Know which channel makes you money

Marketing analyst / growth lead / RevOps / CMO · Assemble in ~15 min; first run in <2 min once installed.

GTMEdit flowon X◐ readiness 47

step 1ready

Define channel data model + event tracking

analytics8 alts

step 2connector

Centralize the cross-channel data

⚡ GA4 / analytics

step 3ready

Query performance/attribution in plain English

data-analysis3 alts

step 4ready

Run multi-touch attribution + incrementality

attribution-modeling1 alts

step 5ready

Build dashboards + KPI tracking

analytics-reporting

step 6partial

Forecast revenue by channel

marketing-performance-analytics

step 7connector

Write report + push to Slack on schedule

⚡ bring your own tool

Did this recipe work for you?

What you get

▸Plain-English answers over your marketing data
▸Channel attribution report (CAC per channel, reallocation advice)
▸Scheduled written report

tools: GA4 free · Slack webhook free

tokens: ~$1-3 per report

≈ tokens only

What stays yours

· Connect the data exports
· Treat attribution as directional — gut-check big calls

The problem

When leadership asks which channel actually drives revenue, I want to centralize the data and query attribution/performance in plain English, with the agent building dashboards, running multi-touch attribution and incrementality, forecasting, and pushing a written report to Slack on a schedule, so I get same-day answers instead of living in spreadsheets and SQL.

Readiness

The reasoning core assembles well from the library: instrumentation, plain-English querying, multi-touch attribution+incrementality, and dashboards are all genuine ready skills (4 of 7 steps). But the two things that make it a *system* — centralizing live cross-channel data and pushing a scheduled report to Slack — are pure BYO holes, and forecasting is only partially covered. You get a strong analyst brain that still depends on you bringing the data pipe, a warehouse, a forecaster, and the Slack+cron plumbing.

Cybernetic layers

Memorybyo

Needs to remember prior periods' attribution numbers and channel baselines to report deltas and detect incrementality shifts run-to-run. No candidate provides state; bring a store — the warehouse itself or a snapshots table — keyed by reporting period.

Background / schedulingbyo

The core ask is a report pushed 'on a schedule', so it must run unattended. No candidate schedules; wire Claude Code's schedule/cron (or external cron) to trigger the pipeline and the Slack push.

Harnessready

Claude Code is the harness — skills load as the analyst, run SQL/analysis on the warehouse extracts, and invoke the Slack webhook. Invoked interactively for ad-hoc 'which channel drives revenue' questions and via the scheduler for the recurring report.

Click any step to see what fills it — and the ranked alternatives you can swap in.