![Aviatrix case study - hero image](http://images.ctfassets.net/tg3z0zgy3dgv/nCNghP61am6J4PRv7PSor/792d7fda1db618e86e1e929f979eb89c/case-study-aviatrix.png)
Editorial note: After a three-week migration to Omni, the Aviatrix team consolidated their Looker, Tableau, and Excel use cases and doubled BI adoption. Here’s how they did it.
Cloud networks have clear advantages over the on-premise solutions they replaced. But widespread cloud adoption has also introduced new complexity, cost, security, and compliance challenges.
Aviatrix consolidates cloud networking, management, automation, and security into a single solution to help enterprise customers save as much as $25 million a year in costs and simplify cloud operations.
“We call our platform the ‘network of one’ because it provides customers with the portability, liberty, and full control of their network that’s required in this multicloud world,” says Cody Pulliam, Aviatrix’s Senior Manager of Business Analytics.
At Aviatrix, Cody leads a small Data team that enables each function to access, explore, and use data. The team previously used Looker and Tableau for different use cases, which resulted in duplicate work and inconsistent results (and still didn’t meet all their needs).
“Compared to Looker and Tableau, Omni was by far the best performer in terms of all-around functionality.” Cody Pulliam, Senior Manager of Business Analytics
By switching to Omni for business intelligence, Cody and his team have:
Consolidated three tools into one: Omni gives Aviatrix the functionality of both Looker and Tableau and reduces the need to export data into Excel for analysis.
Migrated in less than 3 weeks: With support from Omni and Medusa Analytics, the Data team was able to migrate a complex data model and key content quickly.
Increased BI adoption by 2x: Due to increased usability, flexibility, and onboarding support, twice as many Aviatrix employees have full access to features in Omni compared to Looker.
Improved data quality and trust: Migrating to Omni initiated a clean-up of outdated and unnecessary content and allowed teams to customize their workbooks and dashboards without impacting the shared data model.
Saved time for every team: Omni’s intelligent cache makes it possible to run the same queries faster and made interactive exploration intuitive for a wider group of business decision makers without data expertise.
Significantly reduced BI spend: By consolidating tools, switching to Omni freed up budget for new initiatives.
Key elements of Aviatrix's data stack #
![Aviatrix case study - data stack](https://images.ctfassets.net/tg3z0zgy3dgv/4nCbVS8RECOkOwfe8DyUSG/3c7e5cc9cbc4fbbb8b3b2d557270aa08/aviatrix-stack.png?w=1600&fm=webp)
Challenges that multiple BI tools couldn’t solve #
The Data team at Aviatrix has only three members and a lot of ground to cover. “I lead our team and work closely with operations within every function. Then, we have one person in charge of analytics for anything pre-sales, and the other in charge of anything post-sales. Teams within those two groups tend to have different data repositories and ways of logging and managing their data, so it helps to split up the work that way.”
To serve each function’s use cases, Cody and his team historically used both Looker and Tableau. Looker was the company's main BI tool, but it was often slow, hard to use, and prone to the creation of redundant dashboards and data models.
“Our IT team had to do a lot of upfront work before people could interact with the data in Looker — managing ETL, putting snapshots into our data warehouse, and doing all the work on joins and relationships between tables. While they did their best to manage a centralized data model, teams had different practices that created inconsistencies,” explains Cody.
“To combine things like telemetry (product usage) data with Salesforce data, you have to build your own mapping files to join between two disparate systems, and how people did these joins and relationships between tables varied significantly. For example, Sales would try to join the data based on the domain, like nike.com. But Finance would do the join based on the billing account number. Both of these fields existed in the telemetry data and in Salesforce, but it wasn’t standardized.”
Looker also limited how business users could interact with data. “Most reports and dashboards were actually tables, not charts and visuals. People would create these tables, export them to Excel, and then add functions and hard-coded targets on top. BI tools are not supposed to be used like that. They’re supposed to be data visualization tools where people can access up-to-date reports and dashboards by clicking on a link, and explore the data from there. But that’s not how people were using Looker.”
“We needed a faster way to query data because Looker was slow and we needed a drag-and-drop UI to make our end users comfortable building new analyses.” Cody Pulliam, Senior Manager of Business Analytics
Eventually, Cody started using Tableau to support certain use cases that Looker didn’t serve well. “The Finance team came to us with a lot of ad-hoc requests that required very quick analyses or the creation of complex calculated fields, so we added an instance of Tableau to the mix. There were cases where we built something that would have been useful to the broader team, but we couldn’t share it because most people didn’t have accounts in Tableau. Then we’d have to recreate it in Looker.”
But switching everyone to Tableau wasn’t the answer. While it would have improved access for business users in some ways, Tableau wasn’t powerful enough to serve all the use cases — especially IT and Engineering, who preferred a programmatic approach to working with the data. “I wasn’t a fan of Tableau’s web client because it was extremely slow — even just adding another field from an existing data model could take a very long time. But you needed a creative license to get the faster desktop version.”
With neither platform a great option on its own nor together, the team reevaluated the goals for Aviatrix’s BI infrastructure and set out to find one platform that could:
Run faster queries against their database
Make it easy to create and manage metric definitions across functions
Allow all users to explore accurate data with drag-and-drop functionality and Excel-style syntax
Evaluating tools for all-in-one functionality and speed #
Cody was already familiar with many BI tools in the market and was able to disqualify them immediately. “In the past, I’ve used the full SAP suite, but it would have required a full overhaul to migrate to their BI solution because we didn’t have the other SAP products in our tech stack. I’ve used PowerBI, and I was not happy with how slow the visualizations were even for small amounts of data. I’ve also tried Qlik, which caused similar issues to Looker. None of them were a good option for us.”
