How a Mobile Brand Prepared for Big Billion Days Using Real-Time Flipkart Price Feeds

The Client

A mobile handset and accessories brand with a mid-market portfolio, selling on Flipkart through two authorised sellers. Flipkart accounted for a substantial share of the brand's online volume, and Big Billion Days accounted for a disproportionate share of that.

Client details are anonymised. Figures are representative of the engagement.

The Problem: A Post-Mortem Nobody Could Explain

The brand came to us after a Big Billion Days that had gone badly, and — worse — gone badly in a way nobody could account for.

The numbers were not catastrophic. Volume was roughly flat against the previous year in a category that had grown. Margin was down. The brand's hero SKU had sold out on day two, which the team initially read as a success until they worked out that it had sold out at a price several hundred rupees below where they had planned to land.

What nobody could explain was why. The pricing team had approved a sale price. That price had been undercut. Someone had responded. The response had happened. But when the post-mortem team tried to reconstruct the sequence — who moved, when, and how long the brand had been exposed — they found they could not. The data did not exist.

Their monitoring pipeline captured once a day, at 06:00. Big Billion Days ran for six days. They had six rows per SKU covering a sale in which competitors had, as it later emerged, repriced dozens of times.

The pricing lead's summary, which we have since quoted with permission in more than one conversation: we have six photographs of a car crash and no video.

Why the Existing Setup Could Not Work

The problem was not that the pipeline was broken. It ran perfectly. It was simply designed for a different job.

Daily capture in a sale that moves hourly. A once-daily snapshot during Big Billion Days is not low-resolution data. It is a different kind of data — it tells you the state of the world at 06:00 and nothing about the twenty-three hours in between. During a normal week, that is fine. During a sale where deals open and close on the hour, it is close to useless.

Listed price only. The pipeline captured the listed price. It did not capture deal type, deal windows, bank offers, or the effective price. The brand had benchmarked on the layer that moved least.

No stock capture. The pipeline recorded whether the product page existed. It did not record whether the hero variant was purchasable. The day-two stockout appeared in the data only as an absence.

No alerting. Even the daily data went into a dashboard. Nobody was watching the dashboard at 02:00.

No pre-sale baseline. Capture began the day the sale began. When the team later tried to answer whether their discount had been real — whether the sale price was genuinely below the trailing street price — they had no trailing street price to compare against.

The Solution: Tiered Real-Time Flipkart Price Feeds

Product Data Scrape deployed a sale-event monitoring configuration built specifically around the failure modes above, starting thirty days before the next Big Billion Days.

Pre-sale baseline (T-30 to T-1). Daily capture across the brand's SKU universe and a matched competitor basket, establishing a trailing 30-day median price per SKU. This became the reference against which every sale-period discount was measured — including, importantly, the competitors' discounts.

Tiered capture during the sale:

Full-stack capture: listed price, Plus exclusive price, deal type, deal price, deal window, structured bank offers with caps and thresholds, no-cost EMI terms, exchange valuation, SuperCoin earn rate, computed best-case effective price, per-variant stock and stock signal, full seller array.

Three alert rules, with owners:

• Any competitor hero SKU crosses below our effective price → alert to pricing lead.
• Any brand hero variant hits a low-stock signal → alert to supply lead.
• Any brand hero variant goes out of stock → alert to both, escalated.

Rehearsal at T-7. The full sale configuration was run for 24 hours to confirm capacity, parsing, alert routing, and ownership. Two problems surfaced during the rehearsal — an alert channel routing to an unmonitored distribution list, and a competitor SKU mapped to the wrong internal reference — both of which would have been discovered mid-sale otherwise.

Sample Data: The Alert That Paid for the Engagement

An illustrative capture series from day one of the sale.

Illustrative series.

The competitor undercut at 02:15. The alert fired. The pricing lead — who had a defined response mandate agreed before the sale, allowing a re-price within a pre-approved band without escalation — responded at 02:37.

Twenty-two minutes.

In the previous year, the equivalent gap had been measured, after the fact, at approximately nineteen hours.

The 14:20 low-stock alert is the second half of the story. Under the old setup, the hero variant would have sold out and the brand would have discovered it the next morning. Instead, held allocation was released within 35 minutes, and the SKU remained purchasable through the peak.

What Changed Operationally

A pre-approved response band. The single highest-leverage change was not technical. Before the sale, the pricing team agreed a band within which the pricing lead could re-price without escalation. Without it, a 22-minute alert would have produced a 22-minute alert and a six-hour approval chain.

Named owners on every alert. Not a channel. A person.

Effective price as the benchmark. The team stopped comparing listed prices. Twice during the sale, a competitor's listed price dropped and the effective price did not move — the competitor had simply shifted the discount from the offer layer to the price layer. Under the old benchmark, the brand would have chased both moves. Under the new one, it correctly did nothing.

Stockout treated as a competitive event. Escalated to the same channel, with the same urgency, as a competitor price cut.

The Results

Figures are representative of the engagement outcome.

Revenue rose 27 percent and margin rose 14 percent in the same event — a combination that is unusual, because the ordinary way to grow sale revenue is to buy it with margin.

The mechanism is not mysterious. The brand did not discount more deeply. It discounted later, less often, and only when it actually needed to — because for the first time it knew when it needed to.

The Lesson

Every brand in this category knows Big Billion Days is won on price. Far fewer have understood that it is won on price latency.

The competitor in this story was not smarter. They were simply faster, and for a year they had been faster for free, because the brand had no way of knowing when they moved. Sixty-one hours of unopposed undercutting is not a pricing failure. It is a measurement failure that presents as a pricing failure.

The brand's summary, after the second sale: we thought we needed a better price. We needed a faster clock.

Work With Product Data Scrape

Product Data Scrape builds real-time Flipkart price feeds for sale events — 15-minute capture on hero SKUs, deal types and windows, structured bank offers, computed effective price, per-variant stock with live stockout alerting, full seller arrays, and pre-sale baseline capture from T-30.

Big Billion Days is a week. Preparation is a month. If the next one matters to your year, talk to our team before the calendar closes.

Product Data Scrape — turning marketplace complexity into decision-ready data.