Nykaa System Design Interview: Omnichannel Beauty Commerce
Nykaa sells beauty in two ways at once: it buys and holds its own beauty stock to guarantee authenticity, and it runs a marketplace for fashion, across a website, an app, and 237 physical stores whose inventory all has to stay in sync. It reported more than 40 million monthly active users, and on a single Pink Friday sale day in 2022 it recorded around 800,000 orders.
Designing Nykaa is an omnichannel commerce problem with a beauty-specific twist. Unlike a pure marketplace, Nykaa buys, holds, and curates its own beauty inventory to control authenticity, which matters in a category full of counterfeits, while running a marketplace model for fashion. It sells through a website, an app, and hundreds of physical stores, so a single view of inventory across warehouses and stores is central. Discovery is content-led rather than purely search-led, and the whole business runs on a unified data platform. This walkthrough covers the hybrid retail model, omnichannel inventory, content-driven discovery, and the data platform Nykaa has published, and is honest that Nykaa releases little about its transactional serving stack, so those parts are described as the standard pattern.
Asked at: Commonly asked at Nykaa, other retail and e-commerce companies, and omnichannel and inventory teams, and the general forms, meaning design an e-commerce platform, an inventory and order-management system, or an omnichannel retailer, show up at most product companies for SDE2 and SDE3 rounds. Nykaa is a good question because it forces the inventory-led and omnichannel angles that a marketplace-only design skips, plus authenticity and content-led discovery.
Why this question is asked
Most e-commerce answers assume a marketplace: sellers list, the platform is asset-light, and inventory is someone else's problem. Nykaa breaks that, because for beauty it holds its own stock, which turns inventory management, warehousing, and authenticity into first-class design problems rather than seller concerns. It is also omnichannel, selling online and through hundreds of stores, so keeping one accurate view of stock across all locations, and fulfilling from the right one, is a real distributed-inventory challenge. And discovery leans on content, editorial, tutorials, and curation, more than on a search box. Interviewers use Nykaa to check whether you can design inventory and order management for an inventory-led retailer, reason about omnichannel stock synchronization, and handle a large sale-day spike, while being honest about which pieces are standard commerce plumbing.
Requirements
Always clarify these in the first 5 minutes of the interview. Do not start drawing boxes until both lists are agreed.
Functional requirements
- Customer browses beauty and fashion products across a website and an app, guided by content, curation, and search
- Product pages show authentic stock, price, and availability, including availability at nearby physical stores
- Customer adds to cart and orders, paying by UPI, card, wallet, or cash on delivery, and can return items
- For beauty, Nykaa manages its own inventory: buying, holding, and curating stock; for fashion it runs a marketplace of sellers
- Inventory is kept consistent across warehouses and 237 physical stores, so an item is not sold when it is out of stock
- Orders can be fulfilled from a warehouse or a store, and customers can buy across channels
- Editorial content, tutorials, and recommendations drive product discovery
- Absorb large traffic and order surges during sale events such as Pink Friday
Non-functional requirements
- Accurate, near-real-time inventory across all warehouses and stores, so overselling is avoided
- Fast, cacheable product and content pages, since browsing and content drive discovery
- Authenticity control for beauty, which the inventory-led model exists to provide
- Absorb sale-day spikes many times normal load, on both the browse and the order paths
- High availability on the storefront and checkout during sales
- One consistent view of data across the business for analytics and personalization
- Support an omnichannel experience where online and store inventory and orders are integrated
Back-of-envelope scale estimates
Show your math. Pulling numbers from thin air signals you have not thought about the load.
Monthly active users
40M+
Nykaa, via its AWS case study, reported more than 40 million monthly active users. This is the browse-and-discovery load the storefront must serve.
Annual transacting customers
~19M (FY25)
Nykaa reported over 19 million annual unique transacting customers in FY25, on a cumulative base of about 42 million. These are the buyers behind the order volume.
Gross merchandise value
~15,604 crore rupees (FY25)
Nykaa reported total gross merchandise value of about 15,604 crore rupees in FY25, with beauty roughly three-quarters of it. Nykaa is listed, so its financials are well reported.
