Zovda
Price monitoring looks like the simplest dataset in commerce: a product, a date, a number. It is simple to collect and easy to get wrong, because the number on the page is rarely the price.
It is usually the price after a discount that may or may not apply, for a variant that may not be the one in stock.
The three values
- Listed price, before any promotion
- Effective price, what the page shows today
- Paid price, what buyers actually paid, including shipping and tax
A series that mixes them looks like volatility and is actually a change in definition.
Variants break the series
A phone in three storage sizes has three prices. A garment in seven sizes has two in stock. A listing that averages them produces a number no buyer ever saw.
Store the variant, and store whether it was in stock. Out-of-stock prices are not prices; they are decoration.
Shipping and tax belong somewhere
For cross-border comparison, a price without shipping is not comparable to one with it. Either include both consistently or state clearly that you do not.
Keep the capture method
Was the price read from the page, from a feed, or from a transaction record? The three have different error profiles and should not be merged without a field.
Zovda assembles these series through Datuva, where each observation keeps its variant and its capture date. The data research collection shows how a dated observation is carried.
Tell us what price question you need answered. We reply within one business day.