Podcast Analytics Platform
Transformation layer powering podcast performance analytics, artist royalty calculations, and content recommendation features. Processing billions of listening events into actionable creator insights.
Product Telemetry Analytics
Comprehensive dbt project transforming product usage data for self-serve analytics. Enables 500+ employees to answer their own data questions with trustworthy, well-documented models.
Flight Operations Intelligence
Operational data transformations powering delay predictions, customer experience metrics, and revenue management. Critical for flight scheduling and crew optimization decisions.
Marketing Attribution Platform
Multi-touch attribution modeling transforming customer journey data into actionable marketing insights. Powers revenue forecasting and partner commission calculations.
What dbt Developers Actually Build
Before writing your job description, understand what dbt engineers do in practice. Here are real examples from companies using dbt in production:
Media & Streaming Platforms
Spotify uses dbt to power their podcast analytics and content recommendations:
- Transforming billions of listening events into user behavior models
- Building artist royalty calculation pipelines with complex business logic
- Creating content recommendation features from engagement data
- Maintaining data contracts across dozens of data-producing teams
Warner Bros. Discovery leverages dbt for cross-platform analytics:
- Unifying viewership data from multiple streaming services
- Building attribution models for content performance
- Creating subscriber lifetime value calculations
- Documenting transformation logic for regulatory compliance
Technology & SaaS
GitLab runs their entire analytics platform on dbt:
- Product telemetry transformations from billions of events
- Customer health scoring and churn prediction models
- Usage-based billing calculations with audit trails
- Self-serve analytics enabling 500+ non-technical users
Hubspot uses dbt for marketing analytics at scale:
- Multi-touch attribution modeling across customer journeys
- Revenue recognition and forecasting transformations
- Product adoption metrics for customer success teams
- Partner commission calculations with complex logic
Travel & Logistics
JetBlue built their operational analytics on dbt:
- Flight operations data powering delay predictions
- Customer experience metrics from touchpoint data
- Revenue management transformations
- Crew scheduling optimization inputs
Instacart uses dbt for delivery intelligence:
- Real-time delivery optimization data pipelines
- Shopper performance analytics
- Inventory prediction model inputs
- Customer segmentation transformations
dbt in the Modern Data Stack: Understanding the Ecosystem
When evaluating candidates, understanding dbt's role in the broader data ecosystem helps you assess transferable skills.
The ELT Revolution
dbt is part of the ELT (Extract, Load, Transform) paradigm that replaced traditional ETL:
| Traditional ETL | Modern ELT with dbt |
|---|---|
| Transform before loading | Load raw, transform in warehouse |
| Proprietary tools (Informatica, SSIS) | SQL-based, open source |
| Centralized data team | Distributed ownership |
| Limited version control | Git-native workflows |
| Documentation separate | Docs generated from code |
| Testing afterthought | Tests as first-class citizens |
How dbt Fits the Stack
| Layer | Common Tools | dbt's Role |
|---|---|---|
| Extraction | Fivetran, Airbyte, Stitch | Receives raw data |
| Storage | Snowflake, BigQuery, Redshift | Execution engine |
| Transformation | dbt | Core transformation layer |
| BI/Analytics | Looker, Tableau, Metabase | Consumes dbt models |
| Reverse ETL | Census, Hightouch | Pushes dbt outputs to tools |
| Orchestration | Airflow, Dagster, dbt Cloud | Schedules dbt runs |
dbt Core vs. dbt Cloud
Understanding the deployment options helps assess candidate experience:
| Aspect | dbt Core (Open Source) | dbt Cloud (Commercial) |
|---|---|---|
| Cost | Free | Subscription-based |
| Hosting | Self-managed | Managed SaaS |
| Scheduling | External (Airflow, etc.) | Built-in scheduler |
| IDE | Local (VS Code, etc.) | Web-based IDE |
| CI/CD | Configure yourself | Built-in |
| Best for | Teams with DevOps capacity | Teams wanting simplicity |
Both require the same SQL and modeling skills—the difference is operational.
