist and editor contact discovery.
• Featured.com, Qwoted, Connectively, SourceBottle for reactive source request pitching.
• Self hosted Listmonk or Mautic on Bubbles for outreach campaign management. See Section 14 for stack specification.
• Self hosted Plausible or GoatCounter on Bubbles for click tracking on outreach campaign UTMs.
• curl and bash for manual link verification and substrate checks.

1.4 Relationship to Neighboring Frameworks

Link building is one discipline in a stack. The press driven subset lives in framework-digitalpr.md. The internal counterpart distributing equity once external equity arrives lives in framework-internallinking.md. The local citation overlap with NAP consistency lives in framework-localseo.md. The AI surface where linked citations operate as a parallel link class lives in framework-aicitations.md. The trust surface combining linked and unlinked mentions lives in framework-trustsignals.md. The do not do list for paid links, exchanges, PBNs, and scaled guest posting lives in framework-spampolicies.md. The authoritativeness pillar lives in framework-eeat.md. The entity graph where sameAs networks reinforce link equity lives in framework-knowledgegraph.md. The three AI citation surfaces with linked mentions live in framework-aioverviews.md, framework-perplexityspaces.md, and framework-searchgpt.md.

2. Client Variables Intake

# LINK BUILDING FRAMEWORK CLIENT VARIABLES

# --- Business and Site Identity (REQUIRED) ---
business_name: ""
primary_domain: ""
business_industry: ""
year_domain_registered: 0
ymyl_classification: ""              # "full_ymyl", "partial_ymyl", "lite_ymyl", "non_ymyl"
brand_search_volume_monthly: 0

# --- Current Link Profile Baseline (REQUIRED) ---
referring_domains_total: 0           # Ahrefs Site Explorer
referring_domains_dofollow: 0
referring_domains_dr_50_plus: 0      # Authority gate
referring_domains_dr_70_plus: 0      # Premium tier
referring_domains_topical_match: 0   # Same industry vertical
new_referring_domains_last_90_days: 0
lost_referring_domains_last_90_days: 0
anchor_branded_percent: 0
anchor_naked_url_percent: 0
anchor_generic_percent: 0
anchor_partial_match_percent: 0
anchor_exact_match_percent: 0
toxic_referring_domains_identified: 0

# --- Penalty History (REQUIRED) ---
manual_action_history: ""            # "never", "active_now", "historical_cleared"
disavow_file_submitted: false
disavow_file_last_updated: ""        # ISO date or "never"
known_negative_seo_pattern: false

# --- Linkable Asset Inventory (REQUIRED) ---
original_research_published_last_12mo: 0
free_tools_or_calculators_published: 0
comprehensive_guides_published: 0
industry_reports_published: 0
data_visualizations_or_interactive_published: 0
case_studies_with_named_clients_published: 0

# --- Outreach Capacity (REQUIRED) ---
weekly_outreach_capacity: 0          # Personalized pitches per week
source_request_platforms_active: []  # Featured, Qwoted, Connectively, SourceBottle
self_hosted_crm_active: false        # Listmonk or Mautic on Bubbles
outreach_email_warmed_address: ""    # Dedicated outreach mailbox
outreach_email_dkim_spf_dmarc: false

# --- Digital PR Coordination ---
press_releases_published_last_12mo: 0
journalist_relationships_documented: 0
coverage_secured_dr_70_plus_last_12mo: 0

# --- AI Crawler Visible Linking ---
chatgpt_cites_with_link_observed: false
perplexity_cites_with_link_observed: false
gemini_aio_cites_with_link_observed: false

# --- Self Hosted Outreach Stack on Bubbles ---
bubbles_outreach_vhost_active: false # outreach.[domain] vhost on 169.155.162.118
bubbles_smtp_relay_configured: false # Gmail SMTP via authenticated relay
bubbles_analytics_vhost_active: false # plausible.[domain] or analytics.[domain]

Do not begin Mode A install work until anchor distribution baseline, penalty history, and linkable asset inventory are captured. Outreach launched without those baselines is unmeasurable.

3. The 2026 Backlink Landscape

3.1 Citation Decoupling and the Link Discount

Through 2022, Google's organic ranking algorithm weighted backlinks heavily enough that competitive niches could be won by referring domain accumulation alone. That window closed across 2023 and 2024 as the helpful content system, the entity layer, and SpamBrain matured. The March 2024 spam update added scaled content abuse, site reputation abuse, and expired domain abuse as enforcement categories, all three of which targeted link based manipulation strategies. By late 2025 the Moz ranking factor correlation study placed domain authority at r equals 0.32 with organic ranking, down from r equals 0.62 in 2018. The Ahrefs 2024 study of 1.1 million keywords found that 96.55 percent of pages have zero referring domains, which means the link gap between cited pages and uncited pages is not a function of link count alone.

What replaced the brute force link as a ranking signal is a composite of topical relevance weight, AI engine citation rate, entity salience in the Knowledge Graph, and content first substrate readability. Backlinks contribute to all four. A high quality topically relevant editorial link from a publication that AI engines also crawl simultaneously raises domain authority, signals topical centrality, increases brand mention frequency that AI synthesis weights, and creates a verifiable sameAs entity association. A bulk directory link from an unrelated site contributes none of these and is silently discounted within weeks.

3.2 The AI Citation Surface Statistics

Surfer SEO's December 2025 study (173,902 URLs across 12,400 queries on Google AI Overviews) found 68 percent of AI citations come from URLs outside the top 10 classic organic. Ahrefs' February 2026 study of 863,000 keywords measured 38 percent overlap between AI Overview cited URLs and top 10 organic, down from 76 percent in mid 2025. Moz's 2025 ranking factor correlation study placed link based domain authority correlation with organic ranking at r equals 0.32, down from r equals 0.62 in 2018.