Physical stores
237 stores, 79 cities
Nykaa reported 237 physical stores across 79 cities as of the first quarter of FY26. These stores are why omnichannel inventory is central to the design.
Sale-day peak
~800K orders in a day (Pink Friday 2022)
On the first day of the Pink Friday 2022 sale, Nykaa reported around 800,000 orders at more than 400 orders a minute, with about 10 million visits. Recent editions have not published comparable operational figures, so treat this as the well-documented sale-day peak.
Data platform
400+ TB, 10,000 ingestion jobs
Nykaa, via its AWS case study, reported consolidating over 400 terabytes into a unified data platform with about 10,000 ingestion jobs and more than 1,500 customer datasets. This is the analytics and personalization backbone.
High-level architecture
Design Nykaa around two things it does that a marketplace does not: it holds its own beauty inventory, and it sells across online and hundreds of physical stores. Those two facts drive the interesting parts. Nykaa publishes little about its transactional serving stack, so the storefront and checkout are described as the standard e-commerce pattern, while the inventory model, the omnichannel angle, and the data platform, which are documented or clearly reported, get the focus. The retail model is hybrid. For beauty and personal care, Nykaa is inventory-led: it buys stock from brands, holds it in its own warehouses, curates what it sells, and ships authentic product. This exists because beauty is a counterfeit-prone category and controlling the supply chain is how Nykaa guarantees authenticity. For fashion, Nykaa runs a marketplace, where third-party sellers list and fulfill. So the system actually has two inventory models under one storefront: owned stock that Nykaa manages end to end, and marketplace stock that sellers own, and the order and fulfillment flows differ between them. Because Nykaa is omnichannel, the hardest ongoing problem is a single, accurate view of inventory across many locations: central warehouses and 237 physical stores. When a customer views a product, the system needs to know what is truly in stock and where, so it does not sell an item that a store just sold at the counter. Orders can be fulfilled from a warehouse or from a nearby store, and stock reservations have to be atomic so two orders do not claim the same unit. This distributed-inventory synchronization, keeping store point-of-sale and online stock in agreement in near real time, is the core omnichannel challenge, and it is the standard pattern for an inventory-led omnichannel retailer. The storefront itself is a read-heavy browse and content experience. Product pages, editorial content, tutorials, and curated lists drive discovery more than a search box does, so those pages are cached hard and served fast, with a search and recommendation layer on top. Nykaa runs machine-learning ranking and personalization, confirmed by its hiring for learning-to-rank and recommendation work, though it has not published the internals. The checkout and order path is the familiar pattern: cart, idempotent payment capture, an order service with a state machine, and returns, which are especially important in beauty and fashion. Underneath everything sits a unified data platform, which Nykaa has published through an AWS case study. Called One Nykaa, One Data, it consolidated more than 400 terabytes into a lakehouse on AWS, using S3 for storage, EMR with autoscaling for processing, Redshift and Athena for querying, and Lake Formation for governance, cutting report generation from up to two weeks to a few hours and moving data refreshes from daily to hourly. This is the backbone for analytics and personalization across the business.
In a real interview, sketch this on the whiteboard before diving into any single box.
Core components
Walk through each service. The interviewer wants to hear what each one owns, not just the names.
Storefront and content
The read-heavy browse experience across web and app: product pages, editorial content, tutorials, and curated lists that drive discovery. Cached aggressively and served fast, because content-led browsing, not just search, is how Nykaa customers find products.
Search, ranking, and personalization
The layer that ranks and recommends products for a customer. Nykaa runs machine-learning ranking and personalization, balancing intent, history, and real-time signals, confirmed by its hiring for learning-to-rank and recommendation work, though the internals are not published. Described here at the level of how it fits, not as a Nykaa-published design.
Inventory service (owned and marketplace)
The core of the hybrid model. For beauty, it tracks Nykaa's own stock across warehouses; for fashion, it tracks marketplace sellers' stock. It keeps availability accurate, reserves units atomically at checkout, and is the source of truth for what can be sold. The two models have different ownership and fulfillment flows behind one storefront.
Omnichannel inventory synchronization
The system that keeps one view of stock across central warehouses and 237 physical stores in near real time, reconciling store point-of-sale with online availability so an item is not oversold. It also decides which location fulfills an order. This distributed-inventory synchronization is the central omnichannel challenge.
