New Data Team Member Onboarding Guide

Portfolio note: This is an anonymized version of an onboarding guide I wrote for a data and analytics team. Company names, people, and internal resources have been replaced with placeholders.

Work through this guide in order during your first week. Each section builds on the last.

Don’t try to memorize everything — your goal is orientation and access, not mastery.

Section 1 — The Big Picture

By the end of this section: You’ll understand what our data stack is, why it’s built the way it is, and how the pieces connect.

What We’re Building

Our job is to make sure the right data gets to the right people in a reliable, trustworthy, and timely way. That means:

• Pulling data from every system the business runs on
• Storing and organizing it in one central place
• Transforming it into clean, meaningful models
• Making it available for analysis, reporting, and operational use

The Stack at a Glance

Our data stack has five layers. Data flows left to right:

Think of it this way: Fivetran brings data IN to Snowflake. dbt organizes it inside Snowflake. Tableau and Hightouch take data back OUT for people to use.

A Note on Our Stack’s History

• Snowflake has always been our data warehouse. BigQuery is used specifically for GA4 data export staging — it is not and was not our primary warehouse.
• We moved from Hevo Data to Fivetran for ingestion.
• We migrated our CRM from HubSpot to Zoho CRM — but many pipeline names and table names still say ‘HubSpot’ (e.g., hubspot_contacts). This is legacy naming. When you see HubSpot in a table or pipeline name, it refers to Zoho CRM data.
• A previous data engineer built much of the dbt infrastructure. His pipeline training video (recorded with the analytics team) is foundational context — ask the Head of Data for the link.

Section 2 — Your Tools & Access

By the end of this section: You’ll be logged in to every tool you need and know who to contact if something isn’t working.

the Head of Data owns access for all data platform tools. If you’re missing anything below, reach out to the Head of Data on Slack — not IT, not Engineering.

Always use your Company email (not personal) to log in to all data tools. MFA is required for Snowflake.

Your Snowflake Environment

• Your dev schema will be named dbt_[yourname] (e.g., dbt_jordan, dbt_sam)
• Production analytics data lives in the Data Warehouse database — this is what Tableau connects to
• Raw ingested data lives in the Stage database — you generally won’t query this directly

Section 3 — How Data Moves

By the end of this section: You’ll be able to describe how data flows from a source system all the way to a Tableau dashboard or Zoho CRM field.

The Standard Pipeline

1. A source system (e.g., MySQL, Zendesk, GA4) generates or updates data
2. Fivetran detects the change and syncs it to Snowflake on an hourly schedule
3. The raw data lands in Snowflake’s Stage database (a loading dock — unprocessed)
4. dbt runs on a schedule (every 3 hours intraday, plus a full daily job) and transforms Stage data into clean, joined models in the Data Warehouse database
5. Tableau connects to the Data Warehouse and refreshes dashboards from the clean data
6. Hightouch reads from the Data Warehouse and syncs enriched data back to Zoho CRM

A Few Things to Know About Fivetran

• Soft deletes: When a record is deleted in a source system, it’s not removed from Snowflake. Instead, a column called fivetran_deleted is set to TRUE. Our dbt models filter these out automatically.
• New tables don’t auto-sync. Enabling a new source table requires a manual step — ask the Head of Data. This is intentional cost control.
• Fivetran costs are based on Monthly Active Rows (MAR). We actively manage which tables sync to stay within budget.

The GA4 Exception

Google Analytics 4 doesn’t flow through Fivetran. It has its own custom pipeline because GA4 natively exports to BigQuery:

GA4 → Google BigQuery → dbt → Google Cloud Storage → dbt → Snowflake

The end result is the same — queryable GA4 data in Snowflake — it just takes a different route to get there.

Reverse ETL: What Hightouch Does

Most data flows into Snowflake. Hightouch does the opposite — it takes data from Snowflake and pushes it back into operational tools. Today, it syncs contact-level metrics (like GMV last 30 days) from Snowflake into Zoho CRM.

Note on naming: The Hightouch sync source table is called hubspot_contacts. This was built originally for HubSpot before we migrated to Zoho CRM. The table name hasn’t been updated yet — this is expected and known.

Section 4 — Snowflake: Your First Queries

By the end of this section: You’ll be logged in to Snowflake, oriented to the key databases, and able to run your first queries.

