How to catch best offers

Probably if you ask this question to different people, you will get different answers.

User guide

What is an offer?

Offers, offers, offers. 💥.

If everything is an offer, then nothing is an offer. Think about that.

Commonly, an offer is an item with reduced price, but the point is that you need to establish a baseline.

Example: If you find a new sail of this year for €350, looks a really good price, but this is because you know how much a sail with the same characteristics normally costs. If the rest of the items with the same characteristics cost under €350, definitely your sail is not an offer!

Establishing the baseline

To detect real offers in our application, first of all we group the items in segments based on three characteristics:

  • Category: Each items of each category have different prices, so this is an expected first classification to avoid false positives.

  • Year: The more recent the year is, the more expensive is the item.

  • Condition When an item is new it is more expensive than used. In addition, it complements the rest of characteristics perfectly: a new sail of the last year could have a similar price compared with a new sail of the current year.

Grouping the elements based on these properties, we can obtain the minimum, maximum and average price per each segment and score all the items inside the segment with a value between 0 and 1:

  • If the item has a value near to 1, it means that the price of the item is near to the minimum price detected in the segment.
  • If the item has a value near to 0, it means that the price of the item is near to the maximum price detected in the segment.

Sorting results by score

Now that we have all items scored, we can sort the results based on this score.

We have divided the offers into four groups (), depending on how good the offer is:

  • Legendary: Better than 95% of the segment.
  • Epic: Better than 80% of the segment.
  • Rare: Better than 70% of the segment.
  • Uncommon: Better than 50% of the segment.

Also in the application we apply another complementary rule to sort the results: The date on which the item has been part of the search engine. The items fetched in the last 24H have a NEW badge next to the title.

Keep offers up to date

Every time a new item is added in the search, we recalculate the score of each item to keep offers up to date. Here is how it looks like:

catch true offers with prices normalized never was more easy 🎉

You can continue applying filters to improve the search.

Next post

How to write good products

Tips Shops

As search engine, We work all time transforming data into information. To achieve it, we transform the shop information of each item into structured data.

Read More →

Comments