Source influence scoring
Weighted by frequency, prominence, and brand co-occurrence.
Last updated: 2026-05-05
Source influence scoring
Not every citation is worth the same. A mention on a domain that 5 different AI models cite repeatedly is far more valuable than a single citation on an obscure blog. The source influence score quantifies this so you can prioritise outreach and link-building.
How influence is calculated
For each domain in scope, Pranas computes:
influence = log(citation_count) × model_breadth × brand_diversity × recency_weight
- citation_count — total times the domain was cited (log-scaled to prevent one viral page from dominating).
- model_breadth — number of distinct AI models that cite the domain (1–5+ scale). Cross-model citation is the strongest reliability signal.
- brand_diversity — number of distinct brands the domain cites. A domain that cites only one brand is a brand-owned property; a domain that cites many is editorial / aggregator territory (more valuable for new entrants).
- recency_weight — recent citations weight higher than old ones; a domain that stopped being cited 60 days ago is decaying.
The output is normalised to a 0–100 score where 70+ means "high-priority outreach target".
Score band reference
| Influence score | Meaning | Recommended action |
|---|---|---|
| 70–100 | High-priority outreach target | Pitch, contribute, or claim a profile now |
| 40–69 | Moderate influence | Worth pursuing after the top tier |
| 10–39 | Low influence | Only if the effort is trivial |
| Below 10 | Negligible | Ignore |
Factor reference
| Factor | What it captures | Why it's weighted this way |
|---|---|---|
| citation_count (log-scaled) | Raw citation volume | Stops one viral page dominating |
| model_breadth | Distinct models citing the domain | Cross-model citation is the strongest reliability signal |
| brand_diversity | Distinct brands the domain cites | Editorial domains beat brand-owned properties |
| recency_weight | How recent the citations are | Decaying domains shouldn't look healthy |
What high-influence domains usually look like
- Industry comparison sites (G2, Capterra, TrustRadius)
- Major editorial sites in your vertical (TechCrunch for SaaS, Healthline for healthcare, etc.)
- Reddit and Stack Overflow threads (heavy citation by Perplexity especially)
- Wikipedia and structured-data sites
- Your own domain (if you're well-cited, your own pages will appear here — useful sanity check)
Acting on the score
Sort the source table by influence descending and:
- Identify the top 20 domains where you have zero or low citation count.
- For each, click through to see which competitors are cited there and what context.
- Build an outreach list: contributed posts, expert quotes, listicle inclusion, review-site profiles.
The Action Center surfaces high-influence "missing citations" as actionable recommendations automatically.
Caveats
- Self-citations don't count. Mentions on your own domain are excluded from the score.
- Domain consolidation. Subdomains (
docs.example.com,blog.example.com) are rolled up to root unless they're known editorial subdomains. - Score decays. A domain that was high-influence last quarter but hasn't been cited recently will drop. Re-run scans regularly to keep the picture fresh.
Related articles
- Sources overview
- Citation optimization tactics
- Action Center