Join Jorge Beleña, Sr. Data Engineer at Vista DnA, as he sheds light on how he and his team leveraged data to help marketing teams create personalized ads for Vista customers.
Marketing isn’t what it used to be. The last few decades have witnessed a switch from broadcast ads to efficiently curated personalized ads. In addition, Google and Facebook Ad tools have made it easier for small businesses to ramp up on digital marketing campaigns with just a few clicks.
However, it has become much harder for larger companies to run campaigns as they might advertise differently to distinct buyer personas across various channels. In that case, ensuring the buyer gets consistent information across the various channels becomes a key challenge, especially when it is critical to ensure user privacy and consent at all times.
As a global personalization platform, at Vista, we need to deliver targeted ads to our customers through different media such as paid search or email marketing. To do so, our marketing teams are split based on the locale and the specific channels, facing the same problem. This encouraged us to address the challenge of delivering hyper-personalized yet consistent ads in our way.
Our goal was to help and support marketing teams to build audiences for omnichannel activation. We achieved that by designing a proprietary set of tools and automation that they can leverage for usage in future campaigns.
To do so, we bet on a customer-centric marketing strategy and the Audience-based Marketing and Activation Team (AbMA) to make the business reliable, ethical, and adapted to customers’ needs. We used algorithmic tools powered by first party data for more accurate and customized ads that reach the customer at the right moment. Here’s what we learned.
Challenges in Traditional Communication
To get a deeper insight into the challenges we were faced with, here’s a comprehensive list of the problems we intended to target with the new method:
- Negative Customer Experience: Since the data is not unified across channels and levels of marketing, the same customer might receive conflicting marketing messages across different channels, leading to customer dissatisfaction.
- Inter-Departmental Silos: The traditional communication approach isolates campaigns with respect to each other and makes it impossible or too tedious for marketers to understand what other campaigns are currently in flight in other parts of the company.
- Speed: As teams work on separate levels, there is a high dependency on communication within the teams. This increases the time to market and affects the campaign conversion rates.
- Inefficiency: Approaching customers with inaccurate analysis leads companies to invest in inefficient marketing, spending more money without achieving goals.
Audience Passer Tools
Two years ago, the data engineering team at Vista dedicated intensive efforts to creating ad-hoc audiences based on marketing requests. This exercise included large-scale research, manual data manipulation, extracting and uploading files to individual marketing channels, and more. In many cases, it was the same request iterated over a list of countries, product categories, or time frames.
1. Audience Generation
We created an automated process to programmatically define our audiences based on several input parameters, such as the country, category, or target average order value. Then, marketing specialists put this information together, with DnA analysts providing valuable insights. This process uses dbt as the transformation tool and Snowflake as our data warehouse containing the transactional information of the customers or the site tracking data.
Other common audiences are:
- Inactive users in the last year
- Registered non-buyers
- Customer industry verticals
- Custom test/control splits
2. Centralized Audience Manager
Once these audiences were stored and refreshed in our database, we had to activate them in our marketing partners, such as Google Ads, Facebook, Pinterest, or Bing.
Data engineers in the team helped create an abstraction layer on top of these partner APIs so that analysts or marketers didn’t have to deal with the burden of navigating and learning about each endpoint or handling authentication tokens.
By creating this wrapper on top of the APIs, we created an audience in Google Ads and Facebook by following a specific syntax.
This was the foundational work to provide a self-service platform for marketers. From consistent audience definition, tracking, and measurement to storage of metadata and interoperability with other marketing assets, the platform keeps marketers informed through a single UI. We also started a long-term collaboration with Segment to gain speed and scalability on the activation side.
Figure 1. High level Data Flow
Standardization, Automation, Security
Our emphasis on automation allowed us to create and manage multiple audiences, programmatically offer varying possibilities, and select preferences from curated categories. Earlier, teams used different naming conventions and followed different definitions for the audiences. However, with the new Audience Generation UI, we are now able to standardize audience naming and ensure better governance and reusability.
Now, customer handling is also managed via internal customer IDs and hashed emails sent through APIs. This automated functioning prevents leaks of important data and mishandling of customer information. At Vista DnA, we’re also emphasizing category standardization across various parameters and homogenizing queries across the business.
Ensuring Effective Data Quality Checks
While the data products are offering valuable insights to our marketing teams, we also wanted to ensure that the data accessed by the teams is of good quality. As a result, we developed several quality checks for the data sources.
Threshold, for instance, is one of the quality checks we have in place to continuously monitor the audience volume each day. If there is a significant drop in the volume below the threshold limit, we immediately get alerted through automated emails and push notifications on Slack.
The Vista DnA Experience
Our data-driven marketing approach helped us develop effective data products to serve our internal customers in numerous ways. With upstream table checks, they can now quickly get insights on customer uniqueness, incomplete or inconsistent customer attributes, or empty results. With access to authentic data, our marketing teams are also more empowered to offer a better, more personalized experience to our end customers.
At Vista, we realize the importance of a customer-first marketing strategy, which is why we’re always in the loop attending conferences to understand the challenges of other businesses. We’re also constantly evaluating other tools available in the market that can help us build innovative data products and advance toward our goal of becoming one of the world’s most iconic data-driven companies in the world.