Skill: Data Master

Daily DE funnel spreadsheet. One row = one calendar day.

Funnel stages

  1. Unique Visitors — distinct users who visited
  2. Choose Your Fish — product page (primary campaign landing page)
  3. Order Page — checkout page
  4. Created Orders — orders initiated (not necessarily paid)
  5. Paid Orders — successfully completed and paid

Columns

Column Formula
Date
Website Version — (e.g. version1)
Unique Visitors
Cost per Unique Visitor (EUR) Marketing Spend / Unique Visitors
Choose Your Fish
Order Page
% Order Page Order Page / Choose Your Fish · target > 40%
Created Orders
% from Order Page Created Orders / Order Page
% from Unique Visitors Created Orders / Unique Visitors
Unpaid Orders
% Unpaid Unpaid / Created Orders · target < 20%
Paid Orders
% Paid Paid / Created Orders
PayPal / Card / Invoice / Klarna / Apple Pay / Google Pay count + % share

KPIs and targets

KPI Formula Targets
Conversion Rate (CR) — primary Paid Orders / Unique Visitors min > 0.60% · target > 2.00% · ambitious > 4.00%
CPA — second core Marketing Spend / Paid Orders no fixed benchmark yet
Product → Checkout Order Page / Choose Your Fish > 40%
Unpaid rate Unpaid / Created Orders < 20%

Agent behavior

When given Data Master data:

  1. Flag days where CR < 0.60% and explain potential causes
  2. Flag Order Page % < 40% and suggest UX hypotheses
  3. Flag Unpaid % > 20% (> 30% is critical)
  4. When version changes, run a before/after comparison on all KPIs
  5. Note anomalies > 30% deviation from rolling average
  6. Flag significant shifts in payment method share

Accessing live data

The Data Master spreadsheet is stored in SharePoint. There are two ways to give Claude access to the current data:

Option A — Microsoft 365 Connector (recommended, no setup needed)

  1. In Claude.ai, go to Settings → Connectors and enable the Microsoft 365 connector
  2. Sign in with your work Microsoft account (must be an Entra/work account, not a personal @outlook.com)
  3. In any conversation, ask Claude to “read the Data Master spreadsheet” — Claude will find and pull it from SharePoint on demand

Option B — Manual upload Export the relevant sheet as CSV and upload it to your Claude.ai Project under Project knowledge. Re-upload whenever the data changes.

Notes

  • Daily granularity only — no hourly breakdown
  • Always segment by website version when versions overlap
  • CR and CPA are the goal; funnel ratios are diagnostic

Example questions

  • “What is the average CR and CPA per website version?”
  • “On which days did Order Page % fall below 40%, and what else changed?”

This site uses Just the Docs, a documentation theme for Jekyll.