Label Insight: Building the perfect data stack

About Label Insight

Label Insight provides comprehensive data about what’s in the products we buy, delivering transparency to consumers by powering analytics, marketing, merchandising, and ecommerce solutions for retailers and manufacturers. It covers more than 80% of top-selling food, pet, and personal care items in the United States and creates more than 22,000 high-order attributes per product.

Industries: Technology

Location: United States

Products: BigQuery, Cloud Vision API

Tell us your challenge. We're here to help.

About Fivetran and Looker

Fivetran is a Google Cloud Technology Partner that enables companies to centralize their data into Google BigQuery.
Google Cloud Technology Partner Looker provides a powerful data analytics platform that helps companies get real value from their data.

With Fivetran data connectors Label Insight brings all its data into a Google BigQuery database, creating a single source of truth for its CPG information, and uses Looker to run analytics on that data, driving value for corporate customers and end consumers.

Google Cloud Results

Brings together data from multiple sources into a single source of truth
Enables faster and more detailed analysis plus additional insights for clients
Used by 70% of the company’s team

200% ROI through freeing up 140 hours of manual reporting weekly

While using data more effectively is important for every business, at Label Insight using data more effectively is the business. The company is quickly becoming the go-to provider for product data; if consumers, retailers or regulators want to know what’s in a packaged food, drink, or personal care product, Label Insight is often the source. It captures information from product labeling, then creates more than 22,000 unique attributes per product, from micro and macronutrient content, allergens, and sustainability practices, to specialty diet eligibility.

Label Insight then works with retailers and Consumer Packaged Goods (CPG) manufacturers to help them gain deeper insights about their product set and inventory and meet the increasing consumer demand for transparency. Label Insight helps CPG brands participate in the SmartLabel transparency initiative, which provides detailed product information to consumers online.

“We can derive around 22,000 attributes from product data,” explains Jim Shedlick, Director of Architecture at Label Insight, “from whether something is paleo-diet approved to whether ethical claims can be justified.” With the company now holding data on over 400,000 consumer products, that’s a huge volume of information.

Lots of data, no warehouse

Although Label Insight had a lot of processes to collect, transform, digitize, and QA data, the company didn’t have a data warehouse. “We had a bunch of databases and people had to work directly in them to access product information,” Jim says. That could require looking in several different tools to gather the relevant information to respond to queries—a time-consuming process, which meant the company’s ability to scale depended on increasing its headcount of operational staff to answer questions.

The lack of integration also limited the company’s ability to analyze data across categories. Along with product information, Label Insight also captures a substantial volume of event data relating to how consumers access information—using apps, scanning QR codes, or simply searching online—and what kind of searches are most popular. That kind of data has the potential to provide valuable insight for retailers and manufacturers to improve on-package information or support targeted marketing.

The commercial choice

Jim and his colleagues were increasingly aware of these limitations—and particularly the missed commercial opportunities. Having identified more than 40 use cases for improved business intelligence and analytics, they decided to trial three cloud-based data warehouses, all of which appeared to offer the scalable, single platform that Label Insight needed. However, one stood out in several areas: Google BigQuery.

With Google BigQuery, Label Insight wouldn’t have to pay for data ingestion—which one of the rival offerings required. Another important differentiator was that Google BigQuery separates compute from storage—both technically, using different hard disk space within the cluster, and commercially.

“You pay for both storage and compute separately as you use them,” Jim confirms. “Storage is relatively low cost, and when we started, we didn’t really need the compute element; we were building up our use cases, our models, and dashboards. Once we had built up enough data in Google BigQuery to run useful models, we could start paying for compute.” As a result, the upfront costs were lower, reducing the barrier to entry.

Another key advantage for Jim was that Google BigQuery is a completely managed service, so there was no need to acquire additional servers or factor in IT management.

“Fivetran is really simple in a good way. They extract the data and load it into the warehouse in the format you want. We just trust them to do that.”

—Jim Shedlick, Director of Architecture, Label Insight

Seventeen sources collated in minutes

Another crucial factor in the decision to deploy Google BigQuery was the availability of Google Cloud partners Fivetran and Looker. Fivetran simplified the implementation, while Looker offered Label Insight the essential analytical capabilities.

