Analytics Could Flip Event Marketing on Its Head
- Event Analytics. As Super Bowl & Valentine’s campaigns approach, tighter budget scrutiny leads marketers to reassess the value of big event investments.
- Data insights. Time forecast analysis becomes crucial for evaluating event ROI, offering a deeper dive into how such events impact customer activity over time.
- Analytic era. The shift toward data-driven decision-making empowers marketers to navigate the complexities of event participation and optimize marketing budgets.
As the world gears up for another Super Bowl and Valentine’s Day campaigns loom, marketers are facing tighter scrutiny over their event marketing budgets than ever before, highlighting the importance of event analytics. With a plethora of options for launching campaign messages, not all brands are as enamored with big events as they once were.
If your brand is considering an event, consider analyzing your data with a time forecast to assess the campaign’s value. This approach will yield a dataset spanning various time periods, which can help determine whether the event statistically contributed to an increase in activity. Let’s take a look at event analytics.
Marketers invest millions in high-profile events, with this year’s Super Bowl ad slots averaging $7 million. Conducting such an analysis offers insights into an event’s lasting influence, allowing managers to refine campaigns and sustain momentum.
Related Article: What Is Predictive Analytics? And How It Works
Events are the Big Tent of a High-Risk, High-Reward Customer Circus
Events with significant media coverage are often referred to as big tent events. While not all enjoy global media exposure like the Super Bowl, some hold sway within specific industries, such as the Chicago Auto Show. Despite a slight decline in influence, the Chicago Auto Show continues to be the premier event for auto manufacturers, renowned for introducing new car models mid-year.
Marketers have traditionally favored big tent events for their broad audience reach. It was once thought that a large audience could spark customer interest and excitement, ultimately driving sales.
Related Article: Customer Data Analytics and AI: The Smart Path
The Analytics Era
Welcome to the analytics era. Metrics now offer a more precise way to understand how impressions from various mediums — be it a digital ad on a display, an in-stream video, a podcast appearance, or an online post — are leading to registrations, subscriptions and sales.
Furthermore, consumers can now research products and services on their smartphones, tablets, or computers while on the move. This not only increases the value of analytic metrics for analysis but also offers customers the option to bypass large crowds while locating the products and services they need.
Related Article: Customer Data, Analytics Top Priorities for Customer Service Leaders
Event Analytics: Reconsidering Marketing Budgets
The widespread availability of event analytics has led marketing managers to reconsider how their brands allocate marketing budgets, including the evaluation of event participation. For instance, Stellantis recently announced its decision not to display its cars at this year’s Chicago Auto Show, as it reassess its marketing strategy and questions the overall value of auto show participation on a case-by-case basis.
Related Article: What Is Customer Analytics? And Why It Matters
The Significance of Time Series Data for Big Tent Events
With the advent of more readily accessible data, time series data has emerged as the key driver in marketing strategy. Its analysis enables a review of how a digital campaign evolves over time, addressing performance questions like “Is there more engagement or less?” or “Did this event alter our level of engagement?”
Shifts in Customer Behavior
Statistical models derived from time series data can pinpoint these answers, identifying whether a shift in customer behavior is unfolding over time.
Exploring a Range of Data
You can explore a range of time-related data, such as comparing store sales with online conversion data, or examining other metrics like store returns to see if they are increasing over time.
Ensure you have ample data to pose a crucial question: Did the campaign metrics experience a significant lift that indicates a consistent trend into the new year?
Related Article: When to Use AI to Forecast Website Data
How to Create Your Data for Analysis
The initial step involves collecting data from activities related to a big tent event campaign. This data could include clicks on associated digital ads, visits to a landing page, or registrations obtained through a landing page.
The data can be uploaded into either R or Python and formatted as a time series object. It should be structured as time series data, featuring a date index column — a column containing dates — and several columns for the metrics you wish to analyze. For instance, if you’re examining website visits, conversions, and sales, you should organize them in a table as follows:
Both R and Python have data structures designed for this. Users of R turn to xts, while data analysts versed in Python typically use the Pandas module.
The index-dataset structure makes time series data a good starting point for understanding R and Python.
Examining the Lift
Next, time series data can be analyzed in a forecast to examine the lift before and after an event. Marketers and analysts can employ Bayesian analysis for this purpose. CausalImpact, a free framework developed by a team at Google, facilitates this analysis and is available as a library for R and a module for Python. CausalImpact enables users to create a model that visualizes the time series data before and after the event.
Organizing Your Time Series Data
Users should organize their time series data to include a range of dates before and after a major event of interest. CausalImpact recognizes these dates as the pre-period (before) and post-period (after). It calculates a Bayesian model to predict the data trend in the post-period as if the event had not occurred. This process relies on assumptions, such as the stability of the relationship between covariates — external factors being controlled for — and the time series data throughout the post-period.
An Event Analytics Example
Here’s an example of how to represent pre and post-period date data in R, extracted from time series data. The pre-period ranges from the start date to the event date, while the post-period spans from the day after the event to the end date. This example demonstrates the approach for 100 days before and after the Super Bowl, for instance.
The Final Step
The final step involves creating a visualization after running the model. This visualization consists of three panels. The top panel displays the trend before and after, marked by a vertical line indicating the event day. The middle panel illustrates the degree of difference, while the bottom panel shows the cumulative points derived from the middle panel’s data. This example, inspired by a demo from the CausalImpact GitHub site, demonstrates the appearance of a lift after an event — both upper panels exhibit a significant upward shift post-event, indicating that the variables in your time series dataset were influenced.
CausalImpact also provides a statistical output, which includes a convenient text summary. This summary can be incorporated into your report. While it remains statistical in nature, it clearly explains the results, offering text that can be swiftly shared in an email or a report with your team.
There are additional statistical programming frameworks capable of conducting Bayesian analysis of time series data, along with other useful analyses. Prophet, a programming framework developed and released by Meta, also includes functions for Bayesian time series analysis.
A crucial analysis to consider is data sustainability, essentially questioning, “Is there a repeatable pattern independent of the campaign period?” As outlined in a previous post, you can answer this question using an Augmented Dickey-Fuller (ADF) test in advanced time series analysis. If the results suggest that the trend in the data is independent of the specific period, it indicates ongoing demand activity beyond the event campaign, which can be leveraged for forecasting.
Final Thoughts on Leveraging Time Series Discussions
Leverage time series discussions to initiate conversations about changes in customer behavior following an event. This approach uncovers insights that can inform decisions on further analysis and whether event analytics reporting based on digital media (ads, podcasts, videos) could be advantageous.
Crucially, discussions about time series analysis can provide additional information that assists in guiding budgetary decisions regarding the worthiness of further investment in an event.