2023-02-03 | Case study | Insights

Understanding engagement is a proxy for conversion optimization

Understanding engagement is a proxy for conversion optimization

With billions in assets under management, this brand is a globally active player providing a digital platform for private equity investing for people interested in investment instruments. With this offering, the company aims to lower investment barriers.

As part of the registration process, the company conducts an assessment to evaluate the ability of users to make decisions and take appropriate risks. Due to a low number of registrations, engagement-increasing tactics were needed so the client initiated a conversation with Digitl.

sessions discovered that can be engaged by providing engaging content

of campaigns identified measured by the KPI cost per engagement score

Deeper insights into user behavior through engagement scoring

Deeper insights into user behavior through engagement scoring

Digitl developed an engagement score model that enables onsite and offsite optimization by analyzing onsite behaviors such as scrolling down the website, clicking on certain categories, watching videos, etc., to calculate the value of user traffic.

The first step was to implement Google Analytics 4 on all pages of the client’s website in order to gain an adequate database for calculating customer engagement scores. Digitl used the native BigQuery export within GA4 to access and query raw data. The query table results were then stored as a BigQuery table and exported as a CSV file to Google Cloud Storage. The saved CSV was exported to Jupyter Notebook in Vertex AI Workbench for following identifying patterns with the help of exploratory data analysis (EDA). Statistical methods and graphical visualizations were used in the EDA to learn about patterns and frequencies in the data.

Digitl selected a set of features/ events that reflect and measure user engagement based on the frequency of occurrences. The identified events formed the basis of the data set for correlation analysis. The events were weighted according to their relevance to user registration. Lastly, they were analyzed with a view to patterns and distributions per engagement score bucket. The process was automated and the results were provided in real-time so they could be also used for onsite-personalization.

Digitl joined campaign data from Campaign Manager 360 for direct publisher campaigns and programmatic campaigns executed in Display & Video 360. This way the performance of these campaigns that was measured using the KPI "Cost per engagement score" could be evaluated. The results were visualized using Looker Studio.

On- and offsite optimization possibilities based on the engagement score

On- and offsite optimization possibilities based on the engagement score

The distribution of engaged sessions was analyzed using a histogram. In order to accurately reflect different user engagement levels, Digitl constructed four buckets instead of only distinguishing between engaged and unengaged users. The team discovered that 29% were at a lower level of engagement. Additionally, geography, traffic source, device, and other factors could be used to distinguish between engaged and unengaged users.

Digitl helped identify content that was of particular importance to each engagement score group based on the distribution of pageviews per navigation content category. The foundation for onsite personalization was laid here.

Moreover, Digitl was able to determine what campaign activities drove engaged traffic through this project. The KPI "Cost per engagement score" was used to evaluate campaign performance and focus the bulk of budget on the top 20% of the awareness campaigns in the next media planning cycle.

The Goals

  • Identify the unengaged traffic on the brand’s website.
  • Discover what content is most engaging on the website to engage non-engaged users.
  • Understand how engaged users are after being exposed to the brand’s campaigns.
  • Guide users down the funnel by optimizing campaigns based on user engagement.

The Approach

  • Create an engagement score methodology.
  • Apply the engagement score in a follow-up exploratory analysis on GA4 raw data.
  • Identify different user buckets based on the level of engagement and analyze them.
  • Analyze and optimize campaigns based on the engagement score.

The Results

  • Four engagement buckets accurately reflect varying levels of engagement among users.
  • By providing engaging content to users with low engagement levels, 29% can be engaged.
  • To focus the bulk of the budget on the top 20% of campaigns, the KPI "Cost per engagement score" was used for performance evaluation.

Do you need more Info?

Contact us