Executive Summary
Every year, the headline figures from India's largest sale event describe discounts of seventy or eighty percent. Every year, brands plan against those figures. And every year, a gap opens between what the headline says and what the transaction data shows.
The gap is not deception. It is measurement.
This study examines Big Billion Days discount data across a monitored panel of SKUs and finds that the single largest determinant of "how deep was the discount" is not the sale price at all. It is the reference price you measure against — and the two reference prices in common use produce numbers that differ by a factor large enough to change a planning decision.
The study also examines two dimensions that receive far less attention than discount depth and, in our view, matter more: how fast prices move during the sale, and how early hero SKUs run out.
Headline findings:
Discount depth is a function of your reference price. Measured against MRP, discounts look large. Measured against the 30-day trailing median street price, they are substantially smaller — and the gap between the two measures is itself the most useful number in the dataset.
Repricing frequency rises by roughly an order of magnitude during the sale versus baseline, and clusters in identifiable windows.
Stockout is the decisive competitive event, and it arrives earlier than most planning assumes.
This report is published by Product Data Scrape. Figures are representative of observed patterns across our Flipkart monitoring panel and are illustrative rather than a market census.
1. Methodology
- Panel: Monitored high-velocity SKUs across Mobiles, Large Appliances, and Small Appliances, together with a matched competitor basket.
- Capture window: T-30 through T+7 around the sale — the pre-sale baseline is essential and is the component most monitoring programmes omit.
- Frequency: Daily during the baseline window; 15-minute capture on hero SKUs during the sale.
- Fields: Listed price, Plus price, deal type, deal price, deal window, structured bank offers, computed effective price, per-variant stock and stock signal, full seller array.
- Reference prices computed: (a) MRP; (b) the 30-day trailing median of the actual listed price prior to the sale — what we call the street price.
2. Finding One: Discount Depth Depends Entirely on the Reference Price
This is the finding that reframes everything else, and it is a methodology finding rather than an accusation.
A discount percentage is a ratio. The numerator is well defined. The denominator is a choice — and the two available choices produce very different answers.
| Category |
Median Discount vs MRP |
Median Discount vs 30-Day Street Price |
Gap Between the Two Measures |
| Mobiles |
~43% |
~11% |
32 pts |
| Large Appliances |
~48% |
~14% |
34 pts |
| Small Appliances |
~39% |
~9% |
30 pts |
Illustrative figures from the monitoring panel.
Both columns are arithmetically correct. They answer different questions.
The MRP-based figure answers: how far below the manufacturer's stated maximum is this price? For most categories, and for most of the year, that question has very little bearing on anything, because almost nothing sells at MRP.
The street-price figure answers: how much cheaper is this than what I would have paid three weeks ago? That is the question a customer is actually asking, and it is the question a brand planning its own sale depth should be asking too.
Implication: if you plan your sale discount against MRP and your competitor plans theirs against street price, you are not running the same calculation, and you will systematically over-discount. The single most valuable line item in a Big Billion Days dataset is not the sale price. It is the T-30 baseline — and it is the one that almost nobody collects, because you cannot go back and get it once the sale has started.
3. Finding Two: Repricing Frequency Rises by an Order of Magnitude
| Period |
Median Repricing Events per Hero SKU |
Peak Day |
| Baseline (ordinary week) |
~0.9 per week |
— |
| Sale week |
~4.1 per day |
Day 1 |
Illustrative figures.
Roughly a thirtyfold increase in the rate of price movement.
Repricing clustered in identifiable windows. Two were consistent across the panel:
The sale-open window (first 6 hours). The heaviest concentration of movement in the entire event, as sellers discover each other's opening positions and adjust.
The late-evening window. A second, smaller cluster corresponding to peak consumer browsing.
Implication: a monitoring programme capturing once daily observes roughly one of every four price movements, and observes none of them within a window in which a response would have mattered. During the sale-open window specifically, a daily-capture programme is not slow. It is blind.
4. Finding Three: Stockout Arrives Earlier Than Planning Assumes
Discount depth gets the attention. Stockout decides the outcome.
| Metric |
Observed (illustrative) |
| Share of hero variants that went out of stock at some point during the sale |
~38% |
| Median time to first stockout on those variants |
Inside the first 48 hours |
| Share of stockouts occurring on day 1 |
~29% |
| Median duration of a stockout before restock |
Substantial — often the balance of the day |
Illustrative figures.
