Myth 1Evidence: Effect ≈ zero
Add schema to force citations (folklore)
Measured effect on AI citations sits around zero (Ahrefs, 2026). We grade the claim instead of selling a schema audit.
How Seeklyn works
See what AI says about you.Fix the passage. Prove it moved.
See how AI engines cite you, fix the passage that loses the citation, and re-check it, with evidence behind every number.Seeklyn tracks how AI answer engines mention and cite your brand, shows the exact passage that loses you the citation, rewrites it without inventing facts, and re-checks the result, with every number traced to its evidence.
Free, no signup, no card · Pro $24/mo when you want the weekly loop
Illustrative example.
The dashboard
Your dashboard, explained. Click anything.
The sidebar follows the job you’re doing: Observe what AI said, Measure how often, Compete for the sources, Diagnose what to change, and see the Impact. No proprietary 0–100 AI score: every number is a raw count you can open.
“Where do I stand, and what should I do next?”
Overview in the sidebar: Command center
Overview · 1 of 5
KPI band
Mention rate 38%: you appear in 91 of 240 AI answers, up 6 points on the previous 30 days. Raw count first, percentage second. Tap ⓘ for the formula.
Tap a number on the screen, or step through:
Illustrative example. The demo project tracks 12 prompts (a Pro-sized project).
Overview, part by part
Select a number on the screen, or a line below.
The evidence chain
Every number opens the sentence behind it.
Click a metric and you land on the prompt, the actual AI answer with your brand highlighted, and the URL it cited. Every number links to its evidence.
Follow one number to its evidence
Illustrative example.
No proprietary 0–100 AI score. Raw counts, named formulas, real sources.
Every number opens the prompt, then the actual AI answer, then the cited URL, tracked live across ChatGPT, Google AI Overviews, Perplexity, Gemini and more.
The loop
From weak passage to verified citation, in six steps.
The dashboard tells you where you stand. The loop is how you move it. Free runs the analysis; Pro runs the loop every week.
Illustrative example.
Step 01: Score
FreeTwo 0–100 scores from auditable signals: CTR for whether your title earns the click, AEO for passage-level citability. Every analysis ends in a funnel: Retrieval → Citation → Overall.
title: six CTR sub-scores · passages: chunked 100–300 words, sweet spot 134–167
Illustrative example.
Step 02: Diagnose
FreeEvery passage is linted against eight extractability flags, each graded by research strength, and every sentence gets a verdict: liftable, hedged, or orphaned.
website mode also checks robots.txt, noindex/nosnippet and blocked AI crawlers
Illustrative example.
Step 03: Fix
Free + ProEach flag ships a deterministic structural fix that never invents facts. A fix must clear its flag and must not lower any comparable score, or we don’t call it a fix. Pro deep mode adds a fact-locked LLM rewrite that must pass a faithfulness check.
citability 28 → 86 · verified by an immediate BM25 re-run
Illustrative example.
Step 04: Verify
Free & ProAm I Cited? shows the actual AI answers for your query across ChatGPT, Google AI Overviews, Perplexity, Gemini and more, with each citation in place. Verdicts are URL-precise: your exact page vs your domain via a different page.
Free checks are limited daily · Pro 100 checks/month · cited, mentioned or not cited, with source position
Illustrative example.
Step 05: Track
ProPro saves up to 25 projects and re-runs score, retrieval and citations weekly, so a lost citation arrives as an alert while it’s still fixable.
weekly re-check · decay alert email · competitive diff shows the winning passage
Illustrative example.
Step 06: Prove
Free & ProResults leave the app: a public share link with preview image, an SVG badge, a white-label PDF (Pro), a REST API with per-key quotas, an MCP server, and a GitHub Action that lints content PRs.
the PDF carries your client’s branding, not ours
Illustrative example.
Under the hood
How the scoring actually works.
Named mechanisms, not a secret algorithm: chunking, BM25 retrieval, eight extractability flags and query fan-out. The pipeline is deterministic: the same input always gives the same score.
Why a passage gets skipped
Engines skip a passage for detectable reasons. Seeklyn lints every chunk against eight flags, each rated by the strength of the research behind it.
- Strong: replicated research behind the rule
- Moderate: consistent evidence, fewer studies
- Weak: plausible, thinly measured
- Opens with "it / this / these" (tripped by the example passage)strong
- No answer in the first sentence (tripped by the example passage)strong
- Hedged claims (tripped by the example passage)moderate
- Run-on sentencesmoderate
- Low fact density (tripped by the example passage)moderate
- No attributionmoderate
- Entity-sparsemoderate
- Keyword stuffingweak
Illustrative example.
Which passages surface at all
Seeklyn chunks your page into 100–300-word passages and runs BM25, the lexical retrieval family production search is built on, over your own document for your target query. Passages that never surface can’t be cited, however well written.
- Top-k passages are the ones an engine can quote.
- Below the cutoff is invisible, whatever the prose.
Illustrative example. Scores illustrative.
The questions behind the question
One prompt silently fans out into roughly ten sub-queries (Seer Interactive, 2026), and about 95% of them have zero search volume. Seeklyn expands your query into eight angles and grades each covered, weak or missing.