The Analytics Engineering Role: What dbt Created
dbt didn't just create a tool—it created a discipline. Analytics Engineering sits between data engineering and data analysis:
The Role Spectrum
| Data Engineer | Analytics Engineer | Data Analyst |
|---|---|---|
| Pipeline infrastructure | Data transformation | Business insights |
| Python, Spark, Airflow | SQL, dbt, data modeling | SQL, BI tools |
| "Get data into warehouse" | "Make data trustworthy" | "Answer business questions" |
| Cares about scale | Cares about correctness | Cares about relevance |
What Analytics Engineers Own
- Data Models: Designing dimensional models, fact tables, and marts
- Business Logic: Encoding how metrics are calculated consistently
- Data Quality: Testing assumptions and catching data issues
- Documentation: Making data discoverable and understandable
- Stakeholder Enablement: Helping analysts self-serve
Why This Matters for Hiring
When you post for "Analytics Engineer," understand that:
- They expect dbt or similar tooling
- They care deeply about data quality
- They want influence over how data is modeled
- They measure success by analyst enablement, not just pipeline uptime
Recruiter's Cheat Sheet: Spotting Great Candidates
Conversation Starters That Reveal Skill Level
Instead of asking "Do you know dbt?", try these:
| Question | Junior Answer | Senior Answer |
|---|---|---|
| "How do you decide between building a new model vs. adding to an existing one?" | "Whatever is easier" | "I consider grain, reusability, and downstream consumers. If the grain differs or it's a different business concept, it's a new model" |
| "A stakeholder says a metric is wrong. How do you investigate?" | "Check the SQL" | "I trace lineage from source to mart, verify business logic definitions with stakeholders, check for data quality issues at each stage, and compare against known-good historical values" |
| "Your dbt project takes 2 hours to run. How do you speed it up?" | "Use a bigger warehouse" | "I'd analyze the DAG for bottlenecks, convert full-refresh models to incremental where appropriate, reduce unnecessary dependencies, and consider materializing expensive CTEs" |
Resume Signals That Matter
✅ Look for:
- Specific scale context ("Built dbt project with 400+ models serving 200 analysts")
- Business impact language ("Created customer LTV model adopted by finance team")
- Testing and quality mentions ("Implemented data contracts reducing downstream incidents 70%")
- Stakeholder collaboration ("Worked with product to define activation metrics")
- Modern data stack familiarity (Snowflake/BigQuery + dbt + Looker pattern)
🚫 Be skeptical of:
- Listing dbt alongside 10 other tools at "expert level"
- No mention of business context or stakeholder interaction
- Only tutorial-level projects (Jaffle Shop, sample datasets)
- No evidence of testing or documentation practices
- Claiming dbt expertise but unclear on warehouse experience
GitHub/Portfolio Signals
Good signs:
- Well-structured dbt projects with staging/marts separation
- Custom schema tests and macros showing deeper understanding
- Documentation with business context, not just technical descriptions
- Evidence of incremental models and optimization
- CI/CD configuration showing production workflows
Red flags:
- Only the Jaffle Shop tutorial (everyone has done this)
- No tests in the project
- Flat model structure with no organization
- Models named "model_1", "final_v2", "new_model"
- No README or documentation
When dbt Matters Most
High-Impact dbt Scenarios
dbt shines when:
- Multiple data sources need unified business definitions
- Self-service analytics requires trustworthy, documented data
- Business logic is complex and needs version-controlled single source of truth
- Data quality is critical and needs automated testing
- Collaboration is essential between data and business teams
When dbt Might Be Overkill
Consider simpler approaches when:
- You have a single data source with minimal transformation
- Your analytics are ad-hoc with no need for repeatability
- The team is tiny (1-2 people) and velocity matters more than process
- You're in early-stage exploration before committing to data modeling
The dbt + Warehouse Relationship
dbt is tightly coupled to your warehouse choice:
| Warehouse | dbt Support | Notes |
|---|---|---|
| Snowflake | Excellent | Most common pairing |
| BigQuery | Excellent | Strong ecosystem |
| Redshift | Good | Some feature limitations |
| Databricks | Good | Growing fast |
| PostgreSQL | Supported | Less common in production |
When hiring, ask about their warehouse experience—dbt skills transfer, but warehouse patterns differ.
Common Hiring Mistakes
1. Requiring "3+ Years of dbt Experience"
dbt reached mainstream adoption around 2019-2020. More importantly, dbt is a tool—what matters is SQL depth and data modeling experience. Someone with 10 years of data warehouse experience and 6 months of dbt will likely outperform someone with 3 years of dbt but shallow SQL skills.
Better approach: "Experience with dbt or similar transformation tools (Dataform, SQLMesh). Strong SQL and data modeling required."
2. Ignoring SQL Fundamentals for dbt Knowledge
A candidate who knows dbt macros but can't write a window function or explain slowly changing dimensions is limited. dbt is 90% SQL—the other 10% is learnable quickly.
Test this: Give a SQL problem that requires CTEs and window functions. Ask them to explain their approach to modeling a specific business concept.
3. Over-Testing dbt Syntax
Don't quiz candidates on Jinja syntax or specific dbt commands—they can look these up. Instead, test:
- Data modeling decisions ("How would you model subscription revenue for a SaaS product?")
- Quality thinking ("How do you catch data issues before they reach dashboards?")
- Stakeholder communication ("A product manager says the numbers look wrong—walk me through your process")
4. Missing the Collaboration Aspect
Analytics Engineers work with stakeholders more than most data roles. A technically strong candidate who can't explain data concepts to non-technical users or gather requirements effectively will struggle. Include behavioral questions about stakeholder communication.
5. Conflating dbt with Data Engineering
dbt is transformation, not the entire data stack. If you need someone to build extraction pipelines, manage Airflow DAGs, or handle Spark jobs, dbt expertise alone isn't enough. Be clear about whether you need an Analytics Engineer or a Data Engineer who also knows dbt.