How to Supercharge Product Intelligence with Reverse ETL?

• September 26th, 2022

Reverse ETL in Product Intelligence
About the Author 
Ali is the creator of Product Analytics Academy, an online school that provides high-quality product analytics courses. He's also the founder of Mentat Analytics, a top-rated analytics consulting agency. He previously worked on the data teams at Bird and Buzzfeed.

In our previous article, we covered the rise of Product-Led Growth (PLG) and the importance of strong product analytics for any company that plans on taking on a PLG approach. Specifically, we discussed how product intelligence is incorporated into the Modern Data Stack (MDS). In this article, we’ll cover how Reverse ETL– a process of syncing data from warehouse to operational SaaS apps, adds another yet crucial layer to the development of product intelligence. 

As you may recall, the MDS is built around a data warehouse. Data from your applications and all your third-party sources get deposited into the warehouse, where the data gets processed via a tool like DBT and subsequently served to your stakeholders through a business intelligence tool like Metabase. 

The tools used for depositing data from your application and third-party sources into your warehouse are referred to as “ETL tools,” with “ETL” being short for “Extract, Transform, and Load.” Hevo Data is one such excellent tool in this category to automate data replication and transformation from 150+ data sources into your warehouse. 

So, if an ETL tool is used for taking data from such sources to the warehouse, then a reverse ETL tool does exactly as you might think: taking data from the warehouse and depositing it into other destinations. 

Why Should I Use Reverse ETL in Product Intelligence?

When you read the description of what Reverse ETL does, an obvious question follows: 

If I put all this effort into taking data from my third-party tools into the warehouse, why do I need to take it back to these same tools again?

The answer lies in how the warehouse is used. 

Remember that the warehouse is the center of the stack, where all the data processing should take place. What comes into the warehouse is raw data, but through tools like DBT, you process that data to turn it into reliable, actionable information. The question is, how do you deal with this actionable information? 

One way is to crunch numbers and prepare reports that act as a base to arrive at data-based conclusions. Based on these conclusions, your teams would devise a business strategy to be implemented into operational systems like CRM, marketing, or advertising platforms. 

The other, and far superior, method is to use Reverse ETL. Product intelligence with Reverse ETL allows you to take processed and accurate data from your warehouse and send it where it’s needed, so that more people can take strategic action on data. 

You effectively create the ability to integrate data between third-party tools that don’t have an existing integration with one another by passing that data through your warehouse first. 

Through the use of Reverse ETL, you can take data from one application, send it to your warehouse for processing, then share that processed data with a different application. 

Reverse ETLs have another lesser-known but incredibly important use case, and that has to do with product analytics tools like Mixpanel, which rely on event data. Normally If you want to track events from your product’s backend, you have to implement code snippets in your server that send an event to Mixpanel every time a new row is created in a table that contains data that you want to track. The more kinds of events you want to track, the more coding you have to do. 

With simple and speedy Reverse ETL solutions like Hevo Activate, you can connect the tables in your warehouse directly to Mixpanel and send cleaned data to analyze it. Hevo Activate lets you connect your warehouse or database destinations like BigQuery, Snowflake, Redshift, or PostgreSQL to downstream SaaS applications like Intercom, Marketo, Salesforce, Mixpanel, etc. seamlessly. Using its robust and intelligent SQL query editor, you can also filter and transform warehouse data into target objects. Hevo Activate includes options to either continuously sync data between two systems or to define what triggers the synchronization.

By using product intelligence with reverse ETL, your whole process, from event tracking to analysis, becomes extraordinarily easy. Your engineers, product teams, and data teams will all be spared from lots of uninteresting work. And if your team doesn’t have any experience with product analytics tools like Mixpanel or with event tracking, then you can take one of the many product analytics courses we offer over at Product Analytics Academy to get up to speed. 

Putting It All Together: A Case Study of Using Reverse ETL in Product Intelligence

If you’re still confused about how the Reverse ETL is used in practice, don’t worry. In this section, we’re going to revisit our case study from our previous article covering the intersection of product intelligence and the Modern Data Stack. This case study was for GroupGrub, a fictional company offering an app that lets its users place grouped food orders so they can save on delivery and preparation costs. Their data stack looked like this:

GroupGrub Data Stack Using Hevo: Reverse ETL in Product Intelligence
GroupGrub’s Data Stack with ETL

This stack allowed them to funnel data to their Redshift warehouse from their application database Amazon Aurora PostgreSQL, Mixpanel, Salesforce, and Marketo using Hevo. This data was then cleaned using DBT and served to company-wide stakeholders through dashboards built in Metabase. 

In companies with a sales component to their business, their sales teams spend most of their time in their Sales/CRM tool. In GroupGrub’s case, that tool is Salesforce. If you’re in sales operations, you want to be able to see the performance of the enterprise customers you’ve brought onto your platform. For the sales team at GroupGrub, a good indication of this would be how often users view the menu of a restaurant, then proceed to complete an order. 

This conversion rate is a great metric for measuring how appealing that restaurant’s menu is, yet this information is only going to be captured in Mixpanel, GroupGrub’s product analytics tool. The sales team also wants to know how much total revenue the app has brought in for each restaurant since that informs how much they should rely on the app for their business. This information likely exists in GroupGrub’s Redshift warehouse. 

In an ideal world, both of these data points would be displayed front and center for each client’s profile in Salesforce. This is where the Reverse ETL comes in: both the conversion rate data (originally pulled from Mixpanel) and the restaurant revenue information can be retrieved from GroupGrub’s warehouse and synced with Salesforce, all due to the power of the Reverse ETL. This way, the sales team can easily make decisions on their most important clients, all within their CRM.

There are additional uses for the Reverse ETL as well. 

Reverse ETL in Product Intelligence: Up-leveling Customer Retention

Many companies build internal machine learning models that predict users’ propensity to churn. But it’s not enough to predict a user’s churn. The ultimate goal of the prediction is to prevent churn. 

In GroupGrub’s case, if a user is predicted to churn, then the marketing team would want to send the user a coupon that could encourage them to stick with GroupGrub as their app of choice. The coupon is sent to the user through their marketing platform, Marketo, but the output of the predictive model is stored in the warehouse. Yet again, Reverse ETL comes into play: users flagged as being high-risk are synced from Redshift to Marketo, triggering the coupon and reducing the user’s chances of churning. 

Last but not least, the GroupGrub team can supplement their existing event tracking setup with backend events that are synced to Mixpanel from the warehouse. 

Recapping everything so far, here’s what the GroupGrub team accomplished with their smart usage of a Reverse ETL tool:

  • Sales Operations: The sales team sees the data that is most relevant to them directly in Salesforce, allowing them to make better decisions and do so faster.
  • Churn Reduction: The marketing team sets up automated coupons that are sent to users most likely to churn, increasing engagement and reducing user churn.
  • Improved Product Analytics: The product team is able to receive backend events in Mixpanel, enhancing their understanding of the user’s journey throughout the product. 

GroupGrub’s updated stack now looks like this:

GroupGrub's New Data Stack Using Hevo: Reverse ETL in Product Intelligence
GroupGrub’s Data Stack with ETL and Reverse ETL

Final Words

To summarize, an ETL tool’s job is to put data in the warehouse and empower the business teams to derive insights, and a Reverse ETL tool’s main job is to take these holistic insights and sync them to other tools. This allows teams using those tools to operationalize insights, make faster and more intelligent decisions, and compete better.