Przejdź do treści
Biblioteka skilli
Marketing i treści · Planowanie

kai-analytics

/kai-analytics

Konfiguruje analitykę i atrybucję: plan śledzenia, konwencje UTM, projekt dashboardów i wybór modelu atrybucji.

Licencja
MIT
Ostatnia zmiana repozytorium
18.10.2026

Ocena i bezpieczeństwo

90/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść25 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
1
Skrypty
0
Komentarze HTML
0
Adresy URL w SKILL.md
0
Przestarzałość

Brak sygnałów. Repozytorium zmienione 18.10.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co robi i kiedy

Konfiguruje analitykę i atrybucję: plan śledzenia, konwencje UTM, projekt dashboardów i wybór modelu atrybucji.

Wyzwalacze z opisu

analytics setupattributiontracking planUTMmarketing analyticsdashboard setupmeasurement strategyhow do I track
Oryginalny opis (EN)

Analytics and attribution setup — tracking plan, UTM conventions, dashboard design, and attribution model selection. Use when "analytics setup", "attribution", "tracking plan", "UTM", "marketing analytics", "dashboard setup", "measurement strategy", "how do I track", "which metrics", or any request to set up or improve marketing measurement and attribution.

Jak działa

  1. Objective
  2. Done when
  3. Constraints
  4. Context
  5. Escalate when
Sekcje z nagłówków SKILL.md, w oryginale · 81 linii, 946 słów
Podgląd SKILL.md
---
name: kai-analytics
description: Analytics and attribution setup — tracking plan, UTM conventions, dashboard design, and attribution model selection. Use when "analytics setup", "attribution", "tracking plan", "UTM", "marketing analytics", "dashboard setup", "measurement strategy", "how do I track", "which metrics", or any request to set up or improve marketing measurement and attribution.
---

# /kai-analytics — A Measurement System Whose Numbers Can Be Trusted

## Objective

A marketing measurement system the team can operate: an event taxonomy with defined triggers and properties, a UTM convention nobody has to guess at, an attribution model matched to the actual data volume, dashboard specs for three audiences, and a pixel/tag plan that survives a consent check. The result is that "where do customers come from" and "which ads work" have answers someone can reproduce.

Matching the model to the data is the load-bearing judgment. A data-driven attribution model on 50 conversions a month produces confident nonsense.

## Done when

Work type `strategy-plan` — floor **E3/C3/O1** (`harness/eco-floors.yaml`).

- **E3** — a named human approved the exact tracking plan, UTM guide, and dashboard specs.
- **C3** — the deliverables clear `banned_word_check`, and someone other than the author read them end to end. Every metric in every dashboard spec carries a written definition ("conversion rate = signups / unique visitors"); an undefined metric is a craft failure.
- **O1** — each P0 implementation item names the blind spot it closes, its owner, and how anyone will know the event is firing correctly. The plan's own outcome is instrumentation proven, not documents delivered.

A tracking plan nobody implements is not CLOSED. Its first work item is named in the implementation checklist.

## Constraints

- **Read `MARKETING.md` from the project root first.** It carries product, ICP, value prop, monetization, voice, current channels, and competitive landscape. If it does not exist, build it from the codebase — CLAUDE.md, README.md, PROJECT.md, package.json, landing pages, email/ad/analytics config — using the template from `/kai-email-system`, and confirm the draft. Do not ask the user what the product is.
- **Seven things must be known before recommending a model or a stack:** the business model and primary conversion event; current analytics tools (GA4, Mixpanel, PostHog, Amplitude); ad platforms in use; the CRM or email tool; current tracking status (nothing, partial, broken, outdated); the questions they actually need answered; and the sales cycle type (self-serve, sales-assisted, enterprise).
- **Never recommend tracking that violates GDPR or CCPA without a consent mechanism.** Every plan touching EU or California audiences carries a consent-management note.
- **UTM values are consistent or they are useless** — lowercase, no spaces, no special characters, no mixed case.
- **Event names follow one convention** (snake_case preferred), and each event declares its trigger, its properties, and the tools it lands in.
- **Attribution model must match data volume.** Data-driven needs 600+ conversions/month. Recommending above the data is a craft failure, not a stretch goal.
- **Every attribution recommendation ships with its caveats** (below). Attribution language that implies proof of causation without a counterfactual does not ship.
- **Dashboard metrics carry definitions**, not just names.

