In four short years, Instacart, the grocery startup that connects consumers with personal shoppers to offer home delivery in as little as one hour, has grown into a national company with operations in 26 metro areas.
To keep up with a growing customer base, Instacart used Segment to unify their data across the diverse toolsets each team uses, giving the analytics team more time to run powerful new models and analyses.
Instacart operates a platform that allows customers to purchase grocery and other items from local retailers for on-demand delivery. Customers can use Instacart’s website or mobile apps to choose from more than 500,000 category items from the retailers on Instacart’s platform.
Its partner stores now include local grocery stores and national chains like Whole Foods, Publix, Target, and Costco. The company has more than 300 employees and works with thousands of shoppers in cities across the country.
Instacart is a multi-sided marketplace, so it has many products to optimize and measure. Both shoppers and customers can access the service through di erent mobile-responsive websites or mobile apps available on Android and wiOS. The company also maintains relationships with retail partners that share their store inventory, so Instacart customers can often shop at in-store prices.
To understand how users moved in real-time across multiple platforms, Instacart needed a consistent strategy for collecting and managing the data in a way that would be efficient for both technical and non-technical teams.
While its customer base boomed, Instacart developed a data problem. In order to continue to grow its customers base, it wanted to continue to provide an incredible customer experience. This meant understanding how users moved in real-time across multiple platforms by collecting and managing the data eficiently.
"When we first started, we realized we had a lot of disparate data and a lot of customers who were doing a lot of different things, so we needed to make sure our tracking was consistent across platforms," said Fareed Mosavat, Senior Growth Product Manager at Instacart.
The company needed a plan to analyze funnel behavior, but wasn’t sure which technology platform to use. "We had no idea what tools were going to work best for the business because we were doing something unique," Mosavat said. "This business reaches across multiple platforms. It involves a real-world component and a lot of data that happens offline. Multiple user roles further complicate creation of a clean and consistent data taxonomy."
Different teams wanted different levels of access to the data. The growth team wanted an out-of-the box analytics tool, while the analytics team wanted the raw data in SQL, but they didn’t have an infinite number of engineers to work on these problems. Plus, they wanted to pull in data across outside touchpoints like email and support interactions to get a complete picture of the shopper and customer experience.
Instacart uses Segment to collect data from across applications, push that data to analytics and marketing tools, and establish a primary source of truth about customer behavior.
Instacart chose Segment as a single platform for collecting customer data. The company uses Segment Sources to collect data across applications (iOS, Android, web) and third-party cloud services (Zendesk, SendGrid, Facebook Ads). They use Segment Destinations to push the data out to analytics and marketing tools, and Segment Warehouses to schematize and load the data into a data warehouse. As a result, Segment has become the primary source of truth about its customer behavior.
Segment Destinations played an essential role in helping the company develop a dynamic analytics strategy, giving them the opportunity to move from tool to tool as their needs changed. Rather than building out individual integrations to test different platforms, Instacart used Segment to trial five analytics providers with one set of data tracking.
After the testing period, the company chose Amplitude as their primary analytics provider but also kept Google Analytics and Interana turned on to consult for particular questions. They turned off the remaining analytics vendors in their Segment dashboard with a few clicks.
"We try all these tools, and it’s as easy as flipping switches," Mosavat said.
Using Segment Sources, Instacart pairs behavioral data from their apps with customer support data from Zendesk, and email marketing data from Sendgrid.
"Sources lets us pull in third-party data about our customers," said Che Horder, Director of Analytics at Instacart. "With Zendesk, we’re measuring if support interactions increase orders, and with Sendgrid, we’re looking at the impact of our email campaigns on customer behavior."
We can pull that data into Redshift using the Segment Warehouses. And we use the data warehouse every day to nd out our consumers’ behavior and what our shoppers are doing.
Data from Sources is stored in Segment Warehouses, giving Instacart a central data repository that the entire company uses. "Segment collects all of our event data from all of our customer and shopper apps, across all platforms," said Horder. "We can pull that data into Redshift using the Segment Warehouses. Combining this event data with all of our transactional data gives us a comprehensive view of our customer and shopper interactions."
Segment allows Instacart to easily implement consistent events across their platforms, allowing for apples-to-apples analysis
"If we want to look at a sign-up event, it’s the same event for the web, mobile and mobile web," Horder said. "This consistency makes it very easy to analyze data from all of these disparate sources."
Instacart has saved hundreds of hours in engineering time by testing and implementing tools through Segment. "Segment is really another member of our data engineering team," Horder said. "It’s allowed us to focus on other, harder problems."
With the data engineering out of the way, the Growth and Analytics teams could focus on analyzing the customer and shopper experience. Particularly, they’ve been focused on identifying friction points in their apps and working with product team to eliminate them.
For example, by slicing their data across multiple parameters, the team found areas of improvement for their Android shopper app speci cally. A fix they shipped improved shopper productivity metrics significantly, Horder said.
Combining their Segment data with their shoppers table in Redshift, Horder and his team were also able to identify behavioral differences between new and experienced shoppers, and cater the product to help new shoppers become power users faster.
On the customer-facing side, Instacart was able understand details of their customers’ ordering behavior and what steps they take as they move toward purchase and checkout by analyzing that data in SQL.
"We found that users will visit a couple times before they actually place an order," Mosavat explained. "Previously, we were only attributing that to the last click. But now, because we have all the data, we’re actually able to do a longer attribution cycle and understand how many touches a user had before they actually convert. It’s already been really helpful, but it will become even more important as we develop marketing campaigns to reach customers based on that knowledge."
Conversely, Instacart can discover when customers are opting out of the ordering process and insert customer support managers to help them before they leave.
"When we see a consumer get to a point where they get stuck, at that point we want to offer assistance and see if our 'customer happiness team' can help them through that sticking point," Mosavat said.
"We’ll have more satis ed customers as we learn about what experiences are good and what are bad," he added. "We’ll get the best ordering process possible."
Segment for us is the beginning point for all of our event data and our behavioral data. It’s the tool to get the data everywhere it needs to go and then back to our warehouse so we can analyze it to keep improving our business.
Instacart’s data team will continue to rely on Segment as their service evolves and grows. "Segment for us is the beginning point for all of our event data and our behavioral data," Mossavat said. "It’s the tool to get the data everywhere it needs to go and then back to our warehouse so we can analyze it to keep improving our business."