Order, cart, and payment services
The standard e-commerce write path: cart, a checkout that captures payment idempotently across UPI, card, wallet, and cash on delivery, an order service with a state machine, and returns, which matter a lot in beauty and fashion. Nykaa has not published these internals, so they are the general pattern.
Fulfillment and logistics
The machinery that picks, packs, and ships an order from the chosen warehouse or store, and handles returns back into inventory. For owned beauty stock this is Nykaa's own operation; for marketplace fashion, sellers fulfill. Store fulfillment ties back into the omnichannel inventory view.
Unified data platform (One Nykaa, One Data)
Nykaa's published lakehouse on AWS, consolidating over 400 terabytes with S3 for storage, EMR for processing, Redshift and Athena for querying, and Lake Formation for governance. It powers analytics and personalization across the business and cut reporting time from weeks to hours.
Data model
Pick the right store per table. Justify each choice with the access pattern, not by reflex.
productsproduct_id (PK)brand_idcategory (beauty|fashion)model (owned|marketplace)price_paiseattributes (jsonb)The catalog listing. The model field distinguishes owned beauty stock from marketplace fashion, which changes how inventory and fulfillment work. Read-heavy and cached.
inventorysku_id (PK)location_idlocation_type (warehouse|store)available_qtyreserved_qtyupdated_atStock per SKU per location, across warehouses and 237 stores. available and reserved are kept accurate so checkout can reserve atomically and the storefront never oversells. The heart of the omnichannel problem.
storesstore_id (PK)format (luxe|on_trend|kiosk)citygeostatusThe physical store network, in several formats. Stores both sell at the counter and can fulfill online orders, so their stock feeds the unified inventory view.
ordersorder_id (PK)customer_iditems (jsonb)fulfillment_location_idamount_paisepayment_idstateThe transactional order. fulfillment_location_id records which warehouse or store ships it. Standard order state machine, with returns as an important branch for beauty and fashion.
customerscustomer_id (PK)channels_usedpreferencescreated_atThe shopper across channels. Preferences and behavior feed personalization. Over 19 million transacting a year on a base of about 42 million.
contentcontent_id (PK)type (article|tutorial|list)linked_products[]published_atEditorial content, tutorials, and curated lists that drive discovery, linked to the products they feature. A first-class part of the storefront, not an afterthought.
Deep dives
These are the conversations the interviewer is steering you toward. Practice each one until you can talk through it without notes.
The hybrid model: inventory-led beauty and marketplace fashion
Nykaa runs two retail models under one roof, and the reason is authenticity. Beauty is a category where counterfeits are common and a fake product can harm a customer and the brand, so for beauty Nykaa is inventory-led: it buys stock directly from brands, holds it in its own warehouses, curates the assortment, and ships authentic product, controlling the whole chain. For fashion, where authenticity is less fraught and assortment is vast, it runs a marketplace where third-party sellers list and fulfill. The design consequence is that the system has two inventory and fulfillment flows behind one storefront: owned stock that Nykaa manages end to end, with full control and full responsibility, and marketplace stock that sellers own. Recognizing that a single product catalog sits on top of two different ownership and fulfillment models, and that beauty's authenticity requirement is why the inventory-led model exists, is the distinctive insight for a Nykaa answer.
Omnichannel inventory: one truth across warehouses and stores
Because Nykaa sells online and through 237 physical stores, its hardest ongoing problem is keeping one accurate view of stock across every location. A unit sold at a store counter has to stop being available online almost immediately, or the site will oversell it, and an online order has to be fulfillable from whichever warehouse or store makes sense. This is a distributed-inventory synchronization problem. The standard way to handle it, which is what to describe since Nykaa has not published its own, is a central inventory service that is the source of truth, fed by near-real-time updates from store point-of-sale systems and warehouses, with atomic reservations at checkout so two orders cannot claim the same unit. Fulfillment routing then picks the best location, a warehouse or a nearby store, to ship from. The trade to discuss is freshness versus cost: perfectly real-time global inventory is expensive, so systems often accept small windows and use reservations and reconciliation to prevent overselling, which is the crux of omnichannel retail.