Navigating Snowflake

Key Tables to Know First

• order_line_items — the primary orders table. Use this for all order and GMV analysis.
• super_orders — DEPRECATED. Do not use. Data was migrated to order_line_items ~2 years ago.
• hubspot_contacts — contact-level data synced to Zoho CRM via Hightouch. Despite the name, this is Zoho CRM data.

Your First Queries

-- Preview the main orders table
SELECT * FROM order_line_items LIMIT 100;

-- Check record count
SELECT COUNT(*) FROM order_line_items;

-- Write to your dev schema (safe -- won't affect production)
CREATE OR REPLACE VIEW dbt_[yourname].my_first_test AS
SELECT * FROM order_line_items LIMIT 1000;

Best practice: Always add a LIMIT clause when exploring large tables. Avoid expensive full-table scans without a WHERE clause. When in doubt, write to your dev schema first.

Section 5 — Tableau: Finding & Using Dashboards

By the end of this section: You’ll be able to find, navigate, and filter dashboards in Tableau Cloud and understand how our published data sources work.

Tableau Cloud vs. Tableau Desktop

• Tableau Cloud (browser) — for viewing, exploring, and sharing dashboards. Most people use only this.
• Tableau Desktop (app) — for creating and editing dashboards. Requires a separate license from the Head of Data.

How Our Dashboards Are Built

Our setup uses published data sources:

• Each dashboard connects to a published data source — a shared, named connection to a specific Snowflake table
• Multiple dashboards share the same published source, so a single data refresh updates all of them
• All transformation logic lives in dbt — not inside Tableau. If something looks wrong in a dashboard, the fix is almost always upstream in dbt.

If you’re building a new dashboard: connect to an existing published data source rather than writing custom SQL inside Tableau. If the data you need doesn’t exist as a published source, build a dbt model first, then connect Tableau to it.

Section 6 — dbt & Pipelines (Conceptual)

By the end of this section: You’ll understand what dbt is, why we use it, and how it fits in — without needing to make any changes yet.

What dbt Is

dbt (data build tool) is where our data transformation logic lives. If Fivetran brings raw data into Snowflake, dbt turns that raw data into the clean, meaningful tables you actually query. Think of dbt models as SQL files that each define a table or view.

Key Concepts Glossary

You don’t need to write or modify dbt models right now. Browse dbt Cloud to see existing models and their lineage — read-only and safe to explore. A dedicated dbt deep-dive guide will be available when you’re ready to start contributing.

Section 7 — Ways of Working

By the end of this section: You’ll know our team norms, who owns what, and how to get help.

Team & Ownership

Key Norms & Gotchas

• GMV not GMB — always Gross Merchandise Value
• super_orders is deprecated — always use order_line_items
• HubSpot references = Zoho CRM — our CRM migrated but pipeline names haven’t all caught up yet
• The Head of Data owns all data platform tooling — not IT, not Engineering
• Don’t make changes to production dbt models without a pull request and review

Section 8 — Your First Week Checklist

Use this to confirm you’re fully set up by end of week one. Check in with the Head of Data on anything you can’t complete.

Access & Accounts

• Snowflake — logged in with Company email, MFA enabled, dev schema confirmed (dbt_[yourname])
• Tableau Cloud — logged in, can view at least one dashboard
• Tableau Desktop — license received and installed (if applicable to your role)
• dbt Cloud — logged in, can browse models and lineage
• Fivetran — logged in, can view connector list
• Hightouch — logged in
• GitHub — account added to data team repository
• Slack — joined #data-analytics and relevant channels
• Google Drive — Company account connected, data team shared folder accessible

Orientation

• Read Sections 1–3 of this guide and can describe the pipeline end to end in plain language
• Reviewed the D&A Knowledge Base — familiar with key tables, metric definitions, and team notes
• Watched the pipeline training walkthrough video (ask the Head of Data for the link)

Hands-On

• Run at least 3 exploratory queries in Snowflake against the Data Warehouse
• Found and navigated at least 2 Tableau dashboards
• Browsed dbt Cloud lineage — identified one model that feeds a dashboard you looked at in Tableau
• Located the Fivetran connector for one data source and noted its sync schedule

Questions & Next Steps

• Written down at least 3 questions to review with the Head of Data in your first 1:1
• Identified one area to go deeper on first (Snowflake, Tableau, dbt, or pipelines)

References & Further Reading

Welcome to the team. When in doubt, ask. We’d rather answer a question than have you guess.