Grip uses Databricks Data Intelligence Platform to provide smart shipping solutions
Reduction in damage rates
Reduction in shipping costs seen by customers
Increase in exploratory productivity time-to-insights
“Having our data stored in the same format under the same hood that powers some of the modeling and pipelines accessed through the warehouse — you can’t ask for anything better when it comes to governance, management and usability.”
E-commerce companies can ship perishable goods anywhere in the U.S. if they have a good handle on logistics. Powered by data, that’s what Grip helps customers do, processing hundreds of thousands of orders through its platform each month. Grip serves companies that ship meal kits, frozen bread and pasta, pastries, pet food, flowers, and other perishable items across the country by processing and interpreting a variety of data points to make the most effective shipping recommendations. Customers who use the Grip platform see a 25% reduction in damage rates and a 30% reduction in shipping costs. Data helps Grip make the best recommendations on how customers can ship their goods, and from day one the company adopted Databricks to help.
Consolidating data using the Databricks Data Intelligence Platform
Using Databricks, Grip processes hundreds of thousands of orders through its platform each month and anywhere from 10 to 20 times that amount of data in the form of tracking records from shipping carriers. The Grip platform integrates with weather providers, so forecasting is processed through the platform. The platform also tracks customer feedback — including in previous records of shipments to a particular area of the country.
Grip’s data comes from many places, including Shopify, ShipStation, different warehouse management systems, APIs for weather data, carrier pricing and delivery time tracking, and customer support systems like Zendesk and Dynamics. Once processed, the data empowers Grip to suggest which carrier to choose, as well as the ideal refrigerant and insulation, packaging and material, and other shipping logistics. “The more data we have about shipments, the better we can predict the probability of delivery success,” says Jimmy Cooper, Co-founder and CTO at Grip. “We can then dynamically suggest, ‘I know we told you to use A last time, but now you should use B because you’ll have a higher chance of delivering the product successfully.’ Providing visibility into the areas of the country where customer orders are performing well is valuable.”
Once a delivery is made, Grip publishes analytics through tables accessible via Databricks SQL Serverless so customers can see orders that went out, where they went, areas of the country that are bottlenecked, and areas of the country that are performing well. This helps Grip customers discover how they might improve. Additionally, Grip utilizes Delta Lake and Databricks SQL Serverless to drive analytics, which allows it to share data with customers more regularly. “We originally had a report that refreshed once a day for customers, but after making a few optimizations based on already existing functions, we now have it refreshing close to real time,” says Cooper.
Improving customer service with Databricks
Databricks makes it easy for Grip to do everything with one set of tools. The Grip team can build machine learning models and data engineering pipelines end to end, serve end users with analytics, and track internal analytics — all within one platform. “Having our data stored in the same format under the same hood that powers some of the modeling and pipelines accessed through the warehouse – you can’t ask for anything better when it comes to governance, management and usability,” says Cooper. “We have data coming in from different sources. It lands in our data lake and then we leverage Auto Loader to push it into the Delta tables. We like the Bronze, Silver and Gold structure — anything that goes into Bronze is raw data. As it moves through the ecosystem, we’re able to incrementally clean and prepare it in a way that makes sense for consumption.”
Grip’s pipelines are entirely orchestrated through Databricks Workflows. Each of its jobs has dependencies, so being able to manage everything in one place makes troubleshooting easy when something goes wrong. The alerting integration is tied into the company’s teams, so they can rerun or repair run functionality before it becomes problematic to the customer. Unity Catalog allows Grip to point and route different environments so customers have secure access to the right data sets. Schema drift enables its team to easily enforce, change or merge schemas without breaking any downstream processes. Additionally, Grip reports the developer experience of working with Delta Lake is easy. “People on my team that had not been exposed to the format or Databricks before were able to pick it up quickly because it’s well documented, intuitive and supports the languages we interface with,” says Cooper.
Partnering with Databricks keeps cost low
With Databricks Workflows, shared clusters can be created and repurposed across different tasks within the same job, saving Grip money and time. “It’s easy to spin up a cluster once, reuse it for all of the different steps and spin it down when you’re done,” says Cooper. “We have granular control over how much we’re spending and can compute costs as a part of the jobs that are running.”
Databricks SQL Serverless has been a game changer for cost as well. According to Cooper, adopting Databricks SQL Serverless decreased Grip’s customer and internal dashboard update refresh time by 50% and increased its exploratory productivity time-to-insights by 50%. “To know we don’t need a warehouse running unless it’s actively being used is great, and the speed it starts up from being completely off to executing a query is less than 20 seconds,” says Cooper.
Ultimately, the robustness of being able to manage data at a relatively low cost makes using Databricks the best choice for Grip to iterate, move fast, incorporate new features and serve customers better.