Assuming I can collect the demand of the purchase of a certain product that are of different market tiers. Example: Product A is low end goods. Product B is another low end goods. Product C and D are middle-tier goods and product E and F are high-tier goods.

We have collected data the last year on the following 1. Which time period (season - festive? non-festive?) does the different tier product reacts based on the price set? Reacts refer to how many % of the product is sold at certain price range 2. How fast the reaction from the market after marketing is done? Marketing is done on 10 June and the products are all sold by 18 June for festive season that slated to happened in July (took 8 days at that price to finish selling)

How can data science benefit in terms of recommending 1. If we should push the marketing earlier or later? 2. If we can higher or lower the price? (Based on demand and sealing rate?)

Am I understanding it right that data science can help a marketer in this aspect? Which direction should I be looking into if I am interested to learn about it.

  • 1
    $\begingroup$ I am going to edit the title to be more descriptive. Good question, though. $\endgroup$ Jun 12, 2015 at 19:48

2 Answers 2


You should be able to use linear regression to find correlation between the factors which cause your products to sell better (or worse).

There are many correlations you can test against in this data set. Some examples are:

  1. If a product has been marketed aggressively, does it sell more quickly?
  2. If a low tier item is available, do fewer high-tier items sell?
  3. If multiple high-tier items are available, are fewer sold of each item?

Keep in mind that correlation does not necessarily imply causation. Always think about other factors which may cause sales to go up and down. For example, you may sell more high tier items in a season one year than another year. But, this could be due to changes in the overall economy, rather than changes in your pricing.

The second thing you can do is perform A/B tests on your product sales pages. This gives you clear feedback right away. Some example tests could be:

  1. Show the user one high-tier product and one low-tier product (A). Show the user two high-tier products and no low-tier products(B). Which page generates more revenue?
  2. Send out marketing emails for a seasonal sale 5 days in advance to one group of users (A). Send the same email to a different set of users 1 day in advance (B).

There are many possibilities. Use your intuition and think about previous knowledge you have about your products.


This type of questions can be answered by building a demand prediction model that uses product attributes (including product tiers) and marketing activities as input features. Such as model then can be used to do what-if analysis of different marketing scenarios. Boosted decisions trees and LSTMs are often used to implement such a model. One of the articles that describes the process in more details - https://blog.griddynamics.com/predictive-analytics-for-promotion-and-price-optimization/


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.