- Covered: a passage answers this angle
- Weak: mentioned, not answered
- Missing: nothing to retrieve
- DefinitionWhat is a startup CRM?covered
- HowHow do startups set up a CRM?covered
- WhyWhy do startups need a CRM?weak
- ComparisonFree vs paid CRM for startupsweak
- CostWhat does a startup CRM cost?covered
- BestBest CRM for a 5-person teammissing
- WorthIs a CRM worth it before product-market fit?weak
- TemporalBest startup CRM in 2026missing
3 covered · 3 weak · 2 missingwrite the missing angles first
Illustrative example.
Honest by design
What we won’t tell you.
Every Seeklyn rule shows its evidence grade, including the inconvenient ones. Here is the popular advice we grade as folklore.
- Myth 2Evidence: Not read
Publish an llms.txt (folklore)
No major answer engine reads it. Seeklyn shows it as info only.
- Myth 3Evidence: Weak
Hit a keyword density (folklore)
Retrieval moved on years ago.
We can’t guarantee you’ll be cited. Nobody can, and tools that promise it are selling certainty they don’t have. We measure citability against public rules, verify citations across engines, and label every inference as an inference.
Where Seeklyn sits
AI visibility trackers
e.g. Otterly, Peec, Profound
Tell you whether you’re cited, at domain or prompt level.
- No passage-level diagnosis
- No fix, no before/after proof
~$29–$499+/mo
Content optimizers
e.g. Frase, Surfer, Clearscope
Help you write at keyword level, with no retrieval proof.
- No AI citation tracking
- No retrieval proof for a rewrite
~$15–$129+/mo
Seeklyn
Track + fix + proveTracker + optimizer, one loop
Closes the loop from “am I cited?” to “here is the fix, and proof it worked”.
Free · $24/mo Pro
Competitor pricing approximate, from public pages as of September 2026. Verify on each vendor’s site.
Free vs Pro
Free runs the analysis. Pro runs the loop.
Start with the full analysis at no cost. Upgrade when you want Seeklyn to keep watching, rewriting and proving.
Annual: 2 months free
Free
Analyze before you commit.
$0forever
- Headline + content analysis (all scores, flags, funnel)
- AI visibility dashboard: 3 tracked prompts
- “Am I Cited?” checks: 3/day without an account, 10/month with a free account
- 1 saved project
- A populated dashboard from day one
- No signup, no card
Pro
Everything includedRun the full loop, every week.
$24 per month
- Everything in Free
- 25 tracked prompts, re-run weekly
- 100 “Am I Cited?” checks/month
- 25 saved projects
- Weekly automated tracking + decay alerts
- Query fan-out + coverage matrix
- Fact-locked passage rewrites (deep mode)
- Competitive citation diff
- White-label PDF reports
30-day money-back guarantee, one email, no questions. Cancel anytime.
Run the full analysis free first. If the flags match what you suspected, Pro is a straight upgrade; if they don’t, you’ve lost nothing.
Agencies and teams: Enterprise
Seeklyn chunks your content into 100–300-word passages, runs BM25 retrieval over them for your target query, then lints each passage against eight extractability flags — answer-first structure, self-containment, hedging, attribution, fact density and more. The result is a 0–100 citability score per passage with the failing flags listed. The pipeline is deterministic: the same input always produces the same score.
Because the measurements don’t back the claims: schema markup’s measured effect on AI citations sits around zero (Ahrefs, 2026), no major answer engine reads llms.txt, and keyword-density targets predate modern retrieval. Every Seeklyn rule carries an evidence grade — strong, moderate, or weak — so you can see which advice is research and which is folklore before you spend time on it.
No — nobody can, and tools that promise it are selling certainty they don’t have. Seeklyn measures citability against transparent rules and shows you the actual citations across engines, so you can see what changed instead of trusting a promise.
Trackers report whether you’re cited at the domain or prompt level. Seeklyn also tracks citations, but then explains failures at the passage level, hands you a fix, and proves the fix with an immediate retrieval re-run — the diagnose-and-repair half that trackers leave to you. It’s also priced below every tracker entry plan we’ve seen.
When you ask an AI engine one question, it silently expands it into many sub-queries — roughly ten on average, per Seer Interactive’s 2026 research — and about 95% of those have zero search volume. Seeklyn expands your target query into eight fan-out angles and grades your coverage of each, because a page that only answers the visible query misses most of the retrieval surface.
Seeklyn shows you the actual AI answers and the actual citations, tracked in real time across ChatGPT, Google AI Overviews, Perplexity, Gemini and more. For each engine you see whether your page was cited, your brand was mentioned, or neither, plus the source position, and the verdict is URL-precise: your exact page vs your domain via a different page. The engine list grows over time.
For the same reason you optimized for organic before you ranked: citability is a property of the content, not of your current traffic. The passages you fix today are the ones retrieval finds tomorrow — and the free analysis tells you in about a minute whether you have a problem worth fixing.
Most content tools operate at the keyword level; retrieval operates at the passage level. Seeklyn works where retrieval works: it simulates how your passages surface with BM25, names the specific sentence-level failures, and re-runs the simulation after each fix so you can see the change instead of trusting a dial.