Do I need to open the dashboard to understand a traffic drop?

No. For recurring questions — why traffic moved, weekly summaries, install checks — MCP from Cursor or Claude is enough. The dashboard remains useful for ad-hoc exploration.

Which MCP tools help explain traffic changes?

compare_periods for week-over-week KPIs, get_top_referrers and get_top_pages for where the change concentrated, get_utm_breakdown for campaigns, and generate_report for a shareable summary.

Ask Cursor Why Traffic Moved — Numbers Behind the Answer

Q: Can Cursor query my Antlytics data?

Yes. Connect the Antlytics MCP server at https://www.antlytics.com/mcp, then ask in plain language. Tools like compare_periods, get_stats, and get_top_referrers return structured numbers the agent can cite.

Disclosure: We build Antlytics. This walkthrough shows a workflow that ships today via our MCP server — not a future feature demo.

You ship a change. Traffic dips. The old ritual is: open the analytics tab, hunt the right date range, export a CSV, guess.

If you build with AI, there is a better loop: ask Cursor or Claude why, and require the answer to cite real numbers from your analytics — not vibes.

The workflow in four steps

1. Connect MCP once

Add the Antlytics MCP server to Cursor (or Claude Desktop / Claude Code). Auth is OAuth 2.1 — browser sign-in, no API key paste.

Config and client snippets live in the MCP docs. Paid plans (including the 14-day trial) unlock MCP.

2. Ask in plain language

In the editor:

Why did traffic drop last week on myapp.com?

You are not asking the model to invent a story. You are asking it to call tools that read your account.

3. The agent uses shipped tools

A typical investigation uses tools that already exist on the Antlytics MCP server:

Question	Tool
Visitors / pageviews this week vs last	`compare_periods`
Overall KPIs for a range	`get_stats`
Which sources fell	`get_top_referrers`
Which pages fell	`get_top_pages`
Campaign attribution	`get_utm_breakdown`
Shareable summary	`generate_report`

Full list: MCP tool reference.

4. Answers come with numbers attached

What you want back looks like this — illustrative, but the shape matches real tool output:

You: Why did signups drop last week?
Cursor: Signups down 18% vs the prior week. Top drop: /pricing (−34% views). Referrer chatgpt.com fell from #2 to #8. Overall visitors −12%; bounce rate unchanged.

The agent is useful when it cites compare_periods / referrer / page breakdowns. If it answers without numbers, ask it to pull the tools again.

What this does not replace

Ad-hoc dashboard exploration — filtering weird date ranges or scanning a chart is still faster in the UI sometimes.
Product analytics — funnels inside your app, session replay, feature flags. That is PostHog territory; Antlytics is web traffic analytics.
A magic “explain” button with a special name — the product surface today is MCP tools returning structured stats. Use them.

Install path if you are new

Create an account — trial, no card required
npx @antlytics/init in your project (quick install)
Connect MCP (setup guide)
Ask: “Compare last 7 days to the previous 7 for my site”

FAQ

Can Cursor query my Antlytics data? Yes — via the MCP server. See docs/mcp.

Do I need the dashboard open? Not for recurring “why did traffic move?” questions. Keep the dashboard for exploration; use the agent for answers with numbers attached.

Which tools explain a drop? Start with compare_periods, then get_top_referrers and get_top_pages. Finish with generate_report if you want something to paste into Slack or a PR.