Automated deal alerts across 3,000 Myntra products

Client Overview

Kapil Mital runs an online reselling business. In reselling, the margin is decided at the moment of purchase, so his real edge comes from buying at the lowest possible price and moving product on before the window closes. Much of his sourcing runs through Myntra, one of India's largest fashion and lifestyle marketplaces, where the items he cares about cycle through constant price swings and stock changes.

That is a good place to source from and a hard place to watch. A flash discount or a restock on a fast moving product can surface and vanish within the same hour. For a reseller, that timing is the whole game. Kapil came to DataScrape Solutions because the manual routine he had built around Myntra was no longer keeping pace with the platform, and every deal he missed was money left on the table.

The Challenge

On Myntra, the best sale prices and restocks sell out within minutes. Kapil was tracking around 3,000 products that mattered to his business, and the only monitoring tool he had was his own attention. He was refreshing thousands of product pages by hand, tab after tab, hour after hour, trying to catch the exact moment a price dropped or a sold out item came back into stock.

It did not scale, and it could not. There are only so many pages one person can open in a day, and the window on a good deal is short. The second he looked away from a page was often the second the discount landed on it. He was not missing deals because he did not know where to look. He was missing them because no human can watch three thousand pages at once and still react within the few minutes a price stays low.

The problem had two halves that pull against each other: coverage across a large product list, and speed on any single item, both held continuously, without a person sitting in the loop. Solving one at the cost of the other was not good enough. He needed both at the same time.

How We Built It

We built a Telegram bot that does the watching for him. The goal was easy to state and demanding to deliver: every product on his list, checked often enough that a change never sits unnoticed for long, with the alert reaching him fast enough to still act on it.

The bot works through Kapil's list of roughly 3,000 Myntra products on a fixed cadence. Every 10 minutes it re-checks each product on the list. For every item it reads the current price and the current stock status, then compares that reading against the last known state it holds on record. When nothing has changed, it stays silent. When a price moves, or an out of stock product comes back into stock, it fires an alert immediately.

We chose Telegram as the delivery channel on purpose. The alert lands on Kapil's phone the instant a change is detected, in an app he already keeps open through the day, so there is nothing new for him to log into or monitor. Each alert tells him what changed and points him straight at the product, which turns the path from notification to action into a matter of seconds rather than a fresh hunt across the site.

Holding a 10 minute cycle steady across 3,000 products is where the engineering actually lives. The checks have to be paced so the full list finishes inside the window and the next pass starts clean, and the bot has to stay quiet when nothing has changed so that the real alerts are never buried under noise. The whole system was built and tested in 5 days, and it kept evolving as Kapil refined exactly which signals mattered to him and how he wanted them delivered.

The Results

Kapil went from manually refreshing thousands of pages to letting a bot carry the entire watch. The system monitors around 3,000 Myntra products, re-checks each one every 10 minutes for price changes and restocks, and sends an instant Telegram alert the moment something moves. Instead of hunting for a deal and hoping he arrives before it sells out, he now reacts to a deal that has been handed to him, and he acts on it in seconds.

The practical shift is coverage he could never reach by hand. Three thousand products stay under continuous watch at once, and none of them depend on him remembering to check at the right minute. The short window that used to work against him now works for him, because the bot is already looking when a price drops or a sold out item returns, and it never looks away.

The full build took 5 days from start to a tested, working bot.