SEO & Marketing Framing Analyzer — deterministic, no-AI, evidence-backed

In plain words: every page on the internet is trying to move you — hurry, last chance, everyone's switching, experts agree. This tool reads pages and shows you exactly how they're doing it, quoting the very sentence each time. No AI and no guessing: if it can't point to the actual words on the page, it says nothing at all. Marketers use it to study copy at scale; anyone can use it to see the machinery plainly. The rest of this page is the same promise in engineering terms.

Extract on-page SEO signals, capture pages multimodally, and classify the marketing framing of web copy — with an engine that uses no LLM. That means: no model sits in the loop to drift, sample, or hallucinate — so there are no invented fields, every classification is backed by a verbatim evidence quote, and there's no per-result AI cost — so it stays fast and cheap at scale. (Two runs of the same URL can differ when the page changes or a dynamic page renders differently — that's the input moving, never the engine guessing.)

Three jobs in one Actor — read a page's search-engine signals and its persuasion moves, scrape pages cleanly, or tag pages against your own labels:

Why this Actor

• No model, no mood — there's no LLM in the loop, so nothing is sampled, invented, or drifting; results are driven only by what's actually on the page. The engine adds no variance of its own — what can change between two runs of the same URL is the live page itself (a content update, or a dynamic render), never the engine's judgment.
• No hallucinated data — a field is only emitted when it's actually present on the page, and every marketing-framing label carries the verbatim sentence that triggered it. Nothing is invented.
• Internally corroborated — a classification ships only after it is independently cross-checked inside the engine; a weak or conflicting signal is discarded, not reported. When results arrive, they've already survived their own audit. In plain words: it double-checks itself before it speaks, and throws away what it isn't sure of.
• Your data never meets an AI provider — the engine is LLM-free, so page content is processed entirely in-container and never sent to any third-party model. Privacy by construction, not policy.
• No per-result AI cost — LLM-free also means runs stay cheap and fast even across thousands of pages. That's what keeps pay-per-event pricing low.
• Provenance on every row — each result carries a tamper-evident provenance tag; with a secret set, tags are HMAC-signed so you can prove to a client which rows came from your audit, unaltered.
• Bring a proprietary lens, keep it proprietary — ship your own classification framework encrypted (AES-256-GCM); it decrypts only in memory at runtime. Your methodology never sits in the clear. In plain words: your method stays your secret — even from us.

Input

What you tell it: which pages to read, and which of the three jobs to do:

{
  "mode": "seo",                                  // "seo" | "scrape" | "code"
  "startUrls": [{ "url": "https://example.com" }],
  "multimodal": false,                            // render + transcribe on-screen text
  "lensName": "marketing_seo_v1",                 // built-in framing lens (seo/code modes)
  "maxItems": 50,
  "timeoutSecs": 30
}

For the code mode you can pass your own lens instead:

{
  "mode": "code",
  "startUrls": [{ "url": "https://example.com/pricing" }],
  "lensJson": {
    "matching_modes": {
      "pricing_signals": {
        "free_tier":  { "lexical_markers": ["free plan", "free tier", "no credit card"] },
        "enterprise": { "lexical_markers": ["contact sales", "custom pricing", "enterprise"] }
      }
    }
  }
}

Output

What comes back: everything found on each page, with every claim carrying the exact words that earned it.

seo mode — one item per page:

{
  "url": "https://example.com",
  "title": "Example — Get Started Today",
  "title_length": 30,
  "meta_description": "Try it free, cancel anytime.",
  "canonical": "https://example.com/",
  "og": { "title": "Example", "description": "...", "image": "https://.../og.png", "type": "website" },
  "twitter": { "card": "summary_large_image", "title": "Example" },
  "h1_count": 1,
  "word_count": 812,
  "images_total": 14,
  "images_with_alt": 11,
  "images_alt_coverage": 0.786,
  "links_internal": 42,
  "links_external": 9,
  "top_keywords": [{ "term": "pricing", "count": 12 }, { "term": "free", "count": 8 }],
  "marketing_framing": [
    { "dimension": "persuasion_framing", "label": "risk_reversal",
      "evidence_quote": "Try it free, cancel anytime.", "review_needed": false },
    { "dimension": "persuasion_framing", "label": "urgency",
      "evidence_quote": "Get started today.", "review_needed": false }
  ],
  "marketing_framing_counts": { "risk_reversal": 1, "urgency": 1 },
  "provenance": "dove0:9f2c…"
}

code mode — one item per classification, each with its evidence:

{ "dimension": "pricing_signals", "label": "free_tier",
  "evidence_quote": "Start on the free plan, no credit card required.",
  "source_url": "https://example.com/pricing", "source_title": "Pricing",
  "review_needed": false, "provenance": "dove0:1a7b…" }

Great for

• SEO audits at scale — meta/heading/canonical/OG hygiene across a whole site or competitor set.
• Marketing & competitive intel — see how competitors persuade (urgency vs. social proof vs. authority), backed by the exact copy.
• Content & CRO teams — quantify persuasion framing on landing pages, reproducibly.
• Custom tagging pipelines — the code mode is a deterministic, evidence-backed classifier for any label set you define.

