Welcome to our new blog series, INSIDE RUMPL. We’re going to give you a behind the scenes look at the problems we’re facing, the solutions we’re building, and the team that makes it all happen. We'll share some tips and tricks on what makes each Rumpl department tick. For the first installment we’re going to introduce you to the Analytics Department at Rumpl and how we collect and utilize data to run the business and make informed decisions. Our Analytics Manager, Devon Rogers, joined the team in July 2019 and has been hard at work implementing the necessary infrastructure to set Rumpl up for long-term success.

Why Hire An Analytics Manager at a 15 Person Company?

For the past five years, all reporting and analytics has been managed by our Head of Operations, Patrick. We’ve relied on traditional tools and in-platform reporting for our insights. While these tools work, the data is siloed and often has to be manually updated. To better equip Rumpl for the present and the future, we needed all of our systems to talk to each other and to have a team member own this function and take it to the next level. That’s where Devon comes in!

Portrait photo of Devon Rogers, Rumpl Analytics Manager

Our Analytics Manager, Devon. No, he's not a cop.

What To Look For When Hiring Your First Analytics Manager

We decided to hire an Analytics Manager to own the function of connecting all of our data and bringing our insights to the next level. What to look for? Devon brings the perfect background to the table. He’s extremely versed in SQL to derive insights from a data warehouse, he’s skilled in Python and other languages, and he has a knack for learning and figuring things out on the fly.

What To Build and How To Build It

When deciding what data infrastructure to build, it's important to map out all of the data sources and services being used. Here are some basic questions to evaluate a source’s accessibility and importance:

  • Is this software a widely used service?
  • Does the software have an API / extract that you can interact with?
  • Does the software contain information pertinent to decision making?

Once scope is defined, creating a budget comes next. Data solutions rarely come cheap–– the cost is a combination of time and financing. Evaluating need versus nice-to-have is critical.

Steps: 

  1. Decide whether to invest in an external ETL solution, or customize in-house.
  2. Determine a warehousing solution that scales to the volume of data you plan to work with (highly recommend cloud hosting)
  3. Choose an analytics platform to extract and manipulate all of your data.

Devon working with Shane, our Ecom Manager, reviewing some reporting.

We knew where all of our data was housed (in-platform), so we identified ETLs that were able to directly plug in to our existing software.

Data and Analytics Tools We Use

Redshift - Warehouse
We decided to store with Redshift, a data warehouse hosted by Amazon Web Services, because it’s the market leading solution and our team has previous experience.

S3 - Cloud Object Storage
We opted to use AWS Simple Storage Service (S3), given our choice of Redshift for our data warehouse. We utilize this service to load any data we may otherwise not have an ETL pipeline available.

Segment - Customer Data Infrastructure
A service application that handles ETL and data governance.

Stitch - ETL
It serves as our backup ETL service, and a catch-all for anything we aren’t trafficking through Segment. The platform is very clean, with minimal effort required for setup. Stitch can also setup a data sync through Google Sheets. So great!

Mode Analytics - The Best “Analyst First” Tool
Mode was selected given the team’s experience with the platform at previous companies, its robust support for Python and R, integrations with external services, and customization. The platform is also friendly to non-technical analysts.

Python - Language
Python serves a big purpose at Rumpl. At any given time, we may need to clean data for loading into S3, or require a complex, ad hoc analysis. The community support for key libraries, in all levels of analysis, is hard to parallel.

Q&A with Devon

What is the coolest thing you’ve learned about Rumpl from the data?
Articles written 4 years ago still get people engaged on site - that speaks to the Rumpl brand. I’m a sucker for getting a kick from that.

Where do you read about analytics?
YOU CANalytics is definitely worth a once over, regardless of industry. Kaggle competitions are great when you are looking for motivation, even at a glance. Otherwise, it's really peppering Towards Data Science / Hackernoon for those nuggets of value.

What’s the best way to teach yourself a new language or process?
If you want to learn something and retain it, nothing beats practice. But I think I can give more than that - do your research. Find resources that you trust after taking some time to see what is comparable online. I am an advocate of social and visual learning. It quickly becomes clear what I don’t know once I’m talking with co-workers or mapping things out on a whiteboard.

What do you like to do when you’re not working?
I like to make my way into the forests of the PNW. Most of the time it is in hopes of finding some choice culinary mushrooms (pics below), but I’ll take a scenic hike any fair weather weekend.

What’s your favorite Rumpl product?
There’s no doubt - I will forever be rooting for NanoLoft® Cosmic Soul

What music are you listening to on Spotify right now?
Space Trap, Progressive Trance, Psychedelic Rock (Think Tame Impala, and Deep Sub Bass)