To estimate your product's reach, it is essential to understand how easy it is to be sold. Therefore, we have developed an algorithm that assesses the probability with which your product will be bought, knowing that a customer is already on your product's page. We call this KPI the Sales Conversion Rate.
We considered the information displayed on your product's page and looked at the main metrics (e.g., price, number of reviews, ratings, etc.)
Below we illustrate the distribution of such features as a function of the probability that a given customer buys your product if they are already on your product's page.
You can see, for instance, that the higher your product's page is, the less likely the product is to be bought (negative correlation). This is probably because customers need more time to make up their minds when a product is expensive. The opposite is true when considering the number of reviews: the more reviews a product has, the more likely it is to be bought.
We trained two algorithms to predict the Sales Conversion Rate using all these features. On average, the error of the first version was 7.3%. We managed to decrease it to 6% with the second version. Below we compared the distribution of the SCR of those two versions to the real one. Our versions closely follow the real distribution, even if they tend to be a bit more optimistic than actuality.
Updated 6 months ago