AmberStudent is a popular online booking platform for student accommodation. Students from anywhere can search and book student dorms and independent flats in the UK, Australia and other major countries when they move outside their countries for higher education.
As an entirely digital platform, AmberStudent uses a wide variety of tools that generate business-critical data, whether it is user or conversion tracking on their website, ads and other marketing campaign data, customer support data, etc. Siddhartha Agarwal was the first member of the data team, and since then, the team has grown to 7 people. When he joined AmberStudent, the data system within the organization was in a poor shape. The organization was not able to capitalize on the data to make data-driven decisions. Teams were struggling to compile and analyze their data. Siddhartha vowed to build a robust data infrastructure that would work reliably for their analytics needs.
Head of Data & Analytics
The Problem: Overloaded, fragile analytics
A year ago, AmberStudent was using Redash and Zoho Analytics for querying and dashboarding respectively. As the team needed to ingest quite a high volume of data, Redash kept crashing due to the high workload. This would impact the downstream workflows as well. Every few days, the entire analytics system would go down, leading to huge impacts on the real-time monitoring of business metrics. Besides the issues with the tools, the current system had all queries being written on the production databases, which was leading to overload and slowing down of their databases. The organization wished to make strong data-backed decisions, but these remained stalled or delayed due to these frustrating issues.
At this point, Siddhartha decided to make a shift to a modern data stack. He wished to be able to ingest and centralize data from a number of data sources to perform reliable analytics in a consistent fashion. Thus began the setup of Redshift and the search for a more reliable and manageable tool for ELT.
The Solution: A mature data stack with Hevo
Siddhartha evaluated multiple tools like Fivetran, Hevo, and Dataddo on his quest to find an effective and scalable ELT platform. Out of all these options, he found the balance to be quite right in Hevo- platform reliability, credibility, cost-effectiveness, connector availability, and all other aspects that were important for the team.
I would recommend Hevo for its ease-of-use, reliability, and cost-efficiency. Setting up pipelines is very easy, and the pipelines work consistently day in and day out. I’ve only faced one minor issue in that past year and that was resolved very quickly. The savings in expenditure on one full-time engineer, the accuracy of the data, the ease of building pipelines- all of these are great reasons to pick Hevo.
All of their data sources are currently being leveraged for data analytics, specifically for marketing analytics and customer support analytics.
We build analytics on top of all the data that we have. We don’t like to ignore any data- there are reports built on top of every data point, with over 100 Tableau dashboards currently in use. We also have some ML models like lead scoring models built on top of CRM data.
Now, the issues regarding reliability, scalability and ingestion frequency have been completely resolved, and the data team works speedily on advanced data requests from various teams in a much more streamlined way.
The Impact: High-Impact analytics for all teams
Now, the team at AmberStudent is able to save a significant amount of time in building and maintaining pipelines, saving the equivalent of one full-time engineer in terms of both costs and time. This frees up the team to be able to work on very high-impact projects, and the business impact of this has been huge. No business teams have to dig deep into data on their own- all the reports are generated very smoothly and quickly by the data team. They can focus on their own impact areas instead of worrying about the data and its accuracy.
The main use of the data being ingested into Redshift using Hevo Data is for marketing and customer support analytics. Some of the projects using the data from Hevo are very critical for them. An example is the cost monitoring dashboard, which ingests data from the warehouse and tells the team how each of the marketing campaigns are performing in terms of acquisition cost or conversion cost. They have also built a number of ML models which help them analyze and predict data in a more mature way. For example, the lead scoring model allows the sales team to make informed decisions about the leads they pursue.
Now, the data team is also working to set up a cutting-edge system to configure campaigns in an entirely automated manner. They wish to customize spends, reducing the amount of manual thought and effort going towards building effective marketing campaigns. This would also help save costs on platforms like Google Ads and bring the team into a level of effectiveness. Siddhartha predicts a cost-savings of 10-15% on paid marketing campaigns through this initiative.
The evolution of AmberStudent’s data maturity has been rapid and highly impactful, and Hevo Data has played a critical role in this journey!