At PenPath, we’re committed to helping our clients extract valuable insights from their marketing data to help them make better decisions that positively impact their bottom lines. However, accomplishing this is not easy. We tried several data warehouse alternatives, and we realized we needed a solution that’s reliable, scalable, easy to integrate with our existing technologies, and cost effective.
As the Head of Product, I learned that different data management approaches have limitations and that not all data warehouses are created equal. But Snowflake met all of our criteria, and it has changed everything—not just for us but for our clients, too.
What we did before
Initially, we wanted one database per client where we could access all data in a similar area, essentially creating a data warehouse. We tried a couple of different solutions before choosing Snowflake.
At first, we utilized Amazon RDS for PostgreSQL to create a single database per client, but that involved some pretty elaborate steps. The approach became quite expensive because we had to bring on an additional resource to handle the different scripts, cleanup steps, and extensive maintenance and permissions administration that were required.
Next, we explored Google BigQuery for our data warehousing needs. This was better because it was cost-effective with a good web UI, but it had its own SQL and an API we found cumbersome.
The drawbacks eventually outweighed the advantages. We also had clients already using PostgreSQL-style databases. Therefore, our existing models had to be modified to work within Google BigQuery, which meant we had to maintain different versions of our data models. This took too much time and energy. Recognizing this, we went searching—again—for a better solution.
Why we switched to Snowflake
About a year ago, we started using Snowflake because of its potential to solve the issues we had with Amazon and BigQuery. It was an easy transition. The web UI made setup easy, and I was comfortable creating users, roles, and permissions. Immediately, we had complete confidence in our data governance. We started with a month-to-month usage plan for testing. It worked well, so we now have a contract on a larger capacity plan.
We’ve now implemented Snowflake across all of our clients and reporting, and we’ve seen a lot of advantages:
• Easy setup
• Easy scaling
• Familiar SQL
• Simple permissions administration for as many accounts as we want
• Easy connections to external data sources
The value to our clients
Since switching to Snowflake, we have seen a 67% reduction in time needed to go from data to dashboard. For our clients, this means more immediate access to insights. In addition, our clients have increased confidence in the accuracy and freshness of their data.
For example, we now can leverage hourly advertising data. Working with one of our clients, we created a data model that combined several advertising data sources into one data set. Fivetran enabled us to load data from various paid search providers into our data warehouse every hour. Then we built a data model in Snowflake to blend the data into a powerful resource. The client in this example now uses that process to automate bidding strategies for search advertising.
The benefits to us
We’ve seen dramatic improvements including faster query speeds, significantly fewer failed refreshes in Tableau, and a vast reduction in warehouse management hours. With that kind of power and efficiency, we’re providing more services while ensuring security and reducing costs.
With Snowflake, we are better able to understand the query cost, data load cost, and Tableau cost for each client. As a result, we know which clients are using more resources. This insight has helped us optimize our data models, ELT strategies, and Tableau refresh cycles. Overall, the huge advantage of using Snowflake is how it informs our pricing and how we forecast costs for our clients.
What happens next
Now that we have Snowflake set up, we’re ready to evolve our offerings and enhance our services. Snowflake is the springboard for building more advanced data science models, and we’re able to iterate faster knowing we have the required horsepower. Additionally, better data management means better resource management on our end.
One great example is our new sentiment analysis capability. We’re ingesting data from more than 10 different data sources, combining it in Snowflake, and then running our machine learning models on that data daily.
Each day, we’re finding ways to make more insightful analytics tools for our clients. And with the help of Snowflake, we’re doing so in a secure and cost-effective way, making our work easier and helping our clients get better results, faster.
Learn more about gaining value from your marketing data with Snowflake.