What makes Callahan such a unique digital marketing agency? They start with front-end data analysis to inform client strategy—in other words, gathering as much data as possible (far beyond what would be considered standard marketing sources) to understand the client’s baseline business and marketing operations. The goal is pinpointing exactly wherein lies the biggest opportunity instead of simply running a marketing campaign and measuring its impact retroactively. As part of this front-end analysis, Callahan also looks at the historical trajectory of the business in order to predict how it would fare if nothing changed. That helps Callahan set realistic benchmarks and later conclude if they’ve truly made an impact on revenue or market share.In short, Callahan works with a lot of data—and that’s only the beginning. After completing its front-end analysis, the agency is constantly analyzing and adjusting client spend in order to optimize each campaign. At their core, Callahan is an agency of data people that strive to understand the “how” and “why” of each client’s marketing. But that doesn’t make them data management experts. They want to focus on insights, not on the technicalities of managing data pipelines.
Three Key Challenges of Working with Marketing Data
Collecting and preparing marketing data can push most analysts to the brink of frustration. For Callahan, their biggest pain points in this process included:
- Transferring lots of different client data sources—this includes SAP, LinkedIn, Shopify, Facebook and Salesforce, to name just a few.
- Preparing and integrating these datasets in unique ways before they could be used or reported on.
- Creating a singular and trusted view of the data so that all analysts had the same information.
Working with demanding clients only exasperated these challenges. Callahan needed to be able to understand and communicate the results of their marketing campaigns fast—or their client base would be in jeopardy. And it’d be one thing if Callahan could repeat the same data transfer and preparation process for all clients, improving upon the same steps each time. But each new client came to the table with different challenges and different data, which means Callahan needed to be able to think on their feet and onboard new data as quickly as possible.
No Data Engineers, No Problem: An Automated Cloud Data Warehouse Solution
Without data engineers on deck to custom-build data pipelines, Callahan assembled a modern cloud data warehouse stack built upon Google Cloud Platform (GCP) that would enable self-service. All but one of the technologies used are software as a service (SaaS), which meant Callahan got up and running in minutes—there was nothing to install or configure. “This technology structure has fundamentally changed the way we do business, and it has given us a competitive advantage.”—Zack Pike, VP Data Strategy & Marketing Analytics, CallahanAs expected, what most impacted Callahan’s efficiency were the two technologies that automated the process of transferring and preparing data: Fivetran and Cloud Dataprep by Trifacta, respectively. Zack Pike, VP Data Strategy & Marketing Analytics at Callahan, explains how his team uses the two technologies: “Fivetran is great for a lot of our marketing data like Facebook Ads, Google Analytics or Marketo. The great thing about Fivetran is that I can set up a feed of data and it runs with zero maintenance. Cloud Dataprep by Trifacta is what we use for our messy data that’s coming in because it offers us a lot of benefit in cleaning that data up before it goes into the database. Sometimes we clean data before it goes in, and sometimes we push in as is and then clean it once it’s in the database.” The only technology in Callahan’s stack that is not cloud-based is Tableau. Due to the wide familiarity of Tableau amongst the team, it was an easy choice for visualization. However, since Tableau must be installed on a desktop, it has prevented the team from being able to operate entirely from anywhere or from any machine. Callahan is now exploring new BI solutions such as Looker to fully transition to a SaaS-based solution without on-premise or product installation dependencies.
A Closer Look at the Automated Data Warehouse in Action
A typical Callahan use case might look something like this: Analysts are trying to determine the variance in impact of a specific media campaign in different geographical markets across the U.S. To do so, they want to take a closer look at media spend as a percentage of revenue generated by using Facebook Ad data and sales data. Analysts start by pulling in Facebook Ad data with Fivetran. Again, this is a seamless process using Fivetran connectors—there’s no need for data ingestion building or maintenance. Once the ad data has been loaded into the landing zone of data warehouse BigQuery, it’s time to start on the sales data. For this particular analysis, the sales data needs to be cleaned using Google Cloud Dataprep by Trifacta—the dataset includes values of total sales for each region, which should be removed, along with any null values and state names need to be simplified to match their corresponding names in the Facebook Ad data. Designer Cloud is powered by machine learning, which means it will already suggest the user to make these transformations.After the sales data has been cleaned, the two datasets will be joined together. In Designer Cloud, it’s easy to select which columns should be turned off from each table. For example, Callahan might want to remove “Date” but leave “Impressions,” “Reach,” and “Spend.” Finally, the completed table can be published to the production zone of BigQuery.At this point, Callahan can use the same recipe it created in Designer Cloud to run the same necessary data preparation steps for similar incoming data. All analysts have to do is monitor the results to ensure no new issues arise. Should they appear, it would only take a few clicks to open up the dataset in Designer Cloud and make the corrections to get the job back on track. To see the above use case and a typical automated cloud data warehouse in action, watch the full demo here:
What’s Changed? The Impact of an Automated Data Warehouse
Upon implementing its automated cloud data warehouse solution, Callahan has seen tremendous results. Most notably, it has allowed Callahan data analysts to spend the majority of their time (70%) on data analysis by reducing time spent building data pipelines. This means that Callahan analysts get to do what they do best—analyzing data, not managing or maintaining it. With more time dedicated toward data analysis, Callahan clients have reaped huge benefits as well:
- Client A grew its media impact by 90% while reducing its media budget in half
Callahan was able to identify the waste in their media spend while simultaneously more efficiently running the ads that were working.
- Client B saw a 2X increase in customers on a 60% reduction of leads purchased
Callahan culled through all of the leads that were not converting for the business, which vastly reduced the number of leads purchased. However, the smaller list of leads that the business was left with allowed them to focus more attention on the leads that mattered, and convert those leads more readily.
- Client C saw a 5% sales improvement during peak periods
For clients selling into big retailers like Walmart or Home Depot, managing inventory is critical. By alerting the client three weeks ahead of time that, under its current plan, it wouldn’t be supplying the right amounts of product to retailers, the client was able to adjust its shipments. The result? The client saw a 5% improvement in sales during a peak sales period when its performance mattered most.
To watch the full webinar where Callahan explains why they built a modern cloud data warehouse stack and demo it in action, click here. Schedule a demo to get a personalized look at the Designer Cloud data preparation platform OR try it out for yourself for free.