Narad Shopping indias 1st Live Price Comparison App

Now you can compare any product any time any where within indias no1 shopping site .

Price comparison is comparing the price of the same product in different outlets.

Price levels are based on price comparisons with major competitors.

Price comparison websites can take the hassle out of finding the best deal.

Price comparison is comparing the price of the same product in different outlets.

Price Comparison Engines sound very easy on paper and equally difficult to implement in practice. Developing sound, effective, and scalable price intelligence engines require us to tackle two broad challenges:

Large Scale Data Aggregation
Use ML/AI based normalization techniques to clean the data so that we are able to unearth meaningful insights from it
Let’s define data aggregation challenges, first, scale - in most of the business scenarios we need to acquire certainly a couple of million data points. Usually, for sourcing information, the price comparison platform has to integrate with several e-com marketplaces and continuously monitor pricing information on complete product categories to detect new information. According to some estimates, just Amazon(.com) has a product roster of 12 million products, if you add Amazon Marketplace sellers then it swells to staggering 350 million products.

Say if you want to compare one product category of 5000 products across 8 marketplaces every two hours, we are looking at ~500k crawls per day. There are numerous other challenges that a sourcing engine faces and they become even severe as you start operating on a large scale. Just hardware requirements of setting up a server farm would be huge. We have not even gone into language, ZIP codes, multi-page crawling and mobile apps coverage at all.

Usually, price comparison platforms are built to operate across complex web-environments that not only acquire data from multiple e-com marketplaces but also provide access to the historical store of all data acquired, enabling clients to derive unique, time series insights.

The second challenge is even bigger, how to make sense of the information once you acquire all data points from the Web. How do you process it?

The fact is that World-wide-web is not a uniform place, some of the websites are integration-friendly but most websites do not. They post information for visitors to read, not for a smart price comparison engine to track competitive prices. So, comparison engines need to be smart and require constant tweaking for crawling different websites.

It is extremely hard to integrate with websites at a scale as there are no universal standards for website development. MS vs Google war for browser usage is legendary and we don’t really have defined web standards, at least not practically.

Due to the humongous noise and lack of structure in Web data, we need to develop fairly advanced normalization techniques to clean the data, at times using AI for image and text analytics(Some startups focus only on image analytics from e-commerce websites). After organizing the data, we can think about unearthing meaningful information from it.

Each pricing intelligence tool has built upon complex machine-learning-based semantic models that help make sense of the data. However, pricing intelligence tools take a lot of time to mature and develop, once developed they provide brand managers with almost a simulator on real-time data and helps them undertake several business-critical abilities such as Dynamic Pricing, Measuring Price Elasticity, Deriving SKU level profitability and best of all, ability to compare all these metrics with competitors.