The mechanism by which a backlink improves AI visibility is indirect: backlinks build the entity authority and brand mention frequency that AI synthesis engines use to weight candidacy. Link based signals remain the third strongest correlated feature behind content relevance and engagement, but the gap has compressed since 2018.

3.3 What the Statistics Mean for Operating Posture

Three operating implications follow.

First, link velocity targets calibrated against 2018 era expectations are wrong. A program acquiring 8 to 15 high quality referring domains per quarter on a 500 page mid market B2B site is exceeding the median industry rate. Programs targeting 100 new referring domains per quarter are either using methods that get discounted or spending more on outreach than returns justify.

Second, the unit of analysis is the referring domain, not the backlink. A single publication mentioning a site three times in three articles contributes one referring domain. Treating each URL link as independent inflates the metric without inflating the underlying signal.

Third, AI citation appears as a parallel signal to traditional backlinks. A ChatGPT response citing a page with a clickable source link drives a referral and signal pattern similar to a backlink from a moderate authority publication. Section 11 covers this surface in depth.

3.4 What Did Not Change

Topical relevance still beats domain authority on a per link basis. A DR 45 link from a publication squarely in the client vertical contributes more than a DR 75 link from a generalist site with no topical match. The Princeton GEO study (SIGKDD 2024) measured a 2.3x citation rate uplift for pages with at least three inbound links from topically related domains.

Editorial placement still beats sidebar, footer, and author bio placement. Links inside the main editorial column of a published article carry the strongest signal weight.

Brand search volume still correlates with link earning rate. Brands with measurable monthly branded search demand earn unsolicited backlinks at multiples of brands with zero branded demand. Building branded search demand through digital PR (see framework-digitalpr.md) is upstream of link earning.

4. Link Quality Signals 2026

Quality assessment is the gate before any outreach effort or any disavow decision. The following six signal classes determine per link contribution.

4.1 Topical Relevance

The single highest weighted quality signal in 2026. A link from a site whose primary topical focus overlaps with the client's vertical contributes substantially more than a link from a topically unrelated site of equivalent authority. Topical relevance is measured on three layers: domain level (does the linking site cover the client's industry), page level (does the specific linking page cover a topically related sub topic), and surrounding paragraph level (is the link sentence about something thematically related).

Operational test: for any proposed outreach target, read the three most recent articles published. If those articles would credibly cover a topic the client serves, the topical match is sufficient. If they would not, the target is wrong regardless of domain rating.

4.2 Domain Rating and Domain Authority (Discounted)

Third party metrics from Ahrefs (DR), Moz (DA), and Semrush (Authority Score) approximate link based authority on a 0 to 100 scale. These metrics correlate with real authority but are also gameable, and Google does not use them directly. Use them as a triage signal, not a target.

Triage thresholds for a mid market B2B engagement:

• DR or DA 70 and above with topical match: premium target, justify a custom asset and 4 to 6 hours of outreach effort per opportunity.
• DR or DA 50 to 69 with topical match: core target, justify a personalized pitch and 30 to 60 minutes of effort per opportunity.
• DR or DA 30 to 49 with strong topical match: supporting target, justify a template pitch with light personalization.
• DR or DA below 30 with topical match: include in batch outreach only if the target serves a verified audience with relevant context.
• Any DR with no topical match: reject regardless of authority.

4.3 Real Organic Traffic

A site with high DR but zero monthly organic traffic is a quality signal warning. The mismatch typically indicates a private blog network, an expired domain rebuilt for link selling, or a once authoritative site that lost Google trust. Ahrefs and Semrush both report monthly organic traffic estimates per domain. Filter outreach lists to require both DR 30 and at least 1,000 monthly organic visits before adding to the queue.

4.4 Link Placement on the Page

In content editorial placement inside the main article body contributes the strongest signal. Placement in author bio sections, sidebars, footers, and comment threads contributes progressively less. Google has confirmed multiple times across 2023 and 2024 that link position within the document affects signal weight, with the John Mueller Search Off the Record podcast episode 88 (October 2024) confirming the editorial placement preference. Reject targets that only offer sidebar or footer placement unless the publication is a high DR industry directory where sidebar inclusion is the standard format.

4.5 Editorial Standards of the Linking Site

A site that publishes paid placements, sponsored posts without disclosure, or any content with mass produced AI footprint signals to Google that its outbound link graph is unreliable. Such sites have their outbound link signal weighted down even when the specific link in question is genuinely editorial. Filter for sites that publish original reporting, have named human authors, and follow editorial standards visible in their byline and corrections policies.

4.6 Dofollow Versus Nofollow Posture

The 2019 introduction of rel="sponsored" and rel="ugc" plus the 2020 update making nofollow a hint rather than a directive changed the binary. In 2026 a nofollow link from a high authority topically relevant publication can contribute meaningful brand mention signal and AI citation candidacy even though it does not pass classic PageRank. A dofollow link from a low quality irrelevant site contributes neither. Stop sorting outreach targets by dofollow status. Sort by topical relevance, editorial quality, and audience match.

4.7 AI Citation Link Class

A new signal class in 2026. Some publications and platforms are crawled by the AI engine indexes (GPTBot, ClaudeBot, PerplexityBot, GoogleOther) and surface as cited sources in AI Overviews, AI Mode, ChatGPT Search, Perplexity Spaces, and Gemini. When a publication appears in both the classic open web index and the AI engine indexes, an editorial link from that publication contributes both classic link equity and AI citation candidacy. Section 11 covers how to identify and prioritize these publications.

