Indeed Runs Analytics 60% Faster to Help Millions Find Jobs

The web has transformed how people search for and find career opportunities — and Indeed is at the forefront. From office managers to maintenance workers and nurses to nutritionists, job seekers searching for the right career fit easily browse listings, research salaries and get to know companies. With more than 3.5 million employers and 665 million Job Seeker Profiles,¹ Indeed is the world’s #1 job site² and a global leader in job matching and hiring.

Connecting the right person to the right job at scale isn’t always so simple. It involves enormous amounts of data from Indeed’s website and business operations. The company relies on Snowflake’s AI Data Cloud to power mission-critical reports and insights, streamline collaboration with advertising partners via data clean rooms, and achieve a self-service data lakehouse architecture that ultimately creates a better product for the employers and job seekers Indeed serves.

Indeed’s massive data lake — which currently contains 52PB of data — serves numerous purposes, including providing a single source of truth for reporting and analytical systems.

Prior to using Snowflake, Indeed relied on a PostgreSQL data warehouse for reporting and analytics, but performance and storage limitations presented challenges. After evaluating multiple solutions, Indeed selected one for its “gold layer” for mission-critical reporting that combined optimal performance, ease of use, near-infinite scalability, and an extensive network of connectors, drivers, programming languages and utilities: Snowflake. 

Implementing Snowflake on AWS helped Indeed modernize its data ecosystem while considerably shortening time to deliver on service-level agreements for key reporting across marketing, revenue, and environmental, social and governance (ESG). With Snowflake’s column-level security through the Horizon Catalog, Indeed can make sure team members can access its repository of data at the right level. 

With granular masking policies, there’s only ever one logical copy of a data set — which greatly simplifies governance and minimizes storage costs. The enhanced reporting and analytics available with Snowflake surfaces key insights about Indeed’s operations and customer interactions.

Indeed’s data engineering teams found themselves increasingly overwhelmed with a backlog of requests as more teams wanted to self-serve data to help drive timely, business-critical decisions and operations. This kicked off an architecture evaluation, landing the team on a lakehouse design. After evaluating other open table format solutions, such as Delta Lake and Apache Hudi™, Apache Iceberg aligned perfectly with Indeed’s plan for a “write once, read anywhere” paradigm because of its extensive and growing ecosystem. 

With a single source of truth for querying data, Indeed’s data engineering team could break down data silos and cut out extra querying steps that cost both time and money. According to Daniel DeMara, Staff Software Engineer at Indeed, “We decided to convert new and old data sets in our data lake to Iceberg format tables in Amazon S3 and integrate with Snowflake.” For each petabyte that is converted into Iceberg table format, Indeed bolsters its data centralization strategy while making it easier for staff to leverage Snowflake’s powerful platform.

With the data lake in Iceberg format, Indeed’s analysts can choose from a variety of solutions — including Snowflake — for querying data in the data lake. During Indeed’s testing, Snowflake delivered significantly greater query performance, and cost savings ranged from 43–74%, depending on load size compared to competing analytical tools like Amazon Athena. Cost-efficient querying, combined with ease of use, and simplified security and governance make Snowflake the preferred solution for Indeed’s data analysts to self-serve. As a result, they can now independently develop analytics, understand user journeys, and inform decisions that impact both employers and job seekers.

Indeed also extends the AI Data Cloud for less technical teams. By combining Snowflake with GrowthLoop — a customer segmentation platform designed for Snowflake — Indeed’s marketers can more effectively acquire and engage customers. Previously, deploying a single marketing campaign took three months to coordinate with analysts, engineers and other teams. Synchronizing data between Indeed’s source systems and campaign destinations made it difficult to build audiences and understand customers. Now, marketers can quickly go from idea to campaign launch — sometimes on the same day — with considerably less assistance from technical staff. 

“Unifying our data on Snowflake and leaning on partners available on the Snowflake Marketplace allowed us to democratize audience-building for our marketers and save a lot of time," says Suresh Susarla, Senior Director of Software Engineering for Marketing Systems at Indeed.

Being able to reliably test things like advertising impact is a key aspect of running successful paid campaigns, but doing so requires using first-party data that may have personally identifiable information (PII). By doing data collaboration with Snowflake Data Clean Rooms, Indeed and its partners securely join first-party audiences with campaign exposure and conversion data, run aggregations, and then compare test group and control group performance. 

“We can measure differences in their behavior on our site by conducting analyses within the data clean room, where we have access to person-level data with PII obscured and anonymized,” says Joe Zucker, Senior Manager, Marketing Analytics at Indeed. As a result, Indeed has achieved a transparent, repeatable incrementality testing workflow for understanding how marketing translates into job seeker profiles, employer accounts and job applications. Marketers leverage these insights to optimize advertising expenditures and, ultimately, save the company millions of dollars.

“The majority of our partners are on Snowflake Data Clean Rooms because we’re able to generalize most easily with Snowflake and reduce as much overhead as possible,” says Zucker. Onboarding new partners to Snowflake Data Clean Rooms is straightforward, which keeps Indeed and its collaborators focused on assessing marketing impact rather than struggling with data sharing logistics.

For a company connecting millions of job seekers and employers, speed is a non-negotiable. The dashboards employers rely on to track how their job listings are performing need to respond consistently in under a second at any traffic level. As the platform grew and usage patterns became less predictable, Indeed was running 10 separate computing clusters just to maintain that standard. The infrastructure bill was climbing, and so was the complexity of keeping it all running. 

Moving the workload to Snowflake cut the cost of serving those APIs by 70%, from $2,200 to $670 per million requests, and shortened data freshness from 15 hours to 3. Indeed also adopted Snowflake Interactive Analytics to tackle both problems at once. By consolidating onto interactive tables and warehouses, the team reduced its cluster count from 10 to 4. They were able to do so because interactive warehouses maintain consistent, sub-second performance even during spikes in demand. 

The gains were significant: response times improved 60% on average and 66% at the 95th percentile, and infrastructure costs fell a further 45%. For the employers and job seekers depending on Indeed's platform, that means consistently fast dashboards, at any scale, any time.

Snowflake Data Clean Rooms will continue to play an important role in Indeed’s future testing as well, especially for employer-focused campaigns. “Accessing lookalike audiences that are second-party to Indeed, first-party to the publisher, is the frontier for how we can continue refining that audience testing framework,” Zucker says. Indeed also plans to use Snowflake Data Clean Rooms for near real-time campaign insights, enabling marketers to optimize even faster.

As the Indeed data community continues to build on the “gold layer” of business reporting, some teams are also using it upstream for analytics, data transforms and eventually writing to it as well. Snowflake Interactive Analytics is extending that foundation further, maintaining consistent performance as concurrency demands grow. Indeed is standardizing how its core business measures are defined, so a single definition powers its dashboards, its APIs and its AI agents, rather than each system telling a slightly different story.

With a modern data lakehouse that saves costs and a data-driven marketing strategy that connects more customers to the right opportunities, Indeed is making good on its mission: to help more people get jobs. “The trusted partnership we’ve developed has already made an impact on finding technology solutions that will help us achieve our vision,” says Zucker. “We’re very excited about our path ahead with Snowflake.”