Playbook · Analytics

Run your monthly marketing report from Claude Code.

You're sending emails, posting content, and driving link clicks — but do you know what's actually moving the needle? Once a month, paste one prompt into Claude Code and your AI agent pulls every number from Loopi and writes a plain-English report: what worked, what flopped, and three things to do next month. No spreadsheets, no tab-switching, no dashboard to learn.

What it does for you

Right now, knowing whether last month was good means opening three separate dashboards, exporting numbers, and doing the math yourself. This play collapses that into one prompt. By the end you will know:

• Which emails your audience actually read. Open rates and click rates per broadcast — best-performing subject line named, worst performer with a hypothesis for why it missed.
• What your link page is driving. Total clicks on your bio page and your top three links by count — so you can see which destinations people actually want, and which are dead weight.
• Which social posts pulled their weight. Top post per platform with the engagement number that proves it. Any platform where you went dark last month — flagged.
• Two or three things that worked, one or two that flopped — written as specific observations with numbers, not generic advice.
• Three concrete actions for next month — each tied directly to the data, not boilerplate marketing guidance.

Three steps, start to finish

Everything the play needs. No website to build, no list to import — just two one-time connections and a prompt you run at the start of each month.

1. Get a Loopi account

The prerequisite is a free Loopi account — 60 seconds, email and password. Your mail, links, and post analytics all live there. If you have channels already connected (TikTok, Instagram, YouTube), the report will cover them automatically. If not, you can still run the play — it will report on whichever channels are active.

2. Connect Loopi to Claude Code

Loopi is an MCP server — one line wires it to Claude Code. (Claude Desktop, Cursor, Codex, and Gemini take the same https://api.loopi.social/mcp URL in their connector settings.) Authorization is a one-time browser login with a scoped key. After this, every monthly run needs no setup at all.

# Claude Code — add Loopi in one line
claude mcp add --transport http loopi https://api.loopi.social/mcp

# then just tell your agent:
> connect to loopi mcp        # opens a browser to authorize, once

3. Paste the prompt at the start of each month

This is the whole play. Paste it into Claude Code — the agent finds your profiles, determines last month's date range, pulls every channel, and writes report.md to your working directory. No IDs, no further input, no back-and-forth.

Connect to Loopi MCP and run my monthly marketing report for last month.

Steps:
1. Call profile.listProfiles to get all my profiles.
2. Determine the exact date range for last month (first to last day of the
   preceding calendar month, UTC).
3. For each profile, gather analytics:
   - mail.listBroadcasts (status=sent) to find last month's broadcasts, then
     mail.getMailContentAnalytics per contentId (with sentAtFrom/sentAtTo)
     for open rates, click rates, and best/worst subject lines.
   - analytics.getLinkAnalytics (days=30) for total link-page clicks and
     top links by click count.
   - content.listPostAnalytics (postedAtFrom/postedAtTo, page all results)
     for post engagement across every connected platform.
4. Write report.md to the current directory with this structure:

# Monthly Marketing Report — [Month Year]

## Mail
- Broadcasts sent, overall open rate, overall click rate
- Best-performing subject line (highest open rate) with the number
- Worst-performing broadcast (lowest open rate) and one hypothesis why

## Links
- Total link-page clicks last month
- Top 3 links by click count with their labels and numbers

## Social Posts
- Posts published per platform
- Top post per platform (most engagement) with the metric
- Any platform with zero posts last month — flag it

## What Worked
Exactly 2-3 bullet points, each with a number.

## What Flopped
Exactly 1-2 bullet points, each with a number.

## 3 Actions for Next Month
1. [Specific action tied directly to the data above]
2. [Specific action tied directly to the data above]
3. [Specific action tied directly to the data above]

Write like an analyst briefing the owner — not a data dump.
Skip any channel section if that profile had zero activity on it.

Make an account and connect Loopi to your agent →

What the agent pulls

The agent makes one authenticated MCP call per channel per profile. No browser scraping, no CSV exports, no separate logins. Every number comes through the same connection you authorized once.

What Loopi handles for you

The agent doesn't stitch together exports or scrape dashboards. Every number arrives through one MCP connection.

• All channels in one call surface. Mail broadcasts, link-page clicks, and post analytics are all reachable through the same MCP connection — no separate logins, no exports, no copy-paste.
• Date math is automatic. The agent infers last month's exact date range — first to last day, UTC — so you never need to type a date. Run it on the 1st or the 15th; it always covers the right window.
• Multi-brand rollup via profile.listProfiles. Run more than one brand under your account and the agent covers every profile in the same run. One prompt, all your channels.
• Saved locally, no login to read it. report.md lands in your working directory. Share it as a file, paste it into a doc, or just read it in the terminal — it never touches a hosted dashboard.
• Synthesis, not a data dump. The agent reads the numbers across all channels and identifies the signal: which channel over-performed, which floundered, what the numbers suggest you do next. That's the part that used to take an hour.

What the result looks like

Here is a realistic example of report.md from a solo service business — a law firm with one Loopi profile, active on mail and links, posting occasionally to Instagram. The agent found the pattern nobody noticed: the newsletter nobody reads, the bio link everyone clicks, and one month of near-zero social output that cost them audience momentum.

# Monthly Marketing Report — May 2026

## Mail
- Broadcasts sent: 3
- Overall open rate: 41% · Overall click rate: 6.2%
- Best subject line: "Why I turned down a $200K case" — 58% open rate
- Worst broadcast: "May newsletter" — 19% open rate. Generic subject likely
  bypassed by readers who don't recognize the sender name on a crowded day.

## Links
- Total link-page clicks: 284 (vs 201 the prior 30 days — +41%)
- Top links:
  1. Book a free consultation — 147 clicks
  2. Our practice areas — 63 clicks
  3. Recent case results — 41 clicks

## Social Posts
- Instagram: 2 posts published
  Top post: "3 things to do before signing any contract" — 1,840 views, 94 likes
- TikTok: 0 posts — flagged (no activity last month)

## What Worked
- The "turned down a case" story email hit 58% open rate — highest in 6 months.
  Personal-stakes subject lines are outperforming newsletters 3:1 on open rate.
- Bio page clicks up 41% MoM. The consultation booking link is carrying 52%
  of all clicks — the page is converting; the bottleneck is getting people there.

## What Flopped
- "May newsletter" at 19% open rate dragged the monthly average down 8 points.
  Non-specific subject lines are reliably the bottom performers.
- TikTok was dark all month. The Instagram post that got 1,840 views suggests
  short-form video is worth posting — but there was nothing to measure.

## 3 Actions for Next Month
1. Replace every newsletter subject line with a specific story or case angle —
   "what happened when a client ignored my advice" beats "June newsletter."
2. Post at least 4 TikToks in June using the same contract/case angle that
   worked on Instagram. Reuse the content; change the caption hook.
3. Add one more destination to the bio page — case results are at 41 clicks
   with almost no promotion. A direct link from the next email could double it.