---
title: Getting started
description: Access, login, and your first ten minutes in the new BI platform.
---

# Getting started

## Access

| Tool | How you sign in | Who has access |
| --- | --- | --- |
| Metabase | Google Sign-In, restricted to `@joe.coffee` accounts, at https://joe-metabase.fly.dev | brenden, nick, llexi (admins); ryan (Data Analyst); ask an admin to invite others |
| Reports admin | Admin API token (ask David) | Ops/analyst use |
| Raw SQL | Metabase's native SQL editor (recommended), or a read-only Postgres role for external tools | Analysts |

Metabase sends its own emails (password resets, subscriptions) as `reports@joe.coffee`; scheduled platform reports come from the same address.

## Your first ten minutes

1. **Sign in to Metabase** with your `@joe.coffee` Google account.
2. **Open the Home (Parity) dashboard.** It mirrors the Looker Home dashboard: network-level sales, orders, and trend cards.
3. **Browse the parity collection.** The dashboards rebuilt from Looker so far:
   - **Home (Parity)** - network overview, the old Looker landing page
   - **Sales Report (Parity)** - the per-shop sales workhorse (net sales, tax, tips, discounts, by day)
   - **Payout by Arrival Date (Parity)** - Stripe payouts by the date they hit the bank
   - **All Transactions (Parity)** - the transaction-ledger detail view (every order, refund, gift card load, with full money splits)
   - **Network Overview / Network Pulse / Store Spotlight** - internal network health views
   - **Shop Sales Report (Embed)** - the shop-locked dashboard used for merchant embeds and PDF reports
4. **Ask a question.** Click **+ New > Question**, pick the data-plane database, and start from a `bi.*` view (not a raw table - see below). The GUI builder does filtering/grouping without SQL.
5. **Or write SQL.** **+ New > SQL query**. The same `bi.*` views are the right entry point.

## The one habit that matters

Query the **`bi.*` views**, not raw tables, unless you know why you need the raw table. The views encode the metric contract (which rows count, which columns are cents, what "net" means). Raw-table queries that re-derive metrics are how numbers drift apart between surfaces - the exact problem this platform exists to eliminate.

Available views and what they answer are cataloged in the [data dictionary](./data-dictionary.md).

## Data freshness expectations

| Data | Freshness |
| --- | --- |
| Today's orders (live path: `recent_orders`, live dashboards) | ~20 seconds |
| Transaction ledger, fact tables, `bi.*` views | Complete through **yesterday**; nightly sync at ~05:20 UTC (~9:20pm Pacific) also refreshes a trailing 7-day window to pick up late refunds |
| Stripe payouts | Daily; payouts only arrive on banking days (no weekend rows is normal) |

If a closed day's numbers change slightly a day later, that is the trailing-window resync picking up refunds and adjustments - by design.