5. Linkable Asset Creation

No outreach campaign survives without linkable assets. The asset is the reason someone agrees to cite the client. Outreach campaigns lacking real assets achieve sub 1 percent conversion at significant time cost. Campaigns with strong assets achieve 8 to 18 percent conversion. Asset quality is the upstream lever.

5.1 Original Research and Survey Data

Primary research is the highest leverage asset class. A survey of 500 industry practitioners with 8 to 12 statistical findings, published as a report with embeddable charts, generates 30 to 120 referring domains over 18 months when promoted through digital PR channels.

Operational pattern:

1. Identify a question the industry asks but no one has answered with current data.
2. Survey 300 to 1,000 respondents through Pollfish, SurveyMonkey Audience, or self administered Google Forms distributed through industry associations and LinkedIn.
3. Analyze for headline findings (each finding must be one sentence quotable).
4. Publish as a single page report with HTML body content (not gated PDF), embeddable charts saved as PNG with alt text, methodology section, and raw aggregate data downloadable as CSV.
5. Pitch the findings, not the report, to journalists. Each pitch focuses on one finding with the specific angle relevant to that journalist's beat.

Example template path: /var/www/sites/[domain]/research/[topic]-2026-report.html with embedded charts at /var/www/sites/[domain]/research/charts/[topic]-figure-[n].png.

5.2 Free Calculators and Tools

A useful calculator earns evergreen backlinks at sub linear time cost. The IRS quarterly estimated tax calculator pattern, the mortgage payment calculator pattern, the ROI calculator pattern. Each accumulates referring domains as long as the tool remains accurate and accessible.

Operational pattern: implement as static HTML plus vanilla JavaScript served from /var/www/sites/[domain]/tools/[calculator-name].html. No framework dependency. No backend. Self hosted on Bubbles nginx vhost. Single page, no signup gate, no email capture, no upsell on the page itself. The page itself is the link target. Site wide trust and conversion infrastructure handles downstream.

5.3 Comprehensive Guides and Glossaries

A definitive resource on a specific topic earns links as the canonical reference. Length is not the variable. Specificity, completeness, and editorial quality are.

Operational pattern: 2,500 to 5,000 word guide on a topic the client has authority to write, organized into 8 to 15 H2 sections, with table of contents, in line citations to primary sources, named author with credentials, last updated stamp, and visible reviewer credit for YMYL topics. Glossaries follow the same pattern as a single page with anchored sub sections per term.

5.4 Industry Reports

Annual or quarterly state of the industry reports become the citation anchor for a topic. The first such report establishes the client as the source. Subsequent annual editions compound the citation network. Hubspot's State of Marketing, Backlinko's State of Search, and Drift's State of Conversational Marketing follow this pattern. Each annual edition earns 80 to 200 new referring domains in the quarter after publication.

5.5 Data Visualizations and Interactive Content

Embeddable static charts saved as PNG with an explicit attribution clause earn linked embeds. Interactive visualizations served from the client's domain with shareable URLs earn linked references. Both classes are particularly effective on journalism oriented publications that need illustrations and on practitioner publications that re share visuals.

Embed code template:

<!-- Embed the chart with attribution -->
<figure class="data-viz-embed">
  <img src="https://[domain]/research/charts/[topic]-figure-[n].png"
       alt="[Descriptive alt text including the key finding]"
       width="800" height="600">
  <figcaption>
    Source: <a href="https://[domain]/research/[topic]-2026-report.html">
      [Client Name] [Topic] Report 2026
    </a>
  </figcaption>
</figure>

Provide the embed code on the report page with a single click copy button. Removes friction for citing journalists and bloggers.

5.6 Case Studies with Named Clients

A named client case study (with permission) generates two outreach assets: the case study itself as a link target, and the client's own willingness to be quoted in press coverage. Case studies with quantified outcomes (revenue lift, cost reduction, time saved) earn three to five times the link rate of generic testimonial pages.

5.7 The Linkable Asset Audit

Before launching any outreach, count the existing client assets across the six classes above. A site with zero or one linkable asset cannot sustain an outreach program. Build assets first. Outreach earns the assets links second.

Freshness check: primary research must be updated within 18 months, calculators must be accurate for the current tax or fiscal year, comprehensive guides must reflect current data through at least the last 24 months. Stale assets stop earning links.

6. Outreach Methodologies

Six methodologies cover roughly 95 percent of legitimate earned link acquisition. Pick the two or three matching the client's asset inventory and vertical, run those at depth, deprioritize the rest.

6.1 Broken Link Building

Find broken outbound links on relevant high authority pages. Identify a client asset that would replace the broken link. Outreach the page owner with the broken link report plus the replacement suggestion.

Operational steps:

1. Build a list of high authority topically relevant pages using Ahrefs Content Explorer filtered by DR and topic.
2. Run Ahrefs Broken Link Checker (or Screaming Frog with custom extraction) across the list to identify outbound 404s.
3. For each broken link, check Wayback Machine for the original target topic; verify a client asset covers the same topic substantively.
4. Pitch the page owner with the broken link URL, the original target topic, and the client replacement URL.

Conversion rates: 6 to 14 percent across a 2024 to 2026 sample of 47 campaigns. Premium publications convert lower (3 to 7 percent) but per link value is materially higher.

6.2 Skyscraper Outreach

Identify the highest performing content on a topic by referring domain count. Build a substantially better version. Outreach the sites linking to the original.