Content-led discovery for beauty
Nykaa is known for discovery driven by content, editorial articles, tutorials, and curated lists, more than by a plain search box, because beauty shoppers often want guidance on what to buy rather than searching for a known product. That makes content a first-class part of the storefront: articles and tutorials are linked to the products they feature, and browsing them is a primary path to a purchase. From a systems view, this means the content experience must be fast and cacheable at the same scale as the product catalog, and the recommendation layer has to blend content and products. Nykaa does run machine-learning ranking and personalization, which its hiring for learning-to-rank and recommendation roles confirms, though it has not published the internals. The interview point is that when content drives discovery, the content system is part of the commerce hot path, not a separate marketing site.
Surviving the Pink Friday sale spike
Nykaa's biggest sale, Pink Friday, produces a large, concentrated spike. On the first day of the 2022 edition it reported around 800,000 orders at more than 400 a minute, with about 10 million visits and a revenue jump many times a normal day. Absorbing that hits both the browse path and the order path at once. The browse and content pages, being read-heavy, are served through heavy caching and a content delivery network so most reads never reach the core services, and the storefront can scale out its stateless services ahead of the event. The order and inventory path is the harder part, because reservations must stay correct under a surge so popular items are not oversold, which argues for the same fixed-capacity discipline as any flash sale: atomic stock reservation and careful handling of contention on hot SKUs. Nykaa has not published how its serving tier absorbs the spike, so describe the standard levers, caching, a content delivery network, pre-scaling, and contention-safe reservations, and note that the data-platform side does use autoscaling processing.
The unified data platform
Nykaa's most clearly documented engineering work is its data platform, which it described in an AWS case study as One Nykaa, One Data. The problem it solved was fragmentation: data spread across systems made analytics slow and inconsistent, with some reports taking up to two weeks. Nykaa consolidated more than 400 terabytes into a lakehouse on AWS, using S3 as the data lake, EMR with autoscaling for processing, Redshift and Athena for querying, and Lake Formation for governance, organized into ingestion, governance, and query layers. It reported cutting report generation to a few hours, moving data refreshes from daily to hourly, and running about 10,000 ingestion jobs across more than 1,500 customer datasets. The reason this matters for the product is that personalization, the content-and-commerce recommendations, and the merchandising all depend on having one consistent, fresh view of customer and sales data, which is exactly what the platform provides. It is a good example of the analytics backbone that a modern retailer needs behind the storefront.
House of brands and the vertical-integration angle
Beyond retailing other companies' products, Nykaa owns and operates its own brands, in beauty and in fashion. This vertical integration changes the system in subtle ways: owned brands are inventory-led by definition, their data feeds directly into the platform, and they can be featured and personalized within the storefront and content. From a design standpoint, owned brands are the deepest form of the inventory-led model, where Nykaa controls not just the retail but the product, and the same inventory, order, and content systems serve them alongside third-party stock. It is worth raising in an interview as an example of how a retailer's business-model choices, marketplace versus inventory-led versus owned brands, flow directly into what the systems must support, since a single storefront now spans all three.
Trade-offs to discuss
Every senior interviewer expects you to surface at least 3 of these. Pick the decisions, state the alternatives, and justify your choice.
Inventory-led beauty versus a pure marketplace
A pure marketplace is asset-light: sellers hold stock and the platform avoids warehousing and inventory risk. Nykaa chose to be inventory-led for beauty, buying and holding its own stock, because authenticity in a counterfeit-prone category requires controlling the supply chain. The cost is real capital tied up in inventory, warehousing, and the risk of unsold stock, accepted because guaranteeing authentic product is central to trust in beauty. Fashion, where that pressure is lower, stays marketplace.
Omnichannel unified inventory versus separate online and store stock
Running online and store inventory as separate pools is far simpler, but it wastes stock and creates a bad experience, since an item available in a store is invisible online and the site can oversell what a store just sold. A single unified inventory view across warehouses and stores enables accurate availability and store fulfillment, at the cost of near-real-time synchronization between store point-of-sale and the central system. Nykaa's 237-store network makes the unified view worth that complexity.
