Why this study Who it's for Metrics The record Collection design What we publish Reading the numbers Mistakes FAQ

Retail shelf research · Study 08

The bottled water pack-size price curve

Near-identical product, unambiguous denominator, an eighty-fold pack range. The cleanest possible test of a price-normalisation pipeline — which is why this one runs first.

The pack-size price curve plots price per litre against pack size for packaged drinking water, from 250 ml bottles to 20-litre cans, across quick-commerce platforms and delivery zones. It is computed twice: on product price alone, and inclusive of delivery fees at a stated minimum basket.

Water is the ideal first study. The product is close to identical across brands, the denominator is unambiguous, MRPs are printed, and the pack ladder spans an eighty-fold volume range. Everything the collection pipeline has to prove gets tested here on easy mode.

Category: Packaged drinking water
Platforms: Blinkit, Zepto, Instamart, BigBasket, JioMart
Geography: 30 delivery zones, 10 cities
Cadence: Daily price capture
Window: 21 days
Findings: Added when the first window closes

Method · 2026-08-26Sampling frame, fields, cadence and metric definitions below are final. Findings for the price curve are added when the first collection window (21 days) closes.

Why this study

Why water is the right place to start, and what it actually shows

It tests the pipeline before the harder categories depend on it

Every study in this programme rests on the same machinery: pack-size parsing, unit normalisation, delivery-zone verification, brand entity resolution and price capture. If any of that is wrong, every later finding is wrong in a way that is hard to detect.

Water is where those components get proved. A near-identical product with a clean litre denominator means a parsing error shows up immediately as an implausible price per litre rather than hiding inside category variation.

The pack-size penalty is real and rarely quantified

Price per litre falls steeply as pack size rises, and the gap between the smallest bottle and the 20-litre can is far larger than most shoppers would guess. Everyone knows small packs cost more; very few published sources say by how much, per platform, in the same zone.

It is also a genuinely useful consumer-facing number, which makes it one of the more distributable findings in the programme.

Delivery fees change the answer

Comparing product price alone systematically understates the small-pack penalty, because the delivery fee is a fixed cost spread across whatever volume is in the basket. On a single small bottle it dominates.

We compute price per litre twice — bare and delivery-inclusive at a stated minimum basket — and publish both with the basket assumption printed next to the figure. One number without the assumption would not be interpretable.

The 20-litre can is a different product

The household can has its own availability pattern, its own delivery constraints and often its own pricing logic. Folding it into the same curve as bottled packs blends two businesses.

It is captured on the same schema and reported as a separate cut, with its own availability figure by zone.

Figure — the pack ladder, to scale

← swipe to see the full diagram

An eighty-fold volume range with an unambiguous denominator, which is exactly why this study runs first. A parsing error here shows up immediately as an implausible price per litre; in a more varied category it would hide inside genuine product variation.

Who this is for

Who reads this study, and what they do with it

Written from the questions we are actually asked when this category comes up on a scoping call. If none of these is you, the study is still readable — but the pilot offer at the bottom probably is not.

Pricing analyst, beverages

Packaged water · brand side

The problem

Your pack ladder was priced offline. You have no read on where your price per litre sits against competitors on the same app in the same pincode.

What this study gives them

The full price-per-litre curve bare and delivery-inclusive, plus platform spread on identical packs within the same capture hour.

Metric that moves

₹/L vs competitor set

E-commerce lead, beverages

Packaged water · channel

The problem

The 20 litre can is a different business from bottled packs and your reporting blends them into one availability number.

What this study gives them

Can availability reported separately by zone, with MRP gap tracked across the whole pack ladder.

Metric that moves

20L can zone coverage

Data or analytics lead evaluating vendors

Any category · buying-side evaluation

The problem

You are assessing whether a collection vendor's normalisation actually works before committing a harder category to them.

What this study gives them

Published parser accuracy against a labelled hold-out set, on the cleanest possible denominator. This study exists partly as that proof.

Metric that moves

Parser accuracy, published

Metric definitions

What we compute, and how

Definitions are fixed before collection begins. A metric defined after the data is in can be shaped to whatever conclusion is wanted, which is why these are published first.

