You Might Like These Articles
Incrementality - The Adaptable Lineup
Incrementality will not report who touched the ball last. It depends on attribution data to provide the reporting, but uses machine learning and algorithms to understand causality.
Who scored goals? What were the weather conditions? What was the lineup? Was there a change in the lineup? Was there a change in the conditions?
How many of the players touched the ball, and for how long? Did the team win? What happened during last season? How big was the audience? Did any player get a significant raise? Did the team uniform change? Is there a new player?
By observing ALL the observable changes across hundreds of teams and thousands games - data scientists are able to create hyper parameters (a combination of parameters across a time series) that can indicate what is the expected outcome of a change.
The word “testing” does not make justice to the process of evaluating incrementality - as true incrementality testing does not require a seperate “test” in a laboratory setting , but uses the available data and changes in the data to to provide results.
Incrementality testing has been a holy grail in marketing , but only very few invested the resources required to evaluate it.
With the current market conditions eliminating identifiable data companies are forced to make the effort of researching incrementality as the best alternative to measurement.
Last Touch - The Single Player Team
Last Touch Attribution tells you that 100% of the credit goes to the vendor that “touched” the user last. Some even go out on a lim and call this “deterministic attribution”.
But let's imagine what would happen if you make your budget decisions based on the player that touched the ball before the goal.
Last Touch Attribution would tell you, in a deterministic way, that the best player ever scored goals in past games. That is correct.
But from here - to assume that you can, in a determinis way, that this single player is the cause of the goal...well, that’s just not correct.
Multi Touch - The Impossible Lineup
Multi-Touch attribution is a great concept. By tracking every user engagement within their journey to build a model giving various steps in time different weights in order to give appropriate credit to each step.
This would mean reporting who touched the ball, when they touched the ball, how hard they kicked the ball, and how many players touched the ball before scoring a goal, in order to assign a value for each player.
While in theory this sounds great, there are 2 challenges to multi-touch attribution:
11 players, 90 minutes game ends up with an infinite number of combinations given the amount of data.
IT’S NOT POSSIBLE - since the data is NOT AVAILABLE.
Advertisers must utilize a platform to consolidate data from each media vendor (i.e. each player). Unfortunately, some of the largest players in the world (Facebook, Google) do not share even level data with 3rd party independent attribution solutions.
Incrementality Testing & Football
Do I Need an Incrementality Testing Software ?
“True Attribution Focuses on Incrementality” was a quote from LUMA Partners Digital Marketing Summit, showing that the digital marketing world dependence on Last Touch attribution leads to wrong decisions and waste.
While every method of attribution has value - optimization, forming strategy - depending only on a traffic attribution as means for measurement leads to catastrophic results.
The Football Analogy
Football (“Soccer” to our American readers) is a perfect analogy to Marketing. For a team to score a goal (acquire a new customer) a coordinated effort is needed amongst the players.
External factors influence the game (the audience, the weather, the location) and it is typically not ONE THING that makes a team win or lose. Even at a game where victory is achieved by a score of 1:0.
Did you Like this Article ? Share It!