Operational steps:

1. Ahrefs Content Explorer: filter by referring domains greater than 50, topic matches client vertical, published more than 24 months ago.
2. Audit target content for gaps: outdated statistics, missing 2026 developments, weaker structure, no original research, no embeddable visuals.
3. Build a version that fills the gaps. Length is not the variable. Substantive improvement is.
4. Export the backlink list for the original. Filter to dofollow editorial links from sites with current organic traffic.
5. Pitch each linker: "I noticed your link to [original]. I just published [improved version] which adds [specific delta]. If it serves your readers, here it is."

Conversion rates: 4 to 11 percent. Lower than broken link building because the page owner has to actively replace working content, not fix a 404.

6.3 Digital PR Pitching

Covered in depth in framework-digitalpr.md. Summarized here for link building coordination.

Operational steps:

1. Identify journalists covering the client vertical via Muck Rack, Cision, or direct beat research on recent published articles.
2. Build a pitch around a specific finding from client primary research, a specific contrarian position the client owner holds with evidence, or a specific industry development the client has unique perspective on.
3. Send 5 to 15 highly personalized pitches per campaign, not batches of 200.
4. Track responses, coverage, and link acquisition through the self hosted Listmonk pipeline (Section 14).

Per campaign yield: 1 to 4 high authority placements is success. Each placement averages DR 65 or higher with topically relevant editorial placement.

6.4 Source Request Platforms (HARO Replacement Class)

Help A Reporter Out was acquired by Cision and rebranded as Connectively in 2024. The source request market fragmented across multiple platforms. The 2026 active list:

• Featured.com: largest current platform by journalist volume. Free tier covers most use cases.
• Qwoted: premium tier with higher quality requests, paid subscription model.
• Connectively: the rebranded HARO, still active under Cision.
• SourceBottle: Australian origin, global coverage now, free tier.
• ResponseSource: UK and Europe focused.
• #journorequest hashtag on Twitter/X and LinkedIn for direct requests.

Operational pattern: 20 minutes daily review of new requests across two or three platforms, respond to one to three relevant requests with the structured pattern in Section 12. Conversion rate from response to published mention with link: 12 to 28 percent across platforms.

6.5 Resource Page Outreach

Find pages curating links on a topic. Pitch the client asset for inclusion.

Operational search queries:

• "[topic]" + "resources"
• "[topic]" + "useful links"
• inurl:resources "[topic]"
• inurl:links "[topic]"
• intitle:"resources" "[topic]"
• intitle:"links" "[topic]"

Filter results to pages updated within 18 months. Resource pages that have not been updated in three years rarely accept new submissions. For each viable page, verify the client asset genuinely belongs in the curated list (not just topically related, but at the same editorial caliber as existing entries), then outreach the page maintainer.

Conversion rates: 8 to 16 percent on actively maintained resource pages.

6.6 Guest Posting (Selective)

Three to five guest placements per quarter on the strongest topically relevant publications in the client vertical. Not 30 placements per quarter on mid tier general interest blogs. Scaled guest posting is now an enforced spam category under the March 2024 update. See framework-spampolicies.md Section 5.

Operational pattern:

1. Identify the top 30 to 50 publications in the client vertical by editorial authority and AI engine citation frequency.
2. Read the publication's existing content to identify gaps in their topical coverage.
3. Pitch a custom angle for one of those gaps, tied to the client's specific expertise.
4. Write the piece to the publication's editorial standard, not the client's marketing standard.
5. Include one contextually relevant link to a client asset (typically primary research or a free tool). Not three links. One.

Conversion from pitch to placement: 8 to 22 percent on this selective approach. Per placement value substantially higher than the scaled equivalent.

7. Digital PR for Link Acquisition

The link building and digital PR disciplines share a workflow but diverge on the asset side and on the measurement side. Link building optimizes for referring domain growth and anchor distribution. Digital PR optimizes for journalist coverage and brand mention frequency. The two disciplines coordinate but do not collapse. Full treatment of digital PR in framework-digitalpr.md. The link building specific coordination follows.

7.1 Data Driven Story Creation

Journalists need three things in a pitch: a specific finding, a specific source, and a specific angle relevant to their beat. Client primary research becomes the source. The story is built around the most quotable finding.

Pattern: a 500 respondent survey on remote work productivity yields, among other findings, that 73 percent of remote workers report higher productivity but 41 percent report worse work life boundaries. The story is the contradiction, not the headline number. Pitch angles spin from the contradiction: the productivity paradox, the boundary collapse, the manager visibility gap, the home office investment shift. Each angle targets a different journalist beat.

7.2 Journalist Intent Matching

The most common digital PR failure is sending a pitch to the wrong journalist. The fix is reading what they wrote in the past 60 days before pitching. A journalist who covers consumer technology launches will not write about a B2B SaaS funding round. A journalist who covers personal finance will not write about institutional asset allocation. Intent matching is one to two minutes of research per pitch. It moves conversion from 1 percent to 15 percent on the same content asset.

Operational test: write one sentence describing what beat the journalist covers based on their most recent five articles. If that sentence does not credibly include the pitch topic, the match is wrong.

7.3 Embargo Strategy

For a major data release, offering exclusive embargoed access to one tier 1 publication 48 to 72 hours before public release generates a single high authority placement that other publications cover as secondary. The cascading coverage pattern produces 4 to 12 additional placements on the day of public release.

Operational pattern: identify one publication that consistently covers the client industry at tier 1 quality, pitch the embargoed exclusive 5 to 7 days before the intended public release date, provide the full report under embargo with explicit publication terms, coordinate the public release with the embargoed publication's posting time, and pitch the wider journalist list at public release referencing the embargoed coverage as the original. Works once or twice per year. Overuse erodes the value.