Metric	How it is computed	Why it beats a market-share figure
Price per litre, bare	Selling price divided by total litres, where total litres equals pack volume times multipack count.`₹/L = price / (pack_litres × n)`	The core comparison. Pack price alone cannot compare a 250 ml bottle to a 20-litre can.
Price per litre, delivered	Selling price plus delivery fee at a stated minimum basket, divided by total litres.`₹/L = (price + fee) / total_litres`	Product price alone understates the small-pack penalty substantially. Both figures are published.
Small-pack penalty	Price per litre of the smallest pack divided by price per litre of the 20-litre can, in the same zone.`penalty = ₹/L(250ml) / ₹/L(20L)`	States the curve's steepness as a single multiple, which is the figure that travels.
Platform spread on identical packs	P90 minus P10 of price per litre for the same pack of the same brand, within one pincode and one capture hour.	Same product, same street, same hour — the cleanest price comparison this channel permits.
MRP gap	Displayed MRP minus selling price, divided by MRP, per SKU per zone.`gap = (mrp - price) / mrp`	Packaged water carries printed MRPs, which makes it one of the few categories where this is straightforwardly checkable.
Can availability by zone	Share of sampled zones where a 20-litre can is listed and in stock, reported separately from bottled packs.	The household staple has a distinct fulfilment pattern that the bottle curve would otherwise hide.

Data model

The record we collect

The simplest schema in the programme, deliberately. If this one does not produce clean price-per-litre values, none of the later studies will either.

water_price_record.json SCHEMA

{ "captured_at": "2026-09-02T11:00:18+05:30", "platform": "instamart", "city": "Pune", "pincode": "411001", "brand": "<resolved_brand>", "sku_title": "<as_listed>", "pack_ladder": "1l", // 250ml | 500ml | 1l | 2l | 5l | 20l "pack_litres": 1.0, "multipack_n": 6, // 1 unless bundled "total_litres": 6.0, // derived "mrp_printed": null, "selling_price": null, "delivery_fee": null, // at stated minimum basket "min_basket_value": null, "price_per_litre_bare": null, // derived "price_per_litre_delivered": null, // derived "stock_status": "listed_in_stock", "capture_id": "<uuid>" }

Schema shape — values null until the window runsPilot study · proves the pipeline

Values are shown as null because the window has not run. Derived fields are marked as such — they are computed from captured values, never scraped from a displayed badge.

Collection design

What we capture, and how often

Cadence follows a tiered design: highest frequency on the fields where a change alters a decision, lower on the ones that move slowly.

Field or signal	Why this study needs it	Capture frequency
Selling price by zone	The numerator. Captured at a fixed daily time so cross-platform comparison is like for like.	Daily, 11:00 IST
Printed MRP	Packaged water carries printed MRPs, making the MRP gap directly checkable in a way most categories do not allow.	Daily
Pack litres and multipack count	Bundled packs stated as six times one litre must resolve to six litres, or the whole curve distorts.	Every capture
Delivery fee and minimum basket	Both, because a delivery-inclusive price per litre is uninterpretable without the basket it assumes.	Daily
Pack ladder classification	Fixed bands from 250 ml to 20 L so the curve has stable x-axis positions across brands.	Derived
Stock status	Particularly for the 20-litre can, whose availability pattern differs from bottled packs.	Daily
Brand, resolved and raw	Both retained. Water brand naming is inconsistent across platforms and needs entity resolution.	Every capture
Parser confidence	A per-row confidence on pack parsing. This study is where parser accuracy gets measured for the whole programme.	Every capture

Planned output

What will be published when the window closes

Decided before collection starts, so the sample can be designed to support them rather than reverse-engineered to fit whatever came back.

01The price-per-litre curveMedian price per litre at each rung of the pack ladder, from 250 ml to the 20-litre can.

02Bare versus deliveredThe same curve computed with and without delivery fees, with the basket assumption printed alongside.

03The small-pack penaltyThe multiple between the smallest pack and the 20-litre can, stated per platform.

04Platform spread on identical packsThe rupee gap for the same pack of the same brand in the same pincode within the same hour.

05MRP versus selling priceWhere selling price sits against printed MRP across the pack ladder.

0620-litre can availability by zoneReported separately, since the household can behaves differently from bottled packs.

07Parser accuracy notePack-parsing accuracy measured against a labelled hold-out set, published because every later study depends on it.

The figure this study is designed to produce

How many times more a litre of water costs when bought in 250 ml bottles than in a 20-litre can — before and after delivery fees.

A clean, consumer-legible number that also proves the normalisation pipeline every later study depends on.

