Competitors

DataHawk runs a daily analysis of your market and identifies the competitors for your products.

Dataset

One Competitors dataset is currently available on Destinations. It is available to users who have tracked products.

Overview

Using DataHawk Destinations, your DataHawk-powered Snowflake database contains one table about Competitors. This data is available under the PRODUCTS schema.

We compute competitors for your own (labelled as “owned”) tracked products and the products from your connected accounts (Ads or Seller Central).

How do we identify competitors?

To find competitors, we first select your products and look at all the other products that appeared on the same listing page or in related keywords or categories within the past seven days.

Next, we leverage an in-house method to score the pairs based on their frequency of co-occurrence and semantic similarity. We select the ones with the highest score, with a confidence greater than 90%. This score is called Competitiveness Score.

Every week, we run the same analysis and append new competitors to your database table.

Destinations Datasets

SchemaTableAbout
PRODUCTPRODUCT_COMPETITORSDisplays competitor products for your owned products (set as owned, from Ads account, from Seller Central account)

Considerations

Refresh Frequency

Data is updated daily. If you track your product by noon UTC, you will have suggested competitors by 6 a.m. UTC.

Why do you not see competitors?

We are able to find competitors for 95% of products. But if we could not, this is because we could not extract any obvious category for your product and because we never saw your product in a keyword. We would then suggest you refine your product category and/or track keywords relevant to your product.

To find competitors, we look at:

  • products in the same browse nodes
  • products in the same keywords
  • products that appear on your page

Then, we use an ML algorithm based on the frequency of appearance of each potential competitor in each category (browse node, keyword, page) plus some semantic similarity KPIs to estimate the probability that these potential competitors are actually competitors. Hence, if we did not observe products surrounding yours or if our ML algorithm returned a too-low confidence score (<90%), you may not see any competitors. This is because we favoured precision over coverage.