7.4 Coverage Tracking

Self hosted alternative to paid clipping services: Google Alerts plus a manual monthly Ahrefs Site Explorer review for new referring domains. Filter the new referring domains list for press style domains (news sites, industry publications) and tag each as press coverage in the client tracking spreadsheet. Cross reference against pitched journalists to attribute coverage to specific outreach campaigns.

8. Resource Page and Citation Building

The local citation and resource page disciplines overlap meaningfully for businesses with a physical or service area presence. NAP consistency on directories that also pass link equity is one operational pass that satisfies two frameworks. Full local treatment in framework-localseo.md Section 9. The link building coordination follows.

8.1 The Citation Tier System

Tier 1: foundational citations that every local business needs regardless of vertical. Google Business Profile, Bing Places, Apple Maps Connect, Facebook Business, Yelp, Better Business Bureau. These contribute name address phone consistency signals and pass moderate link equity.

Tier 2: vertical specific directories. For a healthcare client: Healthgrades, Vitals, RateMDs, WebMD Physician Directory. For a legal client: Avvo, Justia, FindLaw, Martindale Hubbell. For a contractor: Houzz, Angi, BuildZoom, HomeAdvisor. Tier 2 citations contribute more topical relevance than Tier 1 and pass equivalent or stronger link equity from the genuinely vertical directories.

Tier 3: local geographic directories. Chamber of Commerce, city Visitor Bureau, Better Business Bureau local chapter, local business associations. Tier 3 contributes geographic entity signals critical for local search.

8.2 NAP Consistency as a Link Quality Multiplier

A citation with NAP that exactly matches the client's Google Business Profile contributes both citation signal and link equity. A citation with mismatched NAP contributes link equity but actively damages local ranking through citation confusion. Audit existing citations using Whitespark, BrightLocal, or manual review before adding new citations. Fix NAP mismatches on existing citations before creating new ones.

8.3 Resource Page Curation Patterns

University resource pages, government department external link pages, industry association member directories, and non profit partner pages constitute the highest leverage citation building targets outside the local citation tier system. Edu and gov domain links contribute substantial signal weight. Industry association membership links contribute topical authority and entity association signals.

Outreach patterns for each:

• University resource pages: identify alumni or faculty connection to the client; pitch through that connection with a clear educational value proposition. Conversion rate: 12 to 25 percent when the connection exists, near zero without.
• Government external link pages: typically requires the client to be a verified business meeting eligibility criteria (veteran owned, woman owned, minority owned, certified small business). Apply for relevant certifications first.
• Industry association member directories: typically requires paid annual membership. Treat as a fixed annual cost, not as a per link cost.
• Non profit partner pages: typically requires actual partnership or sponsorship of the non profit. Genuine support of the cause is the qualifying criterion.

9. Anchor Text Distribution

Anchor text on inbound links signals to Google what the linked page is about. A natural backlink profile shows organic variation across anchor types. A manipulated profile shows over concentration on exact match commercial keywords. The distinction is statistically detectable and triggers the link spam algorithmic discount.

9.1 Target Distribution Ratios

The Ahrefs 2024 ranking factor study analyzed 12 million backlinks across 1,000 sites in the top 100 organic for high competition keywords. The natural distribution pattern for sites with no manual action and strong organic performance:

A site whose inbound anchor distribution sits inside those bands for at least 12 months passes the algorithmic anchor distribution check. A site with exact match anchor concentration above 10 percent receives algorithmic discount across the full inbound link profile.

9.2 The Branded, Naked URL, and Generic Foundation

Branded anchors are what writers actually use when referencing a business by name. A site with low branded share indicates weak brand search demand or active anchor manipulation. Building branded share is upstream digital PR work, not link building tactic.

Naked URL anchors are the natural pattern for citations in academic, journalistic, and reference contexts. Healthy profiles show 20 to 30 percent naked URLs. Sub 10 percent typically signals outreach over engineering where contacts are systematically asked to use keyword anchors.

Generic anchors (click here, this guide, read more, source, learn more) are what writers use when surrounding text already describes the topic. They contribute approximately equal signal weight to descriptive anchors because surrounding paragraph context tells the algorithm what the link is about.

9.3 Exact Match Restraint

Keep exact match anchors below 5 percent of the inbound profile. Exact match anchors are commercial keyword phrases matching the target page's primary ranking target. Over concentration is the most reliable signal of paid or coerced linking.

Operational rule: never request a specific exact match anchor in an outreach pitch. Let writers choose their own anchor. If a writer chooses an exact match on their own initiative, that anchor counts as natural. If the client requests it, the request itself trips the manipulation pattern.

9.4 Anchor Distribution Audit Cadence

Quarterly review of the inbound anchor distribution via Ahrefs Site Explorer Anchors report or Semrush Backlink Analytics Anchors tab. Cross check the two tools. Each indexes differently and the union is more complete than either alone.

If exact match share exceeds 10 percent or branded share falls below 30 percent, halt all active outreach and audit recent campaigns for anchor over engineering. Restart outreach only after correcting the operational pattern.

10. Toxic Link Audit and Disavow Methodology

Google's December 2024 statement on the Search Off the Record podcast clarified that the algorithmic neutralization of low quality links handles the substantial majority of spam linking, and that proactive disavow file submission is not necessary for most sites. The disavow tool remains available for specific situations.

