Abakus’ attribution methodology analyzes user populations and their response to marketing. In its simplest terms, if marketing had no impact on the population of users it exposed, then it would get attributed no value (see baseline explanation). Expanding on this to consider latency, the same is true, if marketing has no impact after a certain period of minutes, hours, days or weeks, its attribution will be zero for those days.
So, if users who see an ad on day 1 and then convert on day 7, respond at the same rate as users who are not exposed to marketing (baseline) the value of those conversions would not be attributed to marketing. Counter to that, if users who see an ad and convert on the same day, respond at a rate that is higher than the baseline, most of the value of those conversions would be attributed to marketing
How can we be certain? The best way to prove this is to design an experiment to test the hypothesis. Here is a description of the experiment we ran.
Experiment: Attribute single day of marketing to n days of conversions via two methods: i) single day of marketing attributed to n days conversions, ii) single day of marketing to each day of conversion (days 1 through n) in turn. Demonstrate both approaches align.
Show that Attribution
The results for one test client who was marketing anti-virus software for online purchase and download are shown below.
Attribution Y14-30 = 0.
The Attribution value of X was within 5% of Σ1-13Y
Latency = up to 13 days.
Y attribution values shown in chart below.