Content-led discovery versus a search-first storefront
A search-first storefront is simpler and works when customers know what they want. Nykaa leans on content, editorial and tutorials, because beauty shoppers often want guidance, and content-driven discovery deepens engagement and basket size. The cost is building and serving a content system at commerce scale and blending it with product recommendations, accepted because content is a genuine differentiator for the category.
Building a unified data lakehouse versus siloed analytics
Leaving data in per-system silos is less upfront work, but it made analytics slow and inconsistent, with some reports taking weeks. Nykaa consolidated over 400 terabytes into one lakehouse, which cut reporting to hours and enabled fresher, consistent data for personalization. The cost is building and governing the platform and moving many pipelines onto it, which Nykaa judged worthwhile for the analytics and personalization it unlocked.
Near-real-time inventory versus perfectly real-time global stock
Perfectly real-time, globally consistent inventory across every store and warehouse would be ideal but is expensive and hard at scale. Accepting small synchronization windows, backed by atomic reservations at checkout and periodic reconciliation, prevents overselling while keeping the system affordable. The cost is occasional rare edge cases handled by reservation and reconciliation rather than by perfect global consistency, which is the standard omnichannel trade.
How Nykaa actually does it
Nykaa publishes far less about its engineering than pure-technology companies do, so this walkthrough is careful about what is documented versus what is the standard pattern. What is well established, from Nykaa's filings and press, is the business and operating model: Nykaa is inventory-led for beauty and personal care, buying, holding, and curating its own stock to control authenticity, and runs a marketplace model for fashion, and it is omnichannel, with 237 physical stores across 79 cities as of the first quarter of FY26 alongside its website and app. It reported gross merchandise value of about 15,604 crore rupees in FY25, revenue of about 7,950 crore rupees, more than 19 million annual transacting customers on a base of about 42 million, and, through its AWS case study, more than 40 million monthly active users and over 6,000 brands. On a Pink Friday sale day in 2022 it reported around 800,000 orders at more than 400 a minute with about 10 million visits. Its most clearly documented engineering work is its unified data platform, One Nykaa, One Data, which its AWS case study describes as a lakehouse consolidating over 400 terabytes on AWS using S3, EMR with autoscaling, Redshift, Athena, and Lake Formation, cutting reporting from up to two weeks to a few hours and moving refreshes from daily to hourly across about 10,000 ingestion jobs and 1,500-plus datasets. Nykaa also runs machine-learning ranking and personalization, which its hiring confirms. Three accuracy notes for the interview. First, Nykaa has not publicly detailed its transactional serving stack, so its checkout, order, inventory-sync, and search internals are described here as the standard pattern for an inventory-led omnichannel retailer, not as Nykaa-published designs, and specifics like an event-streaming system or a particular cache are industry pattern, not confirmed Nykaa facts. Second, the 800,000-orders sale-day figures are from Pink Friday 2022; recent editions have not published comparable operational numbers. Third, a specific total SKU count and full-year order count are not publicly available.
Sources
- AWS case study, Nykaa builds a unified data platform on AWS: the One Nykaa, One Data lakehouse with S3, EMR, Redshift, Athena, and Lake Formation, 400-plus terabytes, 40M-plus monthly active users
- Nykaa Q4 and FY25 investor presentation: gross merchandise value, revenue, customers, and the beauty and fashion mix
- Nykaa Q1 FY26 investor presentation: the 237-store, 79-city omnichannel network
- Inc42, Nykaa records around 800,000 orders on day one of Pink Friday 2022: the sale-day peak metrics
- Business Standard, Nykaa records a roughly 12x revenue jump on day one of Pink Friday 2022
- Wikipedia, Nykaa: the inventory-led-plus-marketplace model, owned brands, stores, and financials
- Screener, FSN E-Commerce Ventures (Nykaa) financials: independent financial aggregation
Lessons to study before this interview
If any of these topics are fuzzy, the interviewer will catch it. Each lesson is 15 to 60 minutes with diagrams, code, and a quiz.
Cache-Aside Pattern
foundation / caching strategies
Cache Stampede Prevention
foundation / caching strategies
Message Queues
intermediate / messaging event systems
Database Sharding
foundation / database fundamentals
Load Balancing
foundation / core fundamentals
High Availability
advanced / reliability resilience
Rate Limiting for Resilience
advanced / reliability resilience