One of Cody’s main tests in the evaluation process determined how quickly a tool could execute queries compared to Looker and Tableau. He was also looking for a tool that would give the team the ability to quickly create a model from scratch without needing to know SQL or a tool-specific syntax, but had a back-end structure and built-in semantic layer so that IT could still build more complex models with code.
“I used a specific query to test how quickly a BI tool can execute against our database, and Omni performed really well compared to Tableau and Looker. In Tableau, I would have to schedule the extract to get faster performance, but Omni always has the live version. It’s so nice that Omni’s caching doesn't have to execute a query every single time someone clicks something on a dashboard. Looker failed there too and it bogged down the user experience. People just had to wait forever for things to load.” Cody Pulliam, Senior Manager of Business Analytics
Along with its performance and modeling capabilities, Omni’s Calculations feature sealed the deal. “It was great to know that Omni already had Excel-type functionality where a user could do things like cell references — multiplying column A by Column B — and it would carry down automatically. It’s something that’s really valuable to our end users who previously had to export data from our BI tools and then work with it in Excel.”
The three-week migration #
Cody partnered with Jack on Omni’s Solutions Engineering team and Andrew Searson from Medusa Analytics to speed up the migration and ensure that users had time to learn the new platform. Despite having a complex model, a large backlog of content, and more than 100 users, the Aviatrix team was able to complete the migration in just three weeks.
“We split up the work between Andrew’s team, my team, and Omni’s team, and migrated 15 dashboards and reports from Looker that were used consistently. We realized that the rest were really just different variations of those same dashboards, so the migration was a good opportunity for us to clean house,” says Cody.
“Andrew had a lot of experience with both tools and helped translate between LookML and Omni. He also helped us identify duplicates — like cases where a field was created twice and used in two different dashboards — so we didn’t have to migrate both. If IT had done it, it would have just been a 1:1 migration and we would have lost those additional optimizations.”
In tandem, Jack led an initial training for all users to give them a basic understanding of Omni. “Jack did a great job demonstrating the similarities between Looker and Omni, and the incredible new functionality that we’d also get from Omni.”
Afterward, Cody led a follow-up training for developers and analysts specific to their data and use cases, including the actual business questions, topics, and data models they’d be working with. Cody and the team then serialized these training sessions for everyone with a program they called FIDS (fast insights and data solutions, a nod to Aviatrix’s flight-inspired name and the Flight Information Display Systems that you see in airports).
“These were 30-minute sessions with specific functions. We’d go through a few questions they had and then help them do it themselves. These could be really simple questions like: What are sales by month? Or even: How many customers do we have? After a quick demo, users were the ones working the data. Rather than just showing them, we were coaching them through it, and that was a great way for people to learn.”
Omni’s early impact #
With Omni, each team at Aviatrix can have separate workbooks and run custom analyses without impacting the shared model — something impossible with Looker. “It’s a huge advantage that edits to the shared model flow down to the workbook level but not vice versa,” says Cody. “Before, when you changed or added a field in Looker, you’d have to make the update in two places. It’s a common use case for us, so I love that Omni works this way.”
Team-specific dashboards
The point-and-click UI and the addition of customized workspaces for specific teams has increased the number of people at Aviatrix using BI as well as the depth of their usage. For example, a support manager can use a support-specific dashboard curated with filters and metrics that are most relevant to them, so it’s easier to run complex analyses using just the UI.
“There’s a lot more people drilling down into the data and building their own visuals. That’s something people were hesitant to do in Looker, so they would just download the table and do it themselves in Excel,” says Cody.
“I’m a huge fan of Omni’s product. I’m constantly recommending it to others. When someone asks me what I’m up to at Aviatrix, I say, ‘Let me tell you about Omni. It’s the great things about Tableau, Looker, and Excel all rolled into one.’” Cody Pulliam, Senior Manager of Business Analytics
The pre-filtered dashboards have also made it easier for leadership to stay updated on business performance, without the business analytics or IT teams needing to prepare beforehand. “We have a leadership dashboard that we present in weekly business reviews with our CEO. Executives can also go to the dashboards tailored to their team without being bombarded by a bunch of irrelevant data points.”
Revenue projections
Cody also plays an important role in managing the Finance team's revenue forecast model and that has been made easier by Omni’s Calculations feature. “I run Python scripts in the back-end to populate tables that get consumed into an Omni topic and joined with other tables that show things like the account name, industry, demographic, etc. If the Finance team wants to see the telemetry data, they can just drag in those fields. It’s great that they can do the rudimentary functions — like converting MRR to ARR — in Omni without needing to export,” says Cody.
Long-term learning
After a fast migration to Omni, the Aviatrix team continues to identify new opportunities to increase adoption and expand use cases even more. When new features and resources are released, Cody and his team also incorporate those into the team’s workflows. Cody gives an example. “Omni recently released a community page where users can give and get advice, and it’s ended up being an important cheat sheet for us. I love that Omni is thinking ahead about what users need. It came in at a perfect time for us.”