The SEO agent that shows its work.
Connect Search Console, ask in plain English, and watch it run audits, traces and lookups — every answer cites the data it used.
Replay of a real session: the user asks whether clicks dropped; the agent checks Search Console for free, reuses a cached domain overview at no cost, runs one fresh lookup, and answers with cited numbers — Did my clicks really drop this month? Check my Search Console, then compare my site against photoai.com and aragon.ai with domain overviews. What should I fix first?
Tracetify's SEO agent answers with numbers from your own Search Console and cites every source it uses — the replay above is an unedited session.
Your data, connected
Connect Search Console once, for free. The agent reads it as often as a run needs — querying your own data never costs a credit.
It decides what to research
Ask a question in plain English. The agent chooses the tools — your Search Console, a domain overview, a competitor lookup — and shows every step with its price before moving to the next.
Answers that cite their sources
No generic checklist. The answer quotes your own numbers, names the pages losing clicks, and points to the competitor evidence behind each recommendation.
No — your clicks did not drop. Search Console (28-day window, data through 2026-08-14) shows 56 clicks vs 28 in the prior period (+100%), impressions 16,432 vs 9,388, and average position improving from 38.8 to 28.3. Your real problem is conversion of that growth: CTR is 0.34%, which is roughly 10x below normal for the positions you hold. Fix in this order:
Billing you can see coming.
see the ceiling before you send
pay only for what a run uses
failed steps refund themselves
Every tier unlocks the agent.
- ✓every existing report, free to read
- ✓try a trace before signing up
- ✓no card, no password — just email
- ✓150 credits, never expire
- ✓5 full analyses (trace + timeline + verdict)
- ✓$3.98 per full analysis
- ✓works across every dashboard tool
- ✓usable from Claude Code, Cursor & Codex — via MCP
- ✓100 credits every month
- ✓unused credits roll over — they never expire
- ✓SEO agent — subscription only
- ✓unlimited traces, timeline & verdict unlocks on the site
- ✓fair use applies
Already living in Claude Code?
The MCP server is the better fit — you bring the model, we bring the data. Search and read reports free; traces and unlocks bill the same credits.
$ claude mcp add tracetify -e TRACETIFY_API_KEY=ttfy_... -- npx -y tracetify-mcpcreate an API key →Cursor & Codex config snippets live there too.
Questions, answered
Anything else — support@tracetify.com
How does the SEO agent charge?
Before you send a message, the composer shows the ceiling for that run — the most it can possibly cost — split into model use and paid research. You are charged for what the run actually uses, which is normally well under that ceiling. Querying your own connected data, such as Search Console, is free, and so is reusing research the agent has already paid for. If a paid step fails, its credits come back.
What happens to my Search Console data?
The connection is read-only. What we import stays private to your account — it powers your dashboard and the agent's answers, nothing else. Disconnect whenever you like and the imported data is deleted with it.
How do the subscription tiers differ?
Every tier unlocks the SEO agent — that is the one thing a subscription adds. The dashboard research tools run on credits, so any balance reaches them. The tier decides how many credits arrive each month, and unused credits roll over instead of expiring. The top tier adds unlimited traces, timeline unlocks and verdicts on the site, on top of its monthly credits, within the fair-use cap.
What does fair use mean on Unlimited?
Unlimited covers normal human use: up to 30 fresh traces per calendar day. Unlocks and verdicts are never capped, and reading existing reports doesn't count. The cap exists to keep scripted abuse from ruining it for everyone. Unlimited covers using the site; API and MCP runs always bill credits per call.
It checks your Search Console, runs the research, and answers with the numbers to prove it.