Deterministic vs. an AI analyzer — a structural difference, not a slogan

Most tools that classify web copy at scale run on an LLM. That buys fluency — and it gives up three guarantees you can't get back, not because any vendor is careless, but because a probabilistic model structurally can't make them:

• Reproducibility. Ask a model the same thing twice and the answer can drift. Ask this engine, and the same page yields the same result — no sampling, no temperature, nothing to drift.
• Evidence. A model can return a confident label with nothing on the page behind it, and can't prove it didn't. Here, every label carries the verbatim sentence that triggered it, or it isn't emitted — a guess has no path onto your bill.
• Privacy. A model has to send your content off to judge it. This engine is LLM-free, so your pages never leave the container and never reach any AI provider.

None of that calls an AI tool dishonest. It's the difference between what a probabilistic method can offer and what a deterministic one can guarantee — and we'd rather you check it than take our word. Run the same audit twice and diff the output. Trace any label to its quoted sentence. Watch the network and see nothing leave. The difference isn't a claim you have to trust; it's one you can verify from where you sit.

Pricing

You pay only for what verifiably happened, and the bill is one number you can work out before you run: a small fixed run fee plus a flat per-page price. Two meters:

That's the whole meter — no per-result charge you can't predict. Your total is

run-started + (pages × page-processed)

What a real run costs

Concrete, and knowable before you press run:

• Quick 3-page check — 1 run + 3 pages ≈ $0.02
• 100-page SEO audit — 1 run + 100 pages ≈ $0.70
• 1,000-page audit — 1 run + 1,000 pages ≈ $7.00

A page the engine abstains on — nothing it can quote — costs exactly the same as any other: you're paying for the read, not for findings, so a guess can never appear on your bill and ambiguous copy never inflates it.

How this price was derived

Priced by derivation, not by market — no reference to anyone else's prices. The formula: measured cost ÷ platform share + a stated stewardship wage. From a real measured run (2026-07-22): 2 pages + 4 classifications cost $0.0078 total — a marginal page cost of roughly $0.002–0.003. After the platform's ~20% share, cost-recovery is about $0.003 a page; the rest of the $0.007 is an openly-stated stewardship wage for building and keeping an engine that reads meaning without guessing. Classifications aren't metered separately — they cost ~nothing to emit, so the work, and the price, is the page. The launch is priced once as itself, so a 3-page check and a 3,000-page audit each pay the true shape of their own cost — no cross-subsidy. And one thing you are never charged for: a guess — the engine abstains on ambiguous copy rather than inventing a classification to bill you for. When the meters change, the price is re-derived and this section updated.

The honesty contract (what it will and won't do)

What you can rely on: every classification is evidence-quoted, internally corroborated, and free of model drift — no LLM sits in the loop to hallucinate or change its mind between runs, so the analysis is driven entirely by the page in front of it. When a page hasn't changed, a re-run reproduces it; when it has, what you see is the page's change, never the engine's mood. On ambiguous phrasing it abstains rather than guesses — so when it does speak, the label has already survived cross-examination. Precision over recall is a capability, not an apology: it is the one guarantee AI-based analyzers structurally cannot make.

The true limits, stated plainly:

• It classifies by what is actually on the page. Framing that lives purely in tone, imagery, or implication — with no words to quote — is reported as nothing, on purpose. It stays silent rather than guess.
• multimodal mode renders each page (slower); reserve it for JavaScript-heavy pages or text baked into images — where it will read what non-rendering scrapers can't see at all.
• Respect each target site's terms of service and robots policy when scraping.

Servicing of terms

We flipped the label on purpose: not terms that govern the service — a service that keeps its terms. Here they are, short enough to actually read: your data stays yours (processed in-container, never retained after the run, never sold, never trained on, never shown to any AI provider); no rights are claimed over your inputs or outputs beyond mechanically running the job you asked for; you pay only for receipted events — never for a guess; you can leave anytime with nothing held. We ask one term back: don't point this tool at taking — respect the grounds you aim it at.

Support

Questions or a custom lens for your use case? Open an issue on the Actor's Issues tab.