The promise of the digital Advertising era was that “everything is trackable” – but that is not the case, especially for brands that advertise across platforms and mediums.
An advertiser spending their budget across TV, Cinema, Print, Sponsorships, In-Game, and across websites has no method to understand their user journey from the moment of awareness (being exposed to a new product for the first time) to the point of action (purchasing said product).
Digital First Advertisers enjoyed almost unlimited access to data. Using technologies for tracking and attribution, allowed digital advertisers to receive real-time data points, matching user IDs (e.g. Cookie, IDFA, GAID) to the last ad engagement (impression or click) prior to the conversion, in what is known as Real-Time Attribution.
The challenge with real-time attribution is that all credit to the conversion was given to one ad impression or ad click, ignoring any other form of advertising the user may have been exposed to previously.
Making decisions based on real-time attribution data caused many Advertisers to make wrong decisions. By allocating a majority of the budget towards publishers and media winning the attribution – Advertisers ignored the Top of the Funnel, ending up spending on traffic overlapping with organic conversions, thus, adding very little incremental value.
The holy grail of all advertising activities is to generate an incremental sales lift, or Incrementality.
Incrementality testing measures the true effectiveness of advertising activities regardless of tracking.
There are various ways to measure incrementality in Advertising:
A Blackout test is probably the least preferred measurement method by Advertisers.
The principle of blackouts, as it sounds, is to black out all advertising for a period to understand the baseline with no Advertising, only to restart Advertising gradually, medium by medium, publisher by publisher, to understand the value of each and every publisher.
Blackout tests have several challenges:
- Opportunity costs – While stopping all advertising, sales will take a significant blow, and the optimization outcome may not compensate for the decrease in sales revenues during the blackout test.
- Inconclusive Results, At Best – Marketing Performance is influenced by several factors. Many of which, have nothing to do with the product or the marketing activities. These factors can be anything from weather, special events, the activities of competitors, to a global pandemic, and so on. Making conclusions based on the blackout test as demonstrated below could result in the Advertiser concluding that Social Ads produce a negative impact on sales. While this may or may not be true – it’s impossible to isolate the impact of one channel where the overall results were as volatile as the graph below.
Similar to a blackout test, a partial blackout refers to stopping Advertising activities in a channel and compares this to other channels.
A partial blackout is almost completely redundant, as the result of such tests are very predictable.
While marketing reports may attribute performance to various channels and mediums separately, the reality is that audience overlap amongst channels, as presented in the diagram.
Restricting one channel will naturally cause performance to be attributed with other channels, and this may also reflect an impact over sales revenues.
A series of partial blackouts over time may cause Advertisers to conclude that none of the channels produce incremental results, hence, they are better off NOT ADVERTISING.
Random Audience Group Test
A randomized control group seeks to split the user base being advertised in a homogeneous way so that there is an almost equilibrium between group A and group B.
The groups are served with either PSA ads or the campaign creatives and the performance is measured to reach a conclusive outcome.
The good news is that this method can provide rather conclusive results as long as the Advertiser can fully control the targeting limiting against any spillage or overlap. The challenge in this approach is that given the requirement to identify “users” on publishers’ websites and apps – this method became obsolete in a privacy-first world.
With device identifiers, and cookies, being deprecated – the option of performing such tests became unreliable.
Some media vendors offer Advertisers a randomized control group test using the vendors’ first-party data. These tests however are extremely biased, and uncontrolled. The Advertiser has no way to validate the results, nor control what other ads served by other vendors the control or treatment groups will see and engage with.
The most recent method for incrementality measurement. Causal Inference is an algorithmic process to draw conclusions about the causality of results in a multi-variant and noisy environment. Causal inference is used in epidemiological research, in economics and most recently, in AI. Applying causal inference models requires research and development to come up with millions of time series comparing small and large changes in the data to create a prediction range comparing anticipated and actual outcomes to provide a clear recommendation.
The benefit of using causal inference in incrementality testing is that causal inference can process data regardless of where campaigns ran (i.e. digital, untrackable, and offline).
Incrementality measurement using causal inference can provide tactical recommendations, allowing marketers to utilize those in near real-time. The method is reactive rather than requiring Advertisers to make active changes in order to test. User-level data is not required for causal inference to operate flawlessly.
INCRMNTAL is an incrementality testing platform for Advertisers. The platform provides marketers with the tools to uncover the causality behind their marketing activities.
The software identifies the incremental value of campaigns and traffic vendors, Isolate price changes that lead to no additional traffic, find marketing activities that cannibalize organic user base or across paid marketing.