Backlink Profile Analysis

# Backlink Profile Analysis

## Loadout capability bindings

- Read the live Ahrefs tool contract first with <action-tag>mcp:ahrefs_mcp:doc</action-tag>.
- Establish the profile baseline with <action-tag>mcp:ahrefs_mcp:site_explorer_backlinks_stats</action-tag>.
- Analyze linking-site distribution with <action-tag>mcp:ahrefs_mcp:site_explorer_referring_domains</action-tag>.
- Analyze anchor concentration with <action-tag>mcp:ahrefs_mcp:site_explorer_anchors</action-tag>.
- Identify link magnets and reclamation targets with <action-tag>mcp:ahrefs_mcp:site_explorer_pages_by_backlinks</action-tag> and <action-tag>mcp:ahrefs_mcp:site_explorer_broken_backlinks</action-tag>.
- Review link-level evidence with <action-tag>mcp:ahrefs_mcp:site_explorer_all_backlinks</action-tag> only when aggregate data is insufficient.
- Analyze referring-domain change over time with <action-tag>mcp:ahrefs_mcp:site_explorer_refdomains_history</action-tag>.
- For domain-level Ahrefs calls, use `mode=subdomains` unless the user explicitly wants only the exact hostname. Confirm scope and any paid usage before execution.

## Goal

Produce an evidence-backed backlink profile analysis covering referring domains, anchor text, link quality, linkable pages, competitor gaps, and link trends. Separate observed facts from heuristics and never present a risk flag as proof of a search penalty.

## Required inputs

- Target domain or URL
- Analysis mode: full profile, competitor gap, suspicious-link review, reclamation, or new/lost trend
- Optional competitor domains
- Target market and language when geographic relevance matters
- Optional date range

## Source and scope checks

1. Confirm whether the target is a domain, subdomain, or exact URL.
2. Read the Ahrefs documentation Action for the exact endpoint schemas before executing any analysis Action.
3. Record the database, date range, target mode, limit, order, and filters used for every result set.
4. Prefer aggregate Actions first. Use link-level results only to investigate a specific pattern or produce a bounded evidence sample.
5. If Ahrefs is not connected or returns no data, state the limitation. Analyze a user-supplied export if available; do not invent metrics or silently substitute unrelated sources.

## Analysis framework

Produce the sections relevant to the requested mode. Do not manufacture a complete seven-part audit when the available data supports only a subset.

### 1. Profile overview

Report, when returned by the source:

- Total backlinks and referring domains
- Follow and nofollow distribution
- Domain or URL rating fields, using the provider's exact names
- Change over the selected period
- Concentration in the largest referring domains

Label every metric with its source and scope. Compare results with the site's age, category, and prior period; fixed universal thresholds are directional heuristics, not verdicts.

### 2. Anchor text distribution

Group anchors into branded, naked URL, generic, exact-match commercial, partial-match, long-tail or natural phrase, and image or empty anchor. Report the leading anchors and their share of the returned sample. Flag concentrated commercial anchors for manual review, but account for brand names that also contain target keywords.

### 3. Referring-domain quality

Evaluate topical and audience relevance, editorial context, distribution across independent domains, follow and nofollow mix, geographic fit, estimated authority fields returned by Ahrefs, and repeated sitewide patterns. Do not assume a TLD is high or low quality by itself. Treat authority metrics as vendor estimates, not Google metrics.

### 4. Suspicious-link review

Use bounded link-level samples to investigate:

- Repeated exact-match anchors across unrelated sites
- Large sitewide footer or sidebar placements
- Thin or machine-generated linking pages
- Networks with highly similar templates or ownership signals
- Irrelevant language or country clusters
- Sudden bursts of low-relevance referring domains
- Links from pages that no longer resolve or no longer contain the link

Classify each finding as observed, inferred, or unverified. Do not call a link toxic solely because of a low vendor metric. Recommend a disavow file only after manual review, documented evidence, and confirmation that removal outreach is impractical; never upload or submit one without explicit user authorization.

### 5. Top pages and link reclamation

Identify pages attracting the most backlinks, high-value linked URLs that now return an error or redirect poorly, lost or broken backlinks with a clear replacement destination, and successful asset formats that could be updated or replicated. For each reclamation opportunity, provide the source URL, linked target, observed status, proposed destination, and rationale.

### 6. Competitor gap

Run the same referring-domain query and scope for the target and each competitor. Separate domains linking to competitors but not the target, domains linking to both, and domains linking only to the target. Prioritize opportunities by relevance, editorial fit, evidence of linking to comparable assets, estimated authority, and a concrete outreach angle. Do not treat every competitor referring domain as a viable prospect.

### 7. Trend and change

Use referring-domain history for the requested period. Report the direction and magnitude of change, material spikes or drops, whether change is broad-based or concentrated, and plausible explanations that require verification. Avoid causal claims from correlation alone.

## Evidence and scoring rules

- Use a numeric health score only if the user explicitly asks for one and the required factors have adequate data.
- Show the formula, weights, data coverage, and assumptions next to any score.
- Never convert missing data into a zero.
- Prefer confidence labels such as high, medium, low, or insufficient evidence over false precision.
- Keep source metrics separate when scopes or provider definitions differ.

## Output format

Start with the scope and data source, coverage and limitations, three most important findings, and three highest-leverage next actions. Then provide:

| Finding | Evidence | Source and scope | Confidence | Recommended action |
| ------- | -------- | ---------------- | ---------- | ------------------ |

For competitor-gap or reclamation work, add:

| Prospect or source URL | Why it matters | Outreach or remediation angle | Priority |
| ---------------------- | -------------- | ------------------------------ | -------- |

End with the exact filters and date ranges used so the analysis can be reproduced.

## Pre-delivery review

- Every metric has a source, target scope, and date range where applicable.
- Counts in the summary match the returned result set or are clearly labeled as provider totals.
- No inferred relationship, penalty, or link removal is presented as fact.
- Direct competitors, paid placements, spam, and irrelevant pages are flagged.
- Contact details or link status claims include a public source URL.
- Recommendations distinguish removal, reclamation, content creation, and outreach.

## Upstream attribution

This is a Loadout-native adaptation of the upstream workflow because its Claude SEO CLI, local scripts, and DataForSEO tool contracts are not available as equivalent Loadout capabilities. See [upstream provenance and license](UPSTREAM.md), the [preserved upstream license](UPSTREAM_LICENSE.txt), and the [upstream package license notice](LICENSE.txt).