Like Google BigQuery, Fivetran is also a fully managed service, which focuses on enabling data to be centralized into cloud data warehouses. It has developed over 100 connectors that gather data from different applications, file stores, databases, and event streams into a single central data warehouse—such as Google BigQuery - with minimal set-up. Invaluably, no transformation is required; Fivetran cleans and normalizes the data itself. The result: data can be collated from multiple sources in minutes.

For the complex landscape of Label Insight’s data, which was spread across 17 sources including Salesforce, HubSpot, and Zendesk, this ability was priceless. “Fivetran is really simple in a good way,” says Jim. “They extract the data and load it into the warehouse in the format you want. We just trust them to do that.”

“With Looker, we’ve been able to achieve far greater responsiveness to customer information requests. Instead of having to search separate tools, the team can now find answers in the data far more quickly.”

—Jim Shedlick, Director of Architecture, Label Insight

200% return on investment

With the data combined to form a single source of truth, Label Insight can then use Looker to analyze it—from running ad hoc queries to creating dashboards for different users. Looker has proved to be highly intuitive for the Label Insight team, who use it to quickly find product and manufacturer information from multiple sources.

Jim estimates that his colleagues have now set up over 100 Looker dashboards, providing regular updates on different aspects of Label Insight’s data and operation—from how many new products are waiting to be analyzed to how many views a particular manufacturer’s products have received on SmartLabel. Some are for internal monitoring, others designed to share with corporate customers. It’s also easy to drill down from the dashboards to access more detailed information about individual products or specific topics.

“We’re saving 140 hours of manual reporting time each week. We multiplied that by how much an hour costs the business and compared that cost to the total price of using Google BigQuery, Fivetran, and Looker. The return on investment was double what we’d spent.”

—Jim Shedlick, Director of Architecture, Label Insight

To measure the business value of introducing the new solution, Jim asked staff to report how much time they were saving by using the new solution. On average, two hours per customer per week were saved on completing internal process reports for customers and a further hour per customer per month on SmartLabel analytics reporting.

“In total, we’re saving 140 hours of manual reporting time each week. We multiplied that by how much an hour costs the business and compared that cost to the total price of using Google BigQuery, Fivetran, and Looker. The return on investment was double what we’d spent.” Instead of spending time on data management and administrative tasks, the team can now focus more on generating valuable insights for customers.

The strong return reflects the popularity of the solution within Label Insight. “Over 70% of our team/employees are using the data stack, running queries and viewing Looker dashboards, each week,” says Jim. Together, the combination of Google BigQuery, Fivetran, and Looker is helping Label Insight move to a culture where the whole company is able to perform, review, and benefit from rich data.

New possibilities from consolidated data

The initial returns don’t even include additional opportunities provided by the new solution, particularly in terms of incorporating event data. “With the help of Fivetran, we’ve created an event pipeline into Google BigQuery that allows us to compare and analyze events alongside other data. Fivetran has a connector that allows us to take all the event data directly from our web services into Google BigQuery as immutable event logs. We can then analyze those events in Looker like any other data.”

Data relating to over 400 million events is now stored in Google BigQuery—a figure that increases at a rate of a couple hundred events per minute. It is therefore a real benefit that Fivetran automatically adapts to changes in the source data. With this wealth of additional information at its fingertips, Label Insight is able to provide its customers with richer insights about the way consumers interact with the product information, opening up new use cases. “We can derive deeper insights, identify trends more quickly, and provide customers with talking points that help drive product development,” Jim says.

Working well together

The sheer volume of data is now prompting further exploration as to how to best manage and manipulate it, including taking a closer look at Google Google Cloud Dataflow. This is one of a handful of Google Cloud Services that Label Insight is considering, along with Cloud Vision API with its high-quality optical character recognition (OCR), which could accelerate the initial input of packaging data with accurate scanning and recognition of words on packages. “One of the advantages of consolidating data in Google BigQuery is that it’s easy to adopt other Google Cloud solutions as our business needs evolve,” Jim says.

Whatever direction Label Insight decides to take, Jim is confident that the support and service he will receive from Google and its partners will be a real asset. “They’re all great companies to work with,” he says. “They take our feedback, and they’re quick to respond. We have close relationships with Fivetran, Looker, and Google, and all three have spent time on site providing training and support. Best of all, they really work together.”

Tell us your challenge. We're here to help.

About Label Insight

Industries: Technology

Location: United States

About Fivetran and Looker

BigQuery Cloud Vision API