Systems and methods for generating a true return on advertisement sales value using short-term and long-term factors to analyze advertising effectiveness

A method comprising:

• receiving, for a target product, a first product value;
• determining, for the target product, a cannibalization value using at least one of experimental data and non-experimental data to generate a model of probabilities comprising a neural network that includes a plurality of layers, wherein the model of probabilities is configured to apply one or more weight matrices at each layer of the model of probabilities to predict whether each sale of a plurality of sales of the first product would be realized without an identified advertising characteristic associated with the first product, and wherein the model of probabilities provides, at an output layer of the model of probabilities, a cannibalization probability indicating, at least, the cannibalization value;
• generating a second product value by adjusting the first product value using the cannibalization value;
• determining, for the target product, a return value based on return data;
• generating a third product value by adjusting the second product value using the return value;
• determining, for the target product, a lifetime customer value based on at least the cannibalization value;
• generating a fourth product value by adjusting the third product value using the lifetime customer value;
• determining, for the target product, an organic rank value based on organic placement data and value of organic placement data;
• generating a fifth product value by adjusting the fourth product value using the organic rank value;
• generating, based at least on the fifth product value, an actionable report descriptive of an efficacy rate of the target product in a computer-networked marketplace.

A computing device comprising:

• a processor;
• and
• a memory including instructions that, when executed by the processor, cause the processor to:
• receive, for a target product from a computer-networked marketplace, a first product value;
• determine, for the target product, a cannibalization value using at least one of experimental data and non-experimental data to generate a model of probabilities comprising a neural network that includes a plurality of layers, wherein the model of probabilities is configured to apply one or more weight matrices at each layer of the model of probabilities to predict whether each sale of a plurality of sales of the first product would be realized without an identified advertising characteristic associated with the first product;
• generate a second product value for the target product based on at least the cannibalization value.