## Context

| Need | Load |
|---|---|
| Attribution mechanics and model tradeoffs | `knowledge/playbooks/analytics-attribution.md` |
| Pixels, tags, server-side, event plumbing | `knowledge/playbooks/technical-marketing-tracking.md` |
| Which metrics matter and their benchmarks | `knowledge/playbooks/saas-metrics-guide.md` |
| Business model, channels, tools in use | `MARKETING.md` (project root) |

**Measurement maturity** — diagnose this before prescribing anything:

| Level | State |
|---|---|
| 0 | No tracking beyond platform defaults |
| 1 | Basic GA4 + ad platform pixels |
| 2 | UTMs + event tracking + basic attribution |
| 3 | Multi-touch attribution + cohort analysis + LTV tracking |

**Attribution model selection:**

| Model | Best for | Limitation |
|---|---|---|
| Last-click | Short sales cycles, ecommerce | Ignores awareness channels |
| First-click | Brand-heavy businesses | Ignores nurture channels |
| Linear | Multi-channel, even contribution | Oversimplifies |
| Time-decay | Long sales cycles, B2B | Complex to implement |
| Position-based (U-shaped) | Most B2B SaaS | Requires multi-touch data |
| Data-driven (GA4) | High-volume businesses | Needs 600+ conversions/month |

**Attribution caveats — state these beside every recommendation:**

- Platform dashboards optimize for their own pixel, identity graph, attribution window, and modeled conversions.
- GA4, CRM, payment, and ad-platform revenue will disagree when UTMs, consent mode, offline conversions, refunds, or sales-cycle delays differ.
- Last-click is useful for capture channels; it is not proof that awareness or nurture did nothing.
- Multi-touch models describe observed journeys. They do not prove incrementality without holdouts, geo tests, lift studies, or matched-market tests.
- Attribution stays directional in budget conversations until event QA, UTM hygiene, consent coverage, and CRM joins are verified.

**UTM convention:** `utm_source` = platform name, lowercase (google, meta, linkedin) · `utm_medium` = traffic type (cpc, email, social, organic, referral) · `utm_campaign` = campaign name with date prefix (`2026-03_spring-launch`) · `utm_content` = ad variant or content identifier (`cta-v1`, `hero-image-b`) · `utm_term` = keyword or targeting, paid search only. Ship a builder template alongside the rules.

**Dashboards, three audiences.** Executive (weekly glance): revenue attributed by channel, blended CAC trend, conversion rate by stage, top 5 campaigns. Marketing ops (daily): traffic by source/medium, funnel with drop-off rates, UTM campaign table, ad spend vs revenue by platform. Channel-specific (per platform): platform metrics, audience segment performance, creative variant performance, budget pacing.

**Deliverables:** tracking plan, UTM convention guide, dashboard specs with metric definitions and data sources, pixel/tag setup guide (which tags, GTM vs direct vs server-side, consent requirements, validation steps), a priority-ordered implementation checklist, and a monthly data-quality audit checklist.

## Escalate when

- Data sources disagree materially and the user wants one number to report upward.
- Consent infrastructure does not exist and the requested tracking requires it.
- Conversion volume cannot support the model the user has already decided on.
- Implementation requires engineering time nobody has committed.
- The user asks for an incrementality claim the current setup cannot support.

Wersje

RepozytoriumWynikLicencja
Ta strona · cgallic/kai-cmo-harness90MIT
cgallic/kai-cmo-harness90MIT
cgallic/kai-cmo-harness90MIT
cgallic/kai-cmo-harness90MIT

Identyczna treść występuje jeszcze w 1 innych folderach na GitHubie. Tu pokazujemy jedną, najlepiej ocenioną kopię.

Migawka bazy z 13.11.2026 · wersja paczki adf69f8d0e8a (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100analytics
    Konfiguruje, audytuje i poprawia analitykę marketingową: GA4, UTM, śledzenie konwersji, Google Tag Manager, plan śledzenia, atrybucja i naprawa brakujących konwersji.
  • 90analytics-tracking
    Planuje i audytuje wdrożenie śledzenia w GA4: zdarzenia, konwersje, parametry UTM, Google Tag Manager i plan pomiarów.
  • 100analytics-tracking
    Konfiguruje, poprawia i audytuje śledzenie oraz pomiar analityczny: GA4, Tag Manager, zdarzenia, konwersje, UTM, atrybucja, plan śledzenia, Mixpanel i Segment.
  • 100analytics-tracking
    Konfiguruje, poprawia i audytuje śledzenie analityczne i pomiar: GA4, śledzenie konwersji i zdarzeń, UTM i Google Tag Manager.
  • 100ext-rae-utm-attribution-strategy
    Buduje spójną taksonomię UTM, konfiguruje modele atrybucji i łączy wydatki na marketing z przychodem w różnych kanałach.
  • 95analytics
    Konfiguruje, poprawia i audytuje śledzenie oraz pomiar analityczny: GA4, Tag Manager, zdarzenia, konwersje, UTM, atrybucja, plan śledzenia, Mixpanel i Segment.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.