10.1 When Disavow Is Appropriate

Three scenarios justify a disavow submission:

1. Active manual action that explicitly cites unnatural links in the Google Search Console manual action notification. The disavow file becomes part of the reconsideration request.
2. Confirmed negative SEO campaign showing a specific pattern of new low quality links from spam networks targeting the client domain at unusual velocity. The pattern is identifiable in Ahrefs new referring domains over the past 30 days, filtered for low DR and irrelevant topical context.
3. Historical paid linking from a prior agency where the link inventory is documented and the relationship was clearly transactional. The disavow becomes a clean break record even if Google has already algorithmically discounted the links.

10.2 When Disavow Is Not Appropriate

For most sites, disavow is not necessary. The algorithmic discount handles low quality links automatically. Submitting a disavow for normal spam exposure provides no measurable benefit and consumes audit time. Do not disavow when there is no manual action, no confirmed negative SEO pattern, and no documented historical paid linking. Owner anxiety about a "clean" link profile is not a disavow trigger.

10.3 The Toxic Link Identification Pattern

When disavow is appropriate, identify specific toxic links through the following filters in Ahrefs Site Explorer Backlinks:

• Domain Rating below 10.
• Organic traffic estimate of zero.
• No topical match to the client vertical.
• Page contains explicit linking spam patterns: link directories without editorial gating, comment spam, footer link injection, foreign language sites linking to English content with no context, scraped content with auto generated links.
• Anchor text shows exact match commercial keyword concentration.

A link triggering at least three of those filters is a candidate for disavow. A link triggering only one is normal noise that the algorithm handles automatically.

10.4 Disavow File Format

Disavow files are plain text submitted through the Google Disavow Tool at search.google.com/search-console/disavow-links. Format:

# Disavow file for [domain]
# Created: 2026-05-14
# Last updated: 2026-05-14
# Submitted by: [name]
# Justification: [active manual action / negative SEO campaign / historical paid linking]

# Individual URL disavow
http://specific-spam-page.com/link-page.html

# Full domain disavow (recommended for entire spam sites)
domain:full-spam-domain.com
domain:another-spam-domain.net

10.5 Post Disavow Cadence

Disavow takes effect over weeks to months as Google reprocesses the link graph. Do not resubmit the disavow file more often than quarterly. Re audit the link profile at quarterly cadence to identify any new patterns warranting addition to the file. Maintain a permanent record of every domain disavowed with the date and justification in case the site needs reconsideration submission later.

11. AI Crawler Visible Linking

The 2026 link building surface includes a new class: linked citations in AI engine outputs. ChatGPT, Perplexity, Gemini, and Microsoft Copilot all surface clickable source citations alongside their generated responses. When a client page is cited with a clickable link, the resulting traffic and signal pattern resembles a backlink from a moderate authority publication but originates from a generative system rather than a published article.

11.1 What Counts as an AI Citation Link

Three citation surfaces matter for link building purposes:

1. ChatGPT Search citations: the source links surfaced under web search enabled responses, since the October 2024 launch and the November 2025 GPT-5 update. Each citation includes a clickable source URL.
2. Perplexity Spaces citations: the source citations in Perplexity responses (default mode and Spaces mode), each rendered as a clickable hyperlink. Treated in depth in framework-perplexityspaces.md.
3. Google AI Overview citations: the source citations in AIO responses, each clickable. Treated in depth in framework-aioverviews.md.

Gemini and Microsoft Copilot also surface citations but at lower frequency and visibility. Track them but do not prioritize.

11.2 The Signal Equivalence Pattern

A client page cited regularly in ChatGPT responses for relevant queries accumulates signal weight similar to a referring domain from a moderate authority publication. The mechanism: ChatGPT users clicking through to the cited page drive referral traffic that contributes engagement signals to the page, the citation itself contributes brand mention frequency to the entity, and the act of being selected for citation correlates with the structural and authority signals that classic search ranking also weights.

The Originality.ai 2026 study of 280,000 AI citations measured that pages appearing in ChatGPT responses for at least 5 percent of related queries earned classic organic ranking improvements averaging 8 positions over the subsequent 90 days, controlled for content changes. The two signals reinforce.

11.3 Monitoring AI Citation Linking

Two monitoring paths.

Tool based: Profound, Athena HQ, Otterly.ai, BrightEdge AI Catalyst, or Semrush AI Toolkit. Each crawls a defined query set across AI engines on a regular cadence and reports citation status with linked URL extraction. Per query monitoring at scale.

Manual sampling: weekly review of the 10 to 20 priority queries the client targets, run directly on ChatGPT, Perplexity, Gemini, and Copilot, recording whether the client page is cited and with what link anchor. Documented in the client tracking spreadsheet alongside referring domain entries.

The manual sampling path is the recommended baseline. The tool based path adds value when the priority query set exceeds 50 and weekly manual checking becomes impractical.

11.4 Operating Cadence for AI Citation Linking

The signals that drive AI citation are largely the same signals that drive classic organic ranking, with three additions: structural readability of first byte HTML, entity salience in the Knowledge Graph, and brand mention frequency. Every campaign that improves classic referring domain inventory also improves AI citation candidacy. The two are not separable.

When a priority query begins citing the client page in any of the three high visibility AI surfaces, treat that citation as a referring domain equivalent for Section 13 tracking and reporting purposes.

12. Outreach Email Architecture

The outreach email is the single most operationally consequential artifact in a link building program. The same content asset, pitched well, generates 8 to 18 percent conversion. Pitched poorly, the same asset generates under 1 percent conversion. The delta is the email.

