How does ELT actually fit into the modern data stack?
The modern enterprise is a complex beast that generates, ingests, and emits huge and ever-increasing amounts of data in hundreds of different applications and locations. The average company uses 130 SaaS applications, growing at almost 20% each year, while larger organizations average even more: over 200 apps and tools. In many cases only 20 to 30 of them are mission critical for the entire company, but in some cases that number is much higher.
ELT is the glue that connects these hundreds of applications and, at the same time, the pipeline that orchestrates data movement for maximum utility.
But how?
In a modern data stack, ELT (Extract-Load-Transform) tools serve as the pipeline that moves data from diverse sources into centralized destinations (like cloud data warehouses or lakes) for downstream analysis. We’re talking tools like Extract, or Fivetran, or Stitch, or Airbyte.
(Check a comparison of ELT tools here.)
Unlike traditional ETL which transforms data before loading it into a destination, ELT tools replicate raw data into a data warehouse, deferring transformation to later stages. Those later states often use SQL-based tools like dbt in the warehouse or data lake.
This approach leverages the power of the warehouse for transformations and offers greater flexibility down the line. It also simplifies and speeds up data movement.
Data sources for the modern data stack
There are literally thousands of potential data sources for organizations with a modern data stack. They include a broad range of SaaS applications, databases, and APIs.
Those sources include:
-
Sales and CRM systems Examples: Salesforce, Pipedrive, and HubSpot
-
Marketing platforms Examples: Mailchimp, Facebook, or Google Ads
-
Product and support tools Examples: Jira, Amplitude
-
Revenue and finance systems Examples: Square, QuickBooks, and Stripe
-
Analytics from cloud apps
-
Usage data for websites
-
Mobile app data from app stores or analytics companies
-
E-commerce platforms Examples: Shopify, Magento, or Amazon Seller Central
-
Customer support tools Examples: Zendesk or Intercom
-
ERP systems Examples: SAP or Oracle
-
Databases Examples: Postgres or SQL Server
-
Cloud storage systems Examples: Amazon S3 or Azure
All of it depends on your business model, what your business does, and what critical operation it needs to track, analyze, and improve. A mobile app publisher will have very different needs and very different systems it needs to include in its modern data stack than a brand with third-party physical sellers and owned e-commerce.
Common data destinations for the modern enterprise
There are typically fewer data destinations than sources, though not always. Again, which ones you use really depends on your needs.
Common data destinations in the modern data stack include:
-
Cloud data warehouses Examples: Snowflake, Google BigQuery, Amazon Redshift
-
Data lakes Examples: Amazon S3, Azure Data Lake Storage, Wasabi
-
BI or analytics platforms Examples: Tableau, Excel, Airtable, Looker, Google Sheets
-
Operational destinations Examples: Salesforce, Hubspot, Braze, Zendesk, Slack
-
Databases Examples: MySQL, Oracle, SQL Server, MariaDB
Operational destinations are places where you’re using data gathered from other data sources that enrich actions you want to take with that data. For example, you might take data from website analytics to inform cohort or audience creation in Salesforce or Hubspot.
Other destinations are for longer-term storage as well as use for strategic business analysis.
How ELT connects the spaghetti of the modern data stack
Let’s say you’re one of those big brands with physical products and third-party resellers. How does ELT help you connect all the sources and destinations for your data, and what does that enable?
Perhaps you …
-
Sell through third-party physical retailers (like Target, Walmart, or Best Buy)
-
Sell through owned e-commerce channels (such as Shopify or Amazon DTC)
-
Advertise in a number of places
-
Have your own website with e-commerce
-
Have your own app with a store
-
Use CRM tools to manage customer data
-
Use marketing automation tools to manage campaigns
ELT can play a critical role in unifying data that would otherwise be siloed across a complex mix of platforms. You’ll want to extract raw data from each channel, load it into a centralized data warehouse, and transform it into analysis-ready models.
In doing all that, ELT will enable a brand like yours to have a holistic view of sales, marketing performance, inventory, and customer behavior.
The first step is connecting to all the relevant data sources. ELT tools like Extract, Fivetran, Airbyte, Stitch, can pull in data from:
-
Owned e-commerce platforms like Shopify, WooCommerce, or BigCommerce Get order data, customer behavior, cart abandonment, and product SKUs
-
Amazon Seller Central Get sales, advertising performance, and fulfillment data
-
Retailer portals or EDI feeds from third-party physical sellers like Walmart or Target Get sell-through data, in-store inventory, returns, and chargeback info
-
Ad platforms such as Meta, Google Ads, TikTok Ads, and Apple Search Ads Get ad spend, performance metrics, and campaign-level insights
-
Web analytics tools like Google Analytics or Adobe Analytics Track site visits, conversion funnels, and user flows
-
Payment systems like Stripe or PayPal Get transactions, refunds, and disputes
-
CRM and customer support tools like HubSpot, Salesforce, Intercom, and Zendesk Bring in customer touchpoints, tickets, and lifecycle data
-
Logistics and ERP platforms like NetSuite, SAP, or ShipBob Get information on inventory, fulfillment, returns, and cost of goods sold
Then you’re typically loading your data into a centralized destination like a cloud data warehouse such as Snowflake, Google BigQuery, Amazon Redshift, or even PostgreSQL.
Your ELT tool, whether Extract or another, automates the creation of tables, handles schema changes, and keeps the data synchronized on a regular schedule: hourly, daily, or even near real-time.
(By the way, Extract is super-flexible: you can trigger it via cron or API, or a set schedule.)
Finally, you might want to transform your data and match existing schema, or load it into BI applications for analysis or use. You might use tools like dbt, and you might clean, normalize, and join your data to fit into unified models that reflect how your business actually operates. For example, you might combine orders from Shopify and Amazon into a consolidated order and revenue model, or you might merge advertising data with revenue to calculate cross-channel ROAS (Return on Ad Spend) and blended CAC (customer acquisition cost).
Modern data stacks unlock … everything
By connecting and modeling data across these channels with an ELT tool like Extract, you’ve now put your modern data stack to work and unlocked powerful use cases.
For example, now you can see and impact:
-
Total revenue performance across all channels (owned and retail)
-
Attribution and marketing effectiveness, tying ad spend to both DTC sales and retail lift
-
Inventory optimization by analyzing sell-through rates and fulfillment delays
-
Customer segmentation and personalization based on omnichannel behavior
-
Retailer scorecards for internal benchmarking and account health monitoring
ELT is the glue. It’s the pipeline. It’s essentially the nervous system for the modern enterprise that connects sensory data (all the disparate data sources a brand touches) with processing, strategy, and ultimately action.
And that means a modern data stack connected with a smart and efficient ELT tool like Extract is absolutely critical for data-driven growth.
Want to try Extract at zero cost?
If you want to try Extract, it’s super-easy and completely free. Just create an account and start playing. As you can see on our pricing page, you get:
-
Up to 1 million monthly rows
-
Unlimited sources & destinations
-
Up to 5 live connections
-
2 platform users
-
Hourly syncs
… for exactly $0.
And when you go big, you’ll see savings of up to 70% compared to other ELT tools.
Try it today!
