Serenata Flowers is the largest independent flower company in the UK. From their start in 2003, they have had a culture of leveraging data heavily. They became an entirely online business in 2005. With a founding team drawn from finance and data backgrounds, Serenata developed an institutional skill at SEO, Display and Pay-per-click advertising early on. They have studied clickstream data and used it to improve their ability to drive paid acquisition channels.
Unlike their competitors at the time, they brought everything in-house including design and the checkout process. This is reflected in the highly technical composition of the team – 20 of their 67 employees are in IT.
They have recently launched in France, and are looking to continue to expand internationally.
Having been in business for over a decade, Serenata has gone through a succession of traditional BI tools.
Before finding Metabase, they used a combination of Looker and home-rolled administration web pages. However, these were very complicated and limited wider adoption within the company. They were primarily used by upper management looking at sales reporting.
Serenata added Metabase to their data stack after a colleague showed Martin, Serenata’s Managing Director, an instance he spun up to play with data on BigQuery. The team liked it, and decided to put it into production. The initial goal was to get the data collected by the team in more people’s hands.
A major attraction of Metabase was how quickly they were able to get up and running with Metabase. Additionally, the fact that it was open source was also appealing.
Since adopting Metabase, Serenata has also been able to stop developing additional reports in their admin system. It is far faster, and hence cheaper, to write a question in Metabase than to have an engineer extend the admin system.
A major consequence of how easy Metabase makes asking questions about their data is that more questions get asked, even if there isn’t a clear ROI. This has allowed them to do more free form exploration and ask questions without a rock solid expectation of an outcome. The low cost of each incremental query allows for this kind of lightweight and agile analytics setup.
A wide variety of data sources end up in the Serenata data warehouse. Application databases and behavioural events through Snowplow are piped into the data warehouse. Additionally, a wide variety of third-party data sources, like AdWords, Bing Ads, Facebook Ads, Fifo, and reviews are ingested as well. Serenata have used a number of different underlying data warehouse systems. Currently the primary data warehouse is BigQuery, a move from their previous AWS Redshift cluster.
Serenata’s current overall data structure has lasted for the past three years, and they have a team of engineers and DBAs who keep it optimized for the various analytics workloads needed of it.
Because they already had an existing analytics system setup, once past initial impressions, Serenata ported their old queries. They started by porting all of their LookML queries to SQL, and then copied those over into SQL queries in Metabase.
Now that Serenata is using Metabase, everyone in the company, including accounting, operations and customer service has access and is able to use it. Serenata is also using Metabase dashboards on large format screens in the office, allowing more people to keep track of company data throughout the day.
While report creation is mainly in the hands of a dozen analysts, DBAs and engineers, most people find it simple to modify existing reports to meet their needs. Metabase SQL templates are used heavily to allow end users to perform some limited filtering on their own.
While getting their old reports ported over to Metabase and in everyone’s hands was surprisingly quick and painless, the widespread adoption of Metabase and the ease of creating new reports has lead to a number of unexpected findings at Serenata.
Like all e-commerce companies, Serenata is a target for fraud. One advantage of the number of people looking at reports and trends is that anomalies can be detected faster. By putting these trends in front of employees with different functions, Serenata is able to take advantage of their employees’ specific knowledge of what trends are unusual. This has given them an additional edge on top of their existing mechanisms for spotting anomalous transactions.
Additionally, the lightweight report creation has allowed management to spot operational inefficiencies. For example, the ability to examine the entire unified dataset and slice it by customer service representative allowed them to identify high performing reps and take their methods and use them to improve the performance of the customer service team as a whole.