# Measure

> |

- **Type:** Skill
- **Install:** `agentstack add skill-rmzlb-baaton-skills-measure`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [rmzlb](https://agentstack.voostack.com/s/rmzlb)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [rmzlb](https://github.com/rmzlb)
- **Source:** https://github.com/rmzlb/baaton-skills/tree/main/marketing/measure
- **Website:** https://baaton.dev

## Install

```sh
agentstack add skill-rmzlb-baaton-skills-measure
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Marketing Analytics

Measure what matters. Ignore what doesn't. Iterate based on data.

## How to Use

Action from arguments: $ARGUMENTS

| Action | What it does |
|--------|-------------|
| `report` | Generate weekly/monthly/quarterly report |
| `analyze` | Deep-dive on content performance |
| `test` | Design an A/B test |
| `benchmark` | Compare your metrics against industry benchmarks |

If no action specified, ask what the user needs.

## The Feedback Loop

This is the entire point of analytics:

```
Analyze → Hypothesize → Test → Measure → Iterate → Analyze
```

Every report should end with: "Based on this data, here's what we should try next."
If a report doesn't lead to action, it's a vanity exercise.

## Metrics by Platform

### LinkedIn
| Metric | What it tells you | Good | Average | Poor |
|--------|------------------|------|---------|------|
| Impressions | Reach of your content | >5K (small account) | 1-5K | 5% | 2-5% | 20 | 5-20 | 3/post | 1-2 | 0 |
| Profile views | Curiosity generated | >200/week | 50-200 | 50/week | 10-50 | 10K | 2-10K | 3% | 1-3% | 50 | 10-50 | 5 | 1-5 | 0 |
| Profile clicks | Curiosity | >100 | 20-100 | 10/tweet | 2-10 | 35% | 20-35% | 5% | 2-5% | 2% | 0.5-2% | 0.5% |
| Forward rate | Share-worthy | >1% | 0.2-1% | 2% | 0.8-2% | $4 |
| CPA | Acquisition cost | $80 |
| ROAS | Return on ad spend | >3x | 1.5-3x | 5 |

For full metrics glossary with formulas, see `references/metrics-reference.md`.

**Benchmarks vary by audience size, industry, and stage.** See `references/benchmarks.md`.

## Vanity Metrics vs Actionable Metrics

| Vanity (feels good) | Actionable (drives decisions) |
|---------------------|------------------------------|
| Follower count | Follower growth rate per content piece |
| Total impressions | Impressions-to-engagement conversion |
| Page views | Time on page + scroll depth |
| Email list size | Email-to-revenue attribution |
| "Likes" | Comments + DMs + saves (deep engagement) |

**Rule: if a metric doesn't change what you do next, stop tracking it.**

## Attribution

How did a customer find you? Four models:

| Model | How it works | Best for |
|-------|-------------|----------|
| First-touch | Credit goes to first interaction | Understanding discovery channels |
| Last-touch | Credit goes to last interaction before conversion | Understanding what closes |
| Multi-touch | Credit split across all touchpoints | Complex B2B journeys |
| Self-reported | Ask "how did you hear about us?" | The truth (often different from data) |

**Recommendation:** Use self-reported attribution as your primary signal.
Add a "How did you hear about us?" field to signup/checkout. The answers
will surprise you. Data attribution misses dark social (DMs, word of mouth, podcasts).

## A/B Testing for Content

Most content marketers don't A/B test because they think they need
huge sample sizes. You don't.

### What to Test (ordered by impact)

1. **Hook/headline** — Biggest impact on reach and engagement
2. **CTA type** — Question vs. direct ask vs. none
3. **Format** — Story vs. framework vs. contrarian
4. **Posting time** — Morning vs. evening, weekday vs. weekend
5. **Length** — Short (800 chars) vs. long (2,000 chars)
6. **Platform** — Same content, different platforms (via repurpose)

### How to Test with Small Audiences

You don't need statistical significance for content decisions:

1. Test for 5-10 posts (not 1 post vs 1 post)
2. Compare the same metric across the set
3. If one approach beats the other 7 out of 10 times, it's a signal
4. Act on the signal. If you're wrong, you'll see it in the next 10.

For detailed testing guide, see `references/ab-testing-guide.md`.

## Reporting Templates

### Weekly (5 min, every Monday)
```
This week's content:
- [List pieces published]
- Best performer: [which + why]
- Worst performer: [which + why]
- One pattern noticed: [observation]
- Next week's plan: [based on this week's data]
```

### Monthly (30 min, first Monday of month)
```
Monthly metrics:
- Content published: [count by platform]
- Total reach: [impressions across platforms]
- Engagement: [rate by platform, trend vs last month]
- Conversions: [DMs, signups, sales attributed to content]
- Best content piece: [which + analysis of why]
- Worst content piece: [which + what to learn]
- Audience growth: [by platform, trend]
- Content pillar performance: [which pillar drives results]
- Hypothesis for next month: [what to test]
```

### Quarterly (2 hours, every 3 months)
```
Strategic review:
- ICP still accurate? [validate against new customer data]
- Messaging still resonating? [top-performing language vs flops]
- Channel allocation: [which channels drive conversions, cut losers]
- Content format ROI: [which formats justify the time invested]
- Competitor moves: [anything changed in the landscape?]
- Budget reallocation: [shift spend to what works]
- Strategy update: [feed insights back to strategy]
```

For detailed report templates with examples, see `references/reporting-templates.md`.

## Content Performance Queries

Ask these with your content log data:

- "What are my top 5 posts by engagement rate this month?"
- "Which content pillar drives the most DMs?"
- "What hook style gets the highest engagement?"
- "Compare story posts vs framework posts on LinkedIn"
- "What's my best posting time based on the last 20 posts?"
- "Which repurposed format outperforms the original?"

For ready-to-use analysis prompts, see `references/content-performance-queries.md`.

## References

- `references/metrics-reference.md` — Full metrics glossary with formulas
- `references/reporting-templates.md` — Detailed report templates with filled examples
- `references/ab-testing-guide.md` — Practical A/B testing guide
- `references/content-performance-queries.md` — Analysis prompts for your content log
- `references/benchmarks.md` — Industry benchmarks by platform and audience size

## Related Skills

- `strategy/` — Feed quarterly insights back into strategy
- `create/` — Use performance data to write better briefs
- `repurpose/` — Track which repurposed format performs best
- `audit/` — Correlate audit scores with actual performance

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [rmzlb](https://github.com/rmzlb)
- **Source:** [rmzlb/baaton-skills](https://github.com/rmzlb/baaton-skills)
- **License:** MIT
- **Homepage:** https://baaton.dev

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-rmzlb-baaton-skills-measure
- Seller: https://agentstack.voostack.com/s/rmzlb
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.