DEV Community

Cover image for noon Products Dataset: Is the Featured Offer the Cheapest? 494 UAE and Saudi Offers in Python
Happy Endpoint
Happy Endpoint

Posted on Originally published at Medium

noon Products Dataset: Is the Featured Offer the Cheapest? 494 UAE and Saudi Offers in Python

Every product page on noon shows one price next to the Add to cart button. Below it, sometimes, a line says other sellers offer the same item. Before noon's November sale, I wanted to know two things: is that featured price the lowest one on the page, and do the answers differ between the UAE and Saudi Arabia? So I loaded a free sample of noon products and seller offers into pandas.

The short version: on 49 of the 62 products where at least two sellers competed, the featured offer had the lowest price. That is 79.0%, and the UAE (78.6%) and Saudi Arabia (79.4%) were almost identical. When the featured offer was not the cheapest, it cost a median of 2.9% more, and never more than 13.5%. The interesting part is the rest of the offer list, where the dearest seller often asked more than double the cheapest.

I started out wanting a straight UAE vs Saudi price comparison. The free sample cannot give one, for reasons that come up in the data section, so this article answers the seller question first and then shows what the full snapshot says about the two markets.

The data

The files are the free sample of the noon UAE and Saudi Product Snapshot: 100 products and 494 seller offers on Hugging Face, under CC BY 4.0. The snapshot was taken on 1 November 2026 (UTC), before noon's November sale. The data is compiled by Happy Endpoint from publicly available listings on noon. Happy Endpoint, an independent data provider, has no affiliation with noon.

There are two files:

  • products.csv has one row per product and market: 50 products in the UAE (prices in AED) and 50 in Saudi Arabia (prices in SAR), covering all 30 product types in the full snapshot, from 55-inch TVs and gaming laptops to oud perfume and sunscreen. Each row carries price, original_price, discount_percent, rating, deal_label, promotion_title and the featured seller.
  • offers.csv has one row per seller offer on those products: 249 in the UAE and 245 in Saudi Arabia, from 297 different sellers. The column that matters here is is_buy_box, which marks the featured offer, the one the product page shows first.

The full snapshot behind it has 2,934 products (1,492 UAE, 1,442 Saudi Arabia) and 1,969 seller offers. The sample keeps one or two products per type and market, mostly the top-ranked ones: 56 of the 100 rows are rank 1 and 40 are rank 2.

That is why the sample cannot answer "is noon cheaper in Saudi Arabia?". The two markets hold different products under the same type: the UAE coffee machines in the sample are a Nespresso and a De'Longhi, the Saudi ones are two Black and Decker models. Only 3 products appear in both markets. The seller question, on the other hand, is answered product by product, and every one of the 100 products has offer data.

The code reads both files straight from Hugging Face and needs only pandas.

import pandas as pd

BASE = "https://huggingface.co/datasets/happyendpoint/noon-uae-saudi-products/resolve/main/"

products = pd.read_csv(BASE + "products.csv")
offers = pd.read_csv(BASE + "offers.csv")

# One row per product and market: offer count, cheapest, dearest and median offer price
per = offers.groupby(["region", "sku"])["price"].agg(["size", "min", "max", "median"])
featured = offers[offers["is_buy_box"]].set_index(["region", "sku"])["price"]
per["featured"] = featured

# Only products where at least two sellers compete
multi = per[per["size"] >= 2].copy()
multi["featured_is_cheapest"] = multi["featured"] <= multi["min"]
multi["gap_pct"] = (multi["featured"] / multi["min"] - 1) * 100
multi["spread_pct"] = (multi["max"] / multi["min"] - 1) * 100
multi["vs_median_pct"] = (multi["featured"] / multi["median"] - 1) * 100

print(multi.groupby("region")["featured_is_cheapest"].agg(["sum", "size", "mean"]))
print("Featured cheapest overall: %.1f%%" % (multi["featured_is_cheapest"].mean() * 100))
print("Gap when it is not:", multi.loc[~multi["featured_is_cheapest"], "gap_pct"].round(1).sort_values().tolist())
print("Median spread, dearest vs cheapest: %.1f%%" % multi["spread_pct"].median())
print("Median featured vs median offer: %.1f%%" % multi["vs_median_pct"].median())

# Join back to the product rows on region AND sku (a few skus are listed in both markets)
names = products.set_index(["region", "sku"])[["query", "brand"]]
print(multi[~multi["featured_is_cheapest"]].join(names)[["query", "size", "featured", "min", "gap_pct"]].round(1))
Enter fullscreen mode Exit fullscreen mode

The featured offer is usually the cheapest

38 of the 100 products had a single offer, so there was nothing to compare. The other 62 had between 2 and 46 offers, with a median of 4. Each of them had exactly one featured offer, always in position 0.

On 4 in 5 products with rival sellers, the featured offer had the lowest price

On 49 of those 62, the featured offer was the cheapest (ties included). By market: 22 of 28 in the UAE and 27 of 34 in Saudi Arabia. If you take the median offer on each product as the reference instead of the cheapest, the featured offer was a median of 14.4% below it. Clicking through to "other sellers" mostly shows you dearer prices.

When it is not, the gap is small

When not the cheapest, the featured offer cost 0.1% to 13.5% more than the cheapest seller

The 13 exceptions are spread over 7 Saudi and 6 UAE products. On most of them the difference is a few dirhams or riyals: SAR 0.85 on a smartphone, AED 18.20 on a 55-inch TV. The median gap is 2.9%, a median of 8.9 dirhams or riyals. The widest is an electric shaver in Saudi Arabia, featured at SAR 169.05 while another seller offered it at SAR 149, 13.5% less.

