Your roster’s organic, local, and AI-search numbers stop living in six tabs and a spreadsheet. They land in one role-secured portal, refresh automatically each month, and come with the next move attached: a transparent A–F grade, the content plan with exact internal links, and a priced proposal you can send before the call ends. Pulse is built for marketing agencies. For the fundamentals, see what a marketing dashboard should include.
Pulse is a multi-client command center for SEO and digital-marketing agencies: one typed view per client, assembled from your connected accounts, scored on a published rubric, and paired with the work to move the score.
Organic SEO (Search Console), web analytics (GA4), authority and backlinks (Ahrefs), site speed (PageSpeed Insights), and local presence (Google Business Profile) — plus AI-search visibility as its own pillar. Five collectors run live; three more sources (Google Places, Anthropic, prospect-site scrape) are wired for the proposal engine.
It is not paid-ad-account reporting. "Paid ads" here means competitor ad intelligence and a priced proposal, not connected Google/Meta Ads. It is not a social publishing suite. "Social" here is a link-building property tracker, not scheduling or an inbox. White-label client reporting is on the roadmap, not shipped today.
Most tools stop at charts. Command Center carries the plan and the playbooks: AI content plans with exact internal links, a 33-tactic foundation checklist, an SEO pipeline board with enforced stages, a 20-category launch-readiness gate, and a 10-screen migration suite.
Proof · AI content plan, SEO pipeline board, 33-tactic foundation checklist, 20-category launch gate, 10-screen migration suite — all shipped.
The Chain Reaction scores four funnel stages — Attract → Impress → Convert → Compound — on a published rubric with good/ok/bad bands and visible weights. The Report Card rolls eight weighted pillars into one A–F grade. Both show the literal math, so the number survives a client’s "why?"
Proof · Verified weights: Organic 0.18 · Content 0.15 · Local 0.13 · Interlinking 0.12 · Authority 0.12 · Technical 0.10 · Speed 0.10 · AI Visibility 0.10. Bands: A≥90, B≥80, C≥70, D≥60. Unmeasured pillars excluded, never guessed.
Four AI capabilities run on Claude (claude-sonnet-4-5). Every one recommends or drafts — none auto-executes. Prompts forbid inventing numbers, answers are caged to the data provided, and the AI Analyst reads a live scrape of the client’s actual website.
Proof · AI Analyst, AI Content Plan, Ask-AI/Meetings, AI Proposal generation — all shipped on Claude, all recommend-only, all cached to avoid re-spend.
Postgres Row-Level Security scopes every client-data table. A client reads only the clients they’re a member of. Agency tools — pipeline, proposals, portfolio — are hidden in the nav and redirected at the route.
Proof · RLS policies on every client-data table; analyst vs client roles enforced at the database and route level.
ChatGPT, Gemini, and Perplexity presence, citations, and share-of-voice sit beside classic SEO — and AI Visibility is one of the eight Report Card pillars, not a bolt-on tab.
Proof · The AI-visibility screen ships today; the live collector (Brand Radar) is gated off by default and renders on seeded data until switched on per account.
Paste a prospect’s domain. Command Center pulls public data, scores it on the same Report Card and Chain, ranks "missing opportunity" cards mapped to priced services, and writes an AI cover letter. Curate, then share a public proposal page or save to PDF.
Proof · Proposal generator, priced service catalog, public proposal pages — all shipped (needs Supabase + keys; proposals DB migration not yet applied on the live database).
A Vercel cron pulls every connected source monthly and writes append-only snapshots, so history accrues for free. Run an on-demand benchmark any time and compare any two dated snapshots with a delta table.
Proof · Monthly cron, append-only snapshot history, benchmark + compare — shipped. Five collectors live today (PSI, GSC, GA4, Ahrefs, GBP).
GSC clicks, GA4 sessions, and Ahrefs estimates sit side by side, so the "why don’t these match?" conversation has an answer on screen before the client asks it.
Proof · Reconciliation view — shipped.
One agency-wide grid: clicks, position, Domain Rating, health grade, AI presence, weakest link, opportunities, and concerns per client. Spot the account slipping before the client notices.
Proof · Portfolio screen and API — shipped, analyst-only.
A 10-screen migration suite tracks equity retained, redirect coverage, a 301 map with CSV export, keyword retention, and legacy-vs-destination GSC/GA4.
Proof · 10-screen migration suite — UI shipped; runs on seeded data pending a live migration collector.
Read this table honestly. Where we concede: connected ad-account reporting, social scheduling, and white-label dashboards — we don’t do those yet. Where we win: the work to move the number, scoring a client can trust, AI caged to real data, and a built-in sales engine.
All run on Claude (claude-sonnet-4-5). Every output is advice or a draft. The rule across all four: human-in-the-loop, caged to your real data, never invents numbers. Nothing publishes, edits a live site, or spends ad budget.
Problem: Raw metrics don’t tell you what to do.
How: Claude reads the client’s real signals plus a live scrape of their own website, returns a 3–5 sentence executive read, tier-1/2/3 actions, and a tailored site-architecture map.
Controls: analyst-only to generate; read-only advice; cached per client.
Problem: Content calendars are generic.
How: Topic clusters and per-post briefs built from the client’s actual queries, pages, and products — each brief with exact anchor → target internal links and a phased roadmap.
Controls: analyst generates; recommend-only; cached, regenerated on demand.
Problem: Prepping a client call is slow.
How: Page-aware chat that answers strictly from the client’s provided data and formats answers as presentation-ready Markdown tables.
Controls: clients can ask but not save; the prompt forbids inventing numbers; answers caged to provided data.
Problem: Proposals take hours.
How: Assembles a project from a prospect’s public data, runs Report Card + Chain + AI Analyst, ranks priced opportunity cards, and writes a cover letter.
Controls: the analyst curates (keep/drop cards) before anything is sent.
Next.js on Vercel, Supabase (Postgres) as the source of truth, append-only snapshots for history. No SEO API is ever called from the browser. All third-party keys are server-only env vars; the service-role key never reaches the client.
Postgres Row-Level Security on every client-data table. Analyst-only tables (pipeline, proposals) are gated to the analyst role. Snapshot writes require a secret token; cron requires a secret; mutations require the analyst role.
Not connected — stated plainly: Google Ads, Meta/LinkedIn Ads, any social publishing API, CRM/HubSpot, Stripe/billing, DataForSEO.
No SaaS pricing exists yet — the priced service catalog in the product is what an agency charges its own clients, not this product’s price.
Live collectors (GSC, GA4, Ahrefs, PSI, GBP), transparent Report Card + Chain. AI + proposal engine + suites: TBD.
Talk to usEverything in Starter + all four AI capabilities, the prospect → proposal engine, and migration + launch + pipeline suites.
Talk to usUnlimited clients (to confirm). White-label client reporting is roadmap across all tiers.
Talk to usData costs: Ahrefs usage is metered and cost-capped per run.
Book a walkthrough and see your whole client roster — graded, explained, and ready to act on — in one command center.
Tell us what you’re trying to do and we’ll show you the closest platform live. If it fits, we tailor it to you. If it doesn’t, we’ll say so. Replies within one business day.
Your message is on its way. A real person from the team replies within one business day.