Let’s say a user comes across an advertisement for a product on Snapchat and curiosity sparks. Later that day, the customer turns on the laptop to look for that product. The customer searches for it in the search engine and clicks on the paid advertisement that is displayed at the top and converts.
Although the curiosity was sparked by the Snapchat advertisement, 100% of the conversion is attributed to paid search even though the customer would have likely found a way to purchase the product regardless of whether the paid advertisement was present or not.
So how do we uncover the true value of digital media investments, you may ask?
There are several ways to uncover the true value of our media investments. At Acceleration, we work with a Unified Measurement Framework where we distinguish between three approaches that shed light on the effectiveness of different media and help drive perpetual growth.
At the very top, we have Econometrics also known as Marketing Mix Modeling. These models provide a very accurate and holistic overview of media investments. Additionally, econometrics models are very useful as they help advertisers determine the return on investment (ROI) of their marketing efforts which can be utilized to improve marketing efficiency. Assume a retail company wants to understand the impact of their marketing efforts on revenue. They have been investing in various marketing channels such as TV, print, online advertising, and social media. Additionally, they have been running promotions at different times of the year. econometrics models help the company identify which marketing channels and promotions are driving revenue and by how much.
Next, we have Paradigm which is a cookie-less attribution framework that provides dynamic media evaluation and media activation for businesses. Assume a business wants a future proof and accurate way of allocating its digital media budget while also considering the increasing restrictions around digital privacy and cookies. Paradigm evaluates media investment with the use of live updated cookie-less data together with input from other media evaluations like econometrics models or geographical experiments. The findings are then activated directly in the media buying using automated machine learning algorithms across media platforms. The core benefit of Paradigm is that it provides a futureproof, consistent, and accurate way of evaluating and activating media across platforms.
The last element of the Unified Measurement Framework, and the focal point of this article, is Experiments. Experiments are a very valuable tool as they allow businesses to determine the incremental impact of specific marketing activities. As mentioned previously, media platforms tend to under- or overestimate media performance – thus, experiments can be used to measure the true causal effect of marketing campaigns or media channels.
There are different types of experiments, Brand Lift, Conversion Lift, A/B tests, and geographical experiments – for now, we will focus on geographical experiments.
What is a Geographical Experiment?
A geographical experiment is a methodology that aims to quantify the causal impact of marketing tactics and ascertain if they are generating additional value or whether the result would have been the same independent of the tactic. In these experiments, a target market is split into two groups: a test group and a control group, both of which historically have displayed a similar trend in the KPI being tracked. When all other criteria are the same across the two groups, the test group is subjected to a change in the marketing tactic while the control group is not. After the experiment is complete, the relative difference between the two groups is analyzed to determine the marketing tactic’s incremental effect.
How can geographical experiments help businesses make better marketing decisions?
Media platforms traditionally rely on last-click heavy attribution, which assigns 100% of the conversion value to the last interaction a customer had with a marketing channel before converting. The performance of various media channels is typically under- or overestimated by media platforms since they rely solely on last-click attribution.
Here, going back to the example in figure 1, experiments can be utilized to determine the true incremental value of Snapchat and paid search. In this very simplified example, it is assumed that the businesses only have Snapchat and paid search in their media mix[VF1] [DA2] . To estimate the incremental value of each media channel, the advertiser conducts two experiments – one where Snapchat is disabled in the test group, while remaining active in the control group, and one where paid search is disabled in the test group while remaining active in the control group.
After the test period, the advertiser discovers that the incremental value of Snapchat is 40% and the incremental value of paid search was 60% rather than 100% for paid search cf. figure 1. These findings enable businesses to make more data-driven decisions on how to allocate their budgets and thereby increase their ROI.
In conclusion, geographical experiments are a valuable and futureproof tool for marketing teams as they provide a clear and measurable way to determine the incremental value of marketing tactics without being limited by cookies. Additionally, geographical experiments can be used to test multiple marketing tactics simultaneously, allowing businesses to quickly identify which strategies are most effective. As a result of this, they can make more accurate and data-driven decisions about their marketing strategies, and better understand the impact of the marketing tactics.
Missing out on geographical experiments and incrementality experiments as a whole means that advertisers may be relying on inaccurate performance estimates which essentially prevents them from gaining vital information about the true effectiveness of their marketing tactics.
If you want to futureproof your measurement strategy and get a more detailed and precise evaluation of your marketing tactics or are simply interested in hearing more, do not hesitate to contact Daniel Azadian, Digital Data Consultant (firstname.lastname@example.org) or Kasper Madsen, Senior Measurement Consultant (email@example.com).