The structural point: a stockout on day one is not a lost day. It is a lost sale.
Once a hero variant is unavailable, every subsequent competitor price move is uncontested. The brand is no longer being outcompeted; it has left the field. And the demand does not wait — it converts to a competitor and does not come back when stock returns.
We also observed the reverse pattern in the seller array: on listings where the brand's authorised seller went out of stock, the default position frequently shifted to a lower-rated, non-F-Assured seller. So the day-one stockout does not merely cost volume. It hands the listing to exactly the seller a brand-protection team spends the rest of the year trying to remove.
Implication: stockout should be alerted in real time, with the same urgency as a competitor price cut, and it should be escalated to a supply owner who has pre-approved authority to release held allocation. A stockout discovered in the next morning's dashboard is a stockout that has already done its damage.
5. Finding Four: Discount Depth Tracks Seller Count
Segmenting by the number of sellers on a listing produced a clean and unsurprising relationship.
| Sellers on Listing |
Median Discount vs Street Price |
| 1–2 |
~6% |
| 3–5 |
~11% |
| 6+ |
~17% |
Illustrative figures.
Price competition requires competitors. Where seller count is thin, discount depth is thin — regardless of what the sale banner says.
Implication: this is a distribution lever, not a pricing lever. A brand that wants a deeper competitive discount on its own listings without funding it directly should be recruiting authorised sellers, not cutting price. Sale-period discount depth is downstream of seller count, and seller count is a decision the brand controls.
6. Finding Five: The Effective-Price Gap Widens Through the Event
The gap between listed price and best-case effective price widened materially over the course of the sale, as bank offers deepened, exchange ceilings rose, and SuperCoin earn rates increased.
| Sale Day |
Median Listed-to-Effective Gap |
| T-1 (pre-sale) |
~7% |
| Day 1 |
~13% |
| Day 3 |
~16% |
| Day 6 |
~19% |
Illustrative figures.
Implication: a brand benchmarking on listed price is, by day six, watching a number that has diverged from the transaction price by nearly a fifth — and diverging further every day. Listed-price benchmarking is at its least reliable at precisely the point in the year when pricing decisions carry the most weight.
7. What Brands Should Change
Capture the T-30 baseline. It is cheap, and it is the only thing that makes the sale data interpretable. It cannot be backfilled.
Report discount against street price, not MRP. Report both if you must, but decide against street price.
Raise capture frequency to match repricing frequency. Roughly four movements per day per hero SKU means daily capture is not a resolution problem; it is a blindness problem.
Alert on stockout as a competitive event. Real time, named owner, pre-approved allocation release.
Benchmark on effective price throughout. The listed-to-effective gap widens every day of the sale.
Recruit sellers to deepen discounts you do not fund. Seller count and discount depth move together.
8. Limitations
Findings reflect a monitored panel rather than a platform census, and category composition materially affects every figure presented. Reference-price computation depends on the completeness of the pre-sale baseline. Sale mechanics, deal structures, and offer intensity vary between events and between years, so figures should be read as directional patterns rather than fixed constants. All figures are illustrative of observed behaviour, not audited market statistics.
9. About the Data
This report was produced using Big Billion Days discount data collected by Product Data Scrape. Our sale-event Flipkart datasets capture deal type and deal windows, listed and Plus pricing, structured bank offers with caps and thresholds, no-cost EMI and exchange terms, SuperCoin earn rates, computed effective price, per-variant stock with stock signals, and the full multi-seller array — at capture frequencies down to 15 minutes on hero SKUs, with pre-sale baseline capture from T-30.
Delivered as JSON, CSV, via REST API, or pushed directly to cloud storage and data warehouses.
Want this analysis run on your own catalogue before the next sale? Product Data Scrape will build the T-30 baseline on your SKUs and your competitors' SKUs, so that when the event opens you are measuring against something real.
Product Data Scrape — turning marketplace complexity into decision-ready data.
Singapore
EMAIL
PHONE
.webp)
.webp "thumb")
.webp "thumb")
.webp "thumb")
.webp)