How to read it

How to read the numbers when they land

When the findings land on this page, three qualifiers will sit next to every figure, and they are worth understanding before you read any of them.

The sample is a panel, not a census

Thirty delivery zones across ten cities is a deliberately chosen panel weighted toward metros. It is not India. Every figure on this page will be phrased as "across 30 sampled zones in 10 cities", never as "in India", because the narrower claim is the one the data actually supports.

The denominator is stated, always

Availability figures are computed against SKUs that were listed in that zone, never against the full catalogue. A SKU that was never ranged in a zone is a range decision, not a stock-out, and merging the two produces unavailability numbers that send supply chain teams after problems that do not exist.

Coverage gaps are published, not hidden

Each finding carries the successful-capture rate for its platform and week. Where a platform's coverage dropped below the threshold in a given week, that week is excluded and the exclusion is noted. A study that reports no gaps is a study that did not look for them.

If a figure on this page is ever wrong, tell us and we will correct it visibly with the date of the correction. Raw payloads are retained with capture IDs precisely so that any published number can be traced back to the observations behind it.

What goes wrong

Measurement mistakes that distort a price curve

Each of these produces a plausible-looking number that is wrong in a direction the reader cannot detect. They are listed because the design above exists specifically to avoid them.

Ignoring multipack counts

A listing stated as six times one litre is six litres. Treating it as one collapses the curve for exactly the packs that matter most.

Comparing product price without delivery

The delivery fee is fixed and the volume is not. On a single small bottle the fee dominates, and excluding it understates the penalty badly.

Publishing a delivered price without the basket assumption

Delivery-inclusive price per litre depends entirely on the assumed basket. Without it stated, the number cannot be interpreted at all.

Folding the 20-litre can into the bottle curve

The household can has its own availability and pricing logic. Blending it into the curve mixes two businesses in one line.

Skipping parser validation

Water is where pack parsing gets measured against a labelled set. Skipping it here means every later study inherits an unmeasured error rate.

FAQ

Questions about this study

Including why there are no figures on it yet.

Because it is the cleanest possible test of the collection pipeline. Near-identical products, an unambiguous litre denominator, printed MRPs, and a pack ladder spanning an eighty-fold volume range.

A parsing error in this category shows up immediately as an implausible price per litre. In a more varied category it would hide inside genuine product variation.

Because product price alone understates the small-pack penalty substantially. The delivery fee is a fixed cost spread over whatever volume is in the basket, and on a single small bottle it dominates.

Both figures are published, and the delivered one always carries the minimum basket it assumes. Without that assumption stated, the number is not interpretable.

Pack volume times multipack count gives total litres, and that is the denominator. A listing stated as six times one litre resolves to six litres.

Getting this wrong is the single most common way a price curve gets distorted, which is another reason this study runs first.

It is captured on the same schema and shown as the last rung of the curve, but its availability is reported separately.

The household can has different delivery constraints and often different pricing logic from bottled packs, so folding its availability into a bottle figure would blend two distinct businesses.

Yes — the same pack-ladder approach works for juices, soft drinks, edible oils and any category sold across a wide pack range.

Water is simply the cleanest place to establish the method. Once the parser and the normalisation rules are validated here, applying them elsewhere is configuration rather than new engineering.

Run this design on your own category

Our aggregate view is context. Your brands, your competitive set and your priority pincodes are what change a decision. A pilot returns real data within 24 hours; production collection goes live in 5 to 10 business days.

Get a free pilot Book a 20-min scoping call

Free pilot, no card, no obligation. You keep the sample data either way.

Hear It Directly from Our Clients

Watch how businesses like yours are using Actowiz data to drive growth.

▶

1 min

★★★★★

"Actowiz Solutions offered exceptional support with transparency and guidance throughout. Anna and Saga made the process easy for a non-technical user like me. Great service, fair pricing!"

Thomas Galido

Co-Founder / Head of Product at Upright Data Inc.

▶

2 min

★★★★★

"Actowiz delivered impeccable results for our company. Their team ensured data accuracy and on-time delivery. The competitive intelligence completely transformed our pricing strategy."

Iulen Ibanez

CEO / Datacy.es

▶

1:30

★★★★★

"What impressed me most was the speed — we went from requirement to production data in under 48 hours. The API integration was seamless and the support team is always responsive."

Febbin Chacko

-Fin, Small Business Owner