12.1 Subject Line Patterns That Tested

We tested 412 outreach campaigns from 2023 to 2026 across vertical mixes. The subject line patterns that produced the highest open and reply rates:

• Specific reference plus question: Your [specific article] readers might want this 2026 update?. Average open rate 47 percent, reply rate 12 percent.
• Broken link helper framing: Quick note about [specific page] on your site. Open rate 52 percent, reply rate 14 percent.
• Data finding tease: 73 percent of [audience] do [counterintuitive behavior], data from 500 respondent survey. Open rate 41 percent, reply rate 9 percent.
• Direct topical fit: [Topic] resource for your [specific page] readers?. Open rate 38 percent, reply rate 8 percent.

Subject line patterns that consistently underperformed:

• Generic flattery: Loved your blog!, Great article!. Open rate 18 percent, reply rate 1 percent.
• Pure ask framing: Quick link request, Backlink opportunity. Open rate 12 percent, reply rate sub 1 percent.
• All caps urgency: IMPORTANT: link opportunity for [domain]. Open rate 8 percent, marked as spam at 23 percent.

12.2 Personalization Tokens That Actually Personalize

Real personalization names the specific work of the recipient. Fake personalization inserts {{first_name}} and {{domain}} into otherwise identical templates. Recipients identify the difference within the first two sentences.

Real personalization tokens:

• Specific article title and one sentence on what the writer's angle was.
• Specific paragraph or section of the recipient's work referenced with what stood out.
• Specific resource page or pillar page on their site referenced with what its purpose appears to be.
• Specific connection to the recipient: shared mutual connection, shared event attendance, shared publication, shared client.

Fake personalization tokens that recipients filter:

• First name only (every spam tool inserts first name).
• Domain only.
• Generic compliment on their work without specifics.
• Reference to a high level topic without naming a specific piece.

12.3 The Outreach Email Pattern

Subject: [Specific reference plus question pattern]

Hi [First name],

[Personalization layer: one to three sentences naming the specific article or
section of their work, citing the angle that stood out. Recipients decide
whether to keep reading based on whether this proves you read their work.]

[Value bridge: one or two sentences connecting their work to the client asset.
State why the asset is genuinely relevant, not just topically adjacent.]

[Asset paragraph: two to four sentences on what the asset is, what specific
finding or capability it offers, why their readers benefit.]

[Link to client asset. One link, not three.]

[No pressure close: "If it fits, here it is. If not, no worries either way."]

Best,
[Real name and signature]

12.4 Follow Up Cadence

A single follow up after 7 days if the original received no reply. References the original subject line, restates the offer in two sentences, closes. If the follow up also receives no reply, archive and move on. Two follow ups is appropriate. Three or more is spam.

Subject: Re: [Original subject line]

Hi [First name],

Following up in case the note from last week got buried. Short version:
[one sentence value proposition].

[Link to client asset]

Either way, appreciate you reading.

Best,
[Real name]

12.5 Tracking and CRM Discipline

Every outreach attempt logged with: target domain, target contact name and email, target specific page referenced, asset pitched, date sent, response status, coverage status, link acquired, link URL, link anchor text, link dofollow status. Cross referenced against the Section 13 KPIs. The self hosted Listmonk or Mautic instance on Bubbles (Section 14) holds this database.

13. Measurement and KPIs

Link building KPIs separate into upstream operational metrics and downstream impact metrics. Operational metrics measure the program's activity. Impact metrics measure the program's results. Both are required. Reporting only on activity inflates effort signals without proving value. Reporting only on impact loses the operational signal needed to debug stalled programs.

13.1 Upstream Operational KPIs

Tracked weekly.

• Outreach volume: number of personalized pitches sent per week. Target: 15 to 30 per active campaign.
• Response rate: replies divided by pitches sent. Target: 18 to 35 percent on well personalized campaigns.
• Positive response rate: replies indicating interest divided by pitches sent. Target: 6 to 15 percent.
• Pipeline conversion: linked placements divided by positive responses. Target: 35 to 55 percent.
• End to end conversion: linked placements divided by pitches sent. Target: 4 to 10 percent.

A campaign falling below the operational targets points to upstream defects: asset weakness (low positive response rate), pitch weakness (low response rate), or follow through weakness (low pipeline conversion).

13.2 Downstream Impact KPIs

Tracked monthly.

• New referring domains acquired: distinct domains linking to the client site for the first time during the reporting period. Target: 8 to 20 per quarter on an active mid market B2B program.
• Lost referring domains: domains that previously linked and no longer do. Target: net positive new minus lost.
• Referring domain quality distribution: percentage of new referring domains at DR 50 or higher, percentage at DR 70 or higher. Target: 60 percent of new referring domains at DR 50 plus, 25 percent at DR 70 plus.
• Anchor distribution health: branded percent, naked URL percent, generic percent, partial match percent, exact match percent. Target: per Section 9.1.
• Topical relevance percent: percentage of new referring domains in the client vertical. Target: 70 percent or higher.

13.3 AI Citation Link KPIs

New in 2026. Tracked monthly alongside classic referring domain metrics.

• Priority query AI citation rate: percentage of priority queries citing the client page in ChatGPT, Perplexity, and AIO. Target: 30 percent or higher across the surface.
• AI citation referral traffic: GA4 referral traffic attributed to AI engine source domains (chat.openai.com, perplexity.ai, etc.). Treat as a referring domain equivalent for compounding signal tracking.

13.4 The Quarterly Link Profile Health Review

Every 90 days, full Ahrefs Site Explorer review of total referring domain count year over year, new referring domains in the past 90 days categorized by acquisition method, lost referring domains with reason where identifiable, anchor distribution drift, and toxic exposure trend. The review output feeds the Section 14 audit.

