Good math
We monitor stores directly. Specifically, we monitor the inventory of each store. This is reliable data because store owners can not fake this data in any way.
Let's make an example so you can understand this.
At a timestamp (T1) there are 10 units in stock for a particular product. If at a later timestamp (T2) there are only 3 units in stock for that particular product, the only possible explanation is that 7 units (10-3) were sold.
Sometimes at T2, we find more units in stock for a particular product, and that's why we are able to track restocks and returns.
Good sources
As I mentioned, we go directly to the source, monitoring stores. But, there are also other reliable sources and signals that we monitor that allow us to enrich sales data. These signals include customer conversations we intercept from social activities, forums and more.
This is why we are able, most of the times, to enrich sales data with info on gender and location of sales.
We also track this down to product variants.
