Within the Analytics & Insights team, we have a vision for Analytics at AFG which is:
Everyone has the data and insights they need, at the time they need it.
Decision makers in AFG can readily access a large volume of information, but this information doesn’t always translate readily into insights. For us to successfully and sustainably grow our business, whilst adapting the challenges facing our industry, we need to be truly insight driven. That is, providing insight at the point of action and supporting decision-making at the right place and the right time, every day.
This extends to not just Executives and senior managers, but down to the operational level within AFG, and, importantly, to our brokers and partners.
AFG has had a long history of providing good reporting to the business to inform decision making, as well as providing a comprehensive suite of reports to our brokers and partners via AFG Analytics (in AFG Suite), custom burst reports and data extracts. But all of these aren’t always insights – there’s usually another step in the process to help decision makers (internal and external) answer the question of “so what?”.
So how do we evolve to not just provide what happened, but to start providing answers around why it happened, what could happen and what action should we take?
Over the last couple of months we’ve been working on enhancing our Analytics capability – particularly around self-service analytics and advanced analytics/data science. That has required us to move beyond the traditional enterprise data warehouse (EDW) that our existing BI/reporting is powered by, to a modern Analytics architecture that supports analysis, business hypothesis testing, development of predictive models/machine learning solutions, and enable collaboration between analysts and data scientists.
To that end, complimenting our Oracle EDW, we have built what we call an Exploratory Zone in AWS. This Exploratory Zone is more than just a data lake – it’s an environment that provides core data assets for analysts and the sandbox environments that they need to perform their analysis – as well as a workbench of visual discovery/dashboarding and data science tools.
Our new architecture
The diagram below describes our new architecture. The components in green are the existing capabilities that AFG have had for some time, and provide the reports and certain services that FLEX needs for our brokers, partners and staff.
The components in blue have been recently established and will provide the capabilities needed to support self-service analytics, delivery of machine learning solutions and on-boarding of citizen data scientists. For these we are using:
- AWS S3 as our data lake for source systems to be landed and staged, as well as replicating our EDW into S3 so that Analysts can access that data without impacting the EDW production environment.
- Snowflake as our structured database layer that allows analysts to access the curated data sets (including the replicated EDW) the Analytics & Insights team have made available as well as load additional data into their own user zones.
- A workbench that allows analysts multiple ways in which to interact with the data – either directly through Snowflake, or using data science tools such as R/RStudio and Python or GUI based tools (i.e. little to no code) such as Tableau and KNIME.
The components in grey (our production environment for machine learning solutions) will be built out over the coming weeks and months as we continue to build our data science and advanced analytics capability.
Staying up to date
We’ll keep providing updates on the enhancements we’re making – and how that relates in particular to our customers, brokers and partners. But if you want to reach out at any time, do not hesitate to drop me a note.