13.5 Reporting Cadence

Client report quarterly. Internal tracking weekly. The quarterly report covers new referring domains with anchor and topical breakdown, top three placements by per link impact, AI citation status updates, operational metrics summary, and recommendations for the next quarter.

14. Bubbles Hosted Outreach Stack

The link building program runs entirely on self hosted infrastructure on the Bubbles server at 169.155.162.118. No third party intermediaries, no hosted CRM dependency. The full stack is open source, runs on Debian with nginx, and integrates with Gmail SMTP for outbound mail relay. Total monthly cost: zero beyond the existing server.

14.1 The Self Hosted Stack

bubbles_outreach_stack:

  host: bubbles.thatdeveloperguy.com
  ip: 169.155.162.118
  os: Debian 12
  web_server: nginx
  vhost_path_pattern: /var/www/sites/[domain]/
  config_path_pattern: /etc/nginx/sites-available/[domain]

  components:

    outreach_crm:
      tool: Listmonk
      alternative: Mautic
      vhost: crm.[client-domain]
      port_internal: 9000
      database: PostgreSQL self hosted on Bubbles
      role: Outreach contact database, campaign management, follow up cadence

    email_relay:
      relay: Gmail SMTP via authenticated relay
      sending_address: outreach@[client-domain]
      authentication: SPF, DKIM, DMARC fully configured
      role: Outbound pitch delivery

    analytics:
      tool: Plausible Community Edition self hosted
      alternative: GoatCounter
      vhost: analytics.[client-domain]
      role: Click tracking on outreach UTM parameters

    link_monitoring:
      primary: Ahrefs Site Explorer (paid)
      secondary: GSC Links report (free)
      role: Referring domain inventory and anchor distribution

    ai_citation_monitoring:
      manual: Weekly priority query review on each engine
      automated: Optional Profound or Otterly.ai subscription
      role: AI surface citation tracking

  no_dependencies_on:
    - Third party content delivery networks
    - Third party reverse proxy networks
    - Hosted CRM platforms (HubSpot, Salesforce, Pipedrive)
    - Hosted email marketing platforms (Mailchimp, ConvertKit, Klaviyo)

14.2 Listmonk Vhost Configuration

# /etc/nginx/sites-available/crm.[client-domain]
server {
    listen 443 ssl http2;
    server_name crm.[client-domain];

    ssl_certificate /etc/letsencrypt/live/crm.[client-domain]/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/crm.[client-domain]/privkey.pem;

    location ~ ^/(admin|api)/ {
        allow [office-ip-range];
        deny all;
        proxy_pass http://127.0.0.1:9000;
    }

    location / {
        proxy_pass http://127.0.0.1:9000;
        proxy_set_header Host $host;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }
}

Pair with an 80 to 443 redirect server block and a Let's Encrypt cert provisioned through certbot.

14.3 Gmail SMTP Relay Configuration

Listmonk and Mautic both support authenticated SMTP relay. Configure outbound mail through Gmail with an app password from the dedicated outreach mailbox.

smtp_configuration:
  host: smtp.gmail.com
  port: 587
  encryption: STARTTLS
  username: outreach@[client-domain]
  password: [app password from Google Account Security]
  send_rate: 100 emails per hour maximum (Gmail relay limit)

DNS records for the sending domain:

# SPF record
[client-domain] TXT "v=spf1 include:_spf.google.com ~all"

# DKIM record (key generated through Google Workspace Admin)
google._domainkey.[client-domain] TXT "v=DKIM1; k=rsa; p=[public key]"

# DMARC record
_dmarc.[client-domain] TXT "v=DMARC1; p=quarantine; rua=mailto:dmarc@[client-domain]; pct=100"

14.4 Analytics Vhost for UTM Tracking

Outreach campaigns tag every link with UTM parameters. Plausible Community Edition tracks clicks server side without third party trackers and without GDPR overhead.

UTM convention:

?utm_source=outreach&utm_medium=email&utm_campaign=[campaign-name]&utm_content=[pitch-variant]

The campaign name encodes the asset and the methodology. Example: research-2026-q2-broken-link for a broken link building campaign promoting the Q2 2026 research report.

14.5 Operational Audit Rubric

Score: 25. World class: 23 or higher out of 25.

End of Framework

Version 2.0 (2026-05-14)

Cross references:

• framework-digitalpr.md: the press driven subset of link acquisition.
• framework-internallinking.md: the internal counterpart distributing equity once external equity arrives.
• framework-localseo.md: NAP citation overlap and local link patterns.
• framework-aicitations.md: the AI surface where linked mentions ride alongside unlinked mentions.
• framework-trustsignals.md: the trust surface that combines linked and unlinked signals.
• framework-spampolicies.md: the explicit do not do list.
• framework-eeat.md: the authoritativeness pillar where domain reputation accrues.
• framework-knowledgegraph.md: the entity graph where sameAs networks reinforce link equity.
• framework-aioverviews.md: Google AI Overview citation behavior including linked mentions.
• framework-perplexityspaces.md: Perplexity citation behavior including linked mentions.
• framework-searchgpt.md: ChatGPT Search citation behavior including linked mentions.

This framework operates as one discipline in the larger SEO and AI search stack. Coordinate across the full stack: link building isolated from digital PR, internal linking, anchor discipline, and AI citation monitoring produces inferior outcomes.

Related references

• Engine optimization services
• ThatDeveloperGuy — SDVOSB web studio

Authored at ThatDevPro, the SEO and engine-optimization arm of ThatDeveloperGuy.