North Signal

One loop: measure, fix, prove.

Eight capabilities, one workflow. Everything below is included in Pro — $9.99/month.

Answer Insights

See when AI mentions you, when it actually recommends you, and which competitors it names instead — across ChatGPT, Google AI Overviews, Perplexity, and Claude, broken down by the kind of question asked.

How it works

We build a set of questions your real buyers ask (branded, category, comparison, problem), run each one many times per engine, and read every answer for mentions, recommendations, citations, and sentiment.

Why it's different

AI gives different answers each time you ask. Every number we show carries a confidence range and a run count, so you always know what's signal and what's noise.

Demand Intelligence

Know which questions are worth winning. Each prompt gets a demand tier — high, medium, low — so you spend effort where buyers actually are.

How it works

Tiers come from public signals (Google Autocomplete, community discussion frequency, and real search volumes when connected) and from your own Search Console queries and sales-call questions, imported in one paste.

Why it's different

We never invent a fake 'AI prompt volume' number. Tiers state their derivation, and your own funnel data — the one perfectly unbiased sample — gets first-class treatment.

Experiments

The feature the monitors don't have: proof. Ship a fix and find out whether it actually moved your visibility — or did nothing.

How it works

Pick a fix from your prescription board. We split the matching prompts into a treatment group and a held-out control, re-measure after you ship, and run a significance test.

Why it's different

Verdicts are honest: worked, noise, or kill. When we say a fix worked, it's because it beat the control — not because a dashboard line wiggled.

Monitoring & Alerts

Your visibility, re-checked daily or weekly, with alerts when something real changes — a drop after a model update, a surge after a placement lands.

How it works

Scheduled re-measurement with the same rigor as your baseline, compared batch over batch.

Why it's different

Alerts fire only when a move exceeds the statistical noise band. No 3am panic over sampling variance.

Page Audits

Point at any URL and get a pass/fail checklist of the on-page factors proven to affect AI visibility — plus whether AI crawlers can even read your site.

How it works

We fetch the page and score it: does it open with a direct answer? concrete statistics? credible citations? a visible date? hedging? Then we check llms.txt and robots.txt for AI-crawler access.

Why it's different

Each check carries the strength of evidence behind it — and we tell you which popular 'optimizations' (schema stuffing, visual formatting) measurably do nothing.

Agent Traffic

See which AI crawlers — GPTBot, ClaudeBot, PerplexityBot, Google-Extended — are actually reading your site, and what they read.

How it works

Paste your server access logs; we detect every known AI crawler and summarize requests and top pages.

Why it's different

Parsing runs entirely in your browser. Your logs never leave your machine.

Rewrite Assistant

Turn a weak paragraph into the passage an AI answer-writer wants to quote.

How it works

Paste one passage; it comes back front-loaded, specific, de-hedged, and citation-ready — with a note on exactly which levers were applied and why.

Why it's different

One passage at a time, deliberately. This is a scalpel for your money pages, not a content farm.

Client Reports

A print-ready report — diagnosis, actions, proof of lift — branded with your agency's name and colors.

How it works

One click from any project; your byline and accent color come from settings. Print → Save as PDF is the deliverable.

Why it's different

The methodology footnote is built in, so your clients can trust the numbers the same way you do.