Delivery looks like part of the reason. The featured offer was a noon Express offer on 61 of the 62 contested products, against 52.8% of the 394 other offers on the same products. But on 10 of the 13 exceptions the cheaper rival was an Express offer too, so delivery is not the whole story. The featured sellers are also much bigger: their median rating count is 138,271, against 612 for the other sellers. noon itself held the featured offer on 21 of the 50 UAE products and 19 of the 50 Saudi ones.

The sample cannot tell you how noon picks the featured offer. It can tell you that on these products, the shopper who takes the default pays the lowest price four times in five, and a few percent more the fifth time.

The rest of the offer list is all over the place

The featured offer sits near the bottom of the range. The top of the range is a different matter.

On 35 of 62 contested products, the dearest offer was 50% or more above the cheapest

Across the 62 products, the median gap between the dearest and the cheapest offer was 56.6%. On 20 of them the dearest offer was more than double the cheapest. The UAE was wider than Saudi Arabia: a median spread of 73.8% against 47.6%. The most contested product was a fitness tracker in the UAE, with 46 offers.

So "price on noon" is not one number. For a product with several sellers, the featured offer, the median offer and the dearest offer can be far apart, and an average over all offers says more about how many high-priced sellers list the item than about what shoppers pay. If you build price benchmarks from marketplace data, use the featured offer or the cheapest in-stock offer and say which one.

UAE vs Saudi: what the sample can and cannot say

Some market-level comparisons do work on 50 products each:

Sample, 50 products each UAE Saudi Arabia
Discounted against original price 40 (80.0%) 44 (88.0%)
Median discount, discounted products 39.5% 38.0%
Sale label shown 23 26
Promotion shown 19 26
Featured offer sold by noon 21 19
Contested products, featured cheapest 22 of 28 27 of 34

The promotions differ in kind. In the UAE, 13 products carried "Extra 10% off" and 6 "10% cashback". In Saudi Arabia all 26 promotions were "10% cashback".

For prices, the 3 products listed in both markets give a first hint. Both currencies are pegged to the US dollar (AED 3.6725 and SAR 3.75 per dollar), so 1 AED is 1.021 SAR at a fixed rate. Converted that way, all three cost more in Saudi Arabia: a pair of wireless earbuds by 14.8%, a hair dryer brush by 5.4% and a fitness tracker by 29.3%. Three products are an anecdote, not a finding.

The full snapshot answers the category question. The noon UAE and Saudi price statistics page, built from all 2,934 products, compares the median price of each product type. At the dollar pegs, the median was lower in Saudi Arabia in 16 of 30 types and lower in the UAE in 14, with hair dryers level to within a riyal. The extremes are wide: the median coffee machine was 82.3% cheaper in Saudi Arabia, the median office chair 119.4% dearer. Saudi listings were also discounted more often: 71.5% of products against 59.7% in the UAE, and more often in 20 of the 30 types.

Those category medians compare what is popular in each market, not the same item. The sample's coffee machines (a Nespresso and a De'Longhi in the UAE, two Black and Decker models in Saudi Arabia) hint at a brand mix difference, but two products per market cannot explain a median. A clean like-for-like answer needs the products listed in both markets, matched on sku, across the whole snapshot.

The sample shares are higher than the full snapshot's (80.0% and 88.0% discounted against 59.7% and 71.5%), because it keeps the top-ranked products. The direction matches: Saudi Arabia discounts more often.

Pitfalls in the data

original_price is set per market, so discount_percent is too. The same pair of wireless earbuds had an original price of AED 99 in the UAE and SAR 189 in Saudi Arabia, so a 35% discount in one market and 60% in the other, while the selling prices were within 15% of each other. A hair dryer brush showed an original price of 288 in both markets, AED in one and SAR in the other. Compare selling prices, converted, not discount percentages.

A blank original_price is not a zero discount. 16 of the 100 products have none, so discount_percent is blank too. Count them as "no reference price", or your discounted share drops for no reason.

discount_percent is rounded. It matches 1 - price / original_price to the nearest whole percent on all 84 rows that have both. Recompute it if you need decimals.

Join on region and sku together. 3 skus appear in both markets, with different prices in different currencies. A join on sku alone pairs a dirham price with a riyal price.

The featured offer and the product price can drift. The offers were collected after the product rows, within about 25 minutes (19:46 to 20:11 UTC). On 99 of 100 products the product price equals the featured offer; on one Saudi product it was SAR 553 in products.csv and SAR 579 in offers.csv. Use one file per question.

brand is blank on 19 rows. The brand is usually in name. Fill it from there before grouping by brand.

deal_label values carry an emoji. One of the three labels is stored as "Mega Deal" followed by a megaphone. Strip it before you match on text.

seller_rating is blank on 129 offers. These are sellers without a published rating, not sellers rated zero.

delivery_message is relative text. "Get it by 3 Nov" only means something next to collected_at. In this snapshot, 47 of 50 featured UAE offers promised 3 November and 49 of 50 Saudi ones 7 November, two days against six. Parse it against the snapshot time, not today.

A sale was already running. The snapshot came before the main November sale, but 49 of the 100 products showed a sale label on the day. Treat the discount shares as a "before" picture taken during early promotions, not as an ordinary week.

Next steps with the full snapshot

Three questions are out of reach for 100 rows. The like-for-like UAE vs Saudi price gap, on every product listed in both markets rather than three. The featured-offer question for all 1,969 offers, by product type, to see where the default price is least reliable. And the same products again during the sale, to see how many discounts were real price cuts and how many came from a higher original_price.

The full noon UAE and Saudi Product Snapshot has 2,934 products and 1,969 seller offers in CSV or SQLite, with every column of the free files. A repeat snapshot during the sale, more categories or the other noon markets (Egypt, Kuwait, Qatar, Oman and Bahrain) are available on request. To get a copy or a custom cut, request a quote.

Top comments (0)