Mirage JS Deep Dive: Understanding Factories, Fixtures And Serializers (Part 2)

Mirage JS Deep Dive: Understanding Factories, Fixtures And Serializers (Part 2)

Mirage JS Deep Dive: Understanding Factories, Fixtures And Serializers (Part 2)

Kelvin Omereshone

2020-05-29T11:00:00+00:00 2020-05-29T21:08:33+00:00

In the previous article of this series, we understudied Models and Associations as they relate to Mirage. I explained that Models allow us to create dynamic mock data that Mirage would serve to our application when it makes a request to our mock endpoints. In this article, we will look at three other Mirage features that allow for even more rapid API mocking. Let’s dive right in!

Note: I highly recommend reading my first two articles if you haven’t to get a really solid handle on what would be discussed here. You could however still follow along and reference the previous articles when necessary.


In a previous article, I explained how Mirage JS is used to mock backend API, now let’s assume we are mocking a product resource in Mirage. To achieve this, we would create a route handler which will be responsible for intercepting requests to a particular endpoint, and in this case, the endpoint is api/products. The route handler we create will return all products. Below is the code to achieve this in Mirage:

import { Server, Model } from 'miragejs'; new Server({ models: { product: Model, }, routes() { this.namespace = "api"; this.get('products', (schema, request) => { return schema.products.all() }) }
}); },

The output of the above would be:

{ "products": []

We see from the output above that the product resource is empty. This is however expected as we haven’t created any records yet.

Pro Tip: Mirage provides shorthand needed for conventional API endpoints. So the route handler above could also be as short as: this.get('/products').

Let’s create records of the product model to be stored in Mirage database using the seeds method on our Server instance:

 seeds(server) { server.create('product', { name: 'Gemini Jacket' }) server.create('product', { name: 'Hansel Jeans' }) },

The output:

{ "products": [ { "name": "Gemini Jacket", "id": "1" }, { "name": "Hansel Jeans", "id": "2" } ]

As you can see above, when our frontend application makes a request to /api/products, it will get back a collection of products as defined in the seeds method.

Using the seeds method to seed Mirage’s database is a step from having to manually create each entry as an object. However, it wouldn’t be practical to create 1000(or a million) new product records using the above pattern. Hence the need for factories.

Factories Explained

Factories are a faster way to create new database records. They allow us to quickly create multiple records of a particular model with variations to be stored in the Mirage JS database.

Factories are also objects that make it easy to generate realistic-looking data without having to seed those data individually. Factories are more of recipes or blueprints for creating records off models.

Creating A Factory

Let’s examine a Factory by creating one. The factory we would create will be used as a blueprint for creating new products in our Mirage JS database.

import { Factory } from 'miragejs' new Server({ // including the model definition for a better understanding of what’s going on models: { product: Model }, factories: { product: Factory.extend({}) }

From the above, you’d see we added a factories property to our Server instance and define another property inside it that by convention is of the same name as the model we want to create a factory for, in this case, that model is the product model. The above snippet depicts the pattern you would follow when creating factories in Mirage JS.

Although we have a factory for the product model, we really haven’t added properties to it. The properties of a factory can be simple types like strings, booleans or numbers, or functions that return dynamic data as we would see in the full implementation of our new product factory below:

import { Server, Model, Factory } from 'miragejs' new Server({ models: { product: Model }, factories: { product: Factory.extend({ name(i) { // i is the index of the record which will be auto incremented by Mirage JS return `Awesome Product ${i}`; // Awesome Product 1, Awesome Product 2, etc. }, price() { let minPrice = 20; let maxPrice = 2000; let randomPrice = Math.floor(Math.random() * (maxPrice - minPrice + 1)) + minPrice; return `$ ${randomPrice}`; }, category() { let categories = [ 'Electronics', 'Computing', 'Fashion', 'Gaming', 'Baby Products', ]; let randomCategoryIndex = Math.floor( Math.random() * categories.length ); let randomCategory = categories[randomCategoryIndex]; return randomCategory; }, rating() { let minRating = 0 let maxRating = 5 return Math.floor(Math.random() * (maxRating - minRating + 1)) + minRating; }, }), },

In the above code snippet, we are specifying some javascript logic via Math.random to create dynamic data each time the factory is used to create a new product record. This shows the strength and flexibility of Factories.

Let’s create a product utilizing the factory we defined above. To do that, we call server.create and pass in the model name (product) as a string. Mirage will then create a new record of a product using the product factory we defined. The code you need in order to do that is the following:

new Server({ seeds(server) { server.create("product") }

Pro Tip: You can run console.log(server.db.dump()) to see the records in Mirage’s database.

A new record similar to the one below was created and stored in the Mirage database.

{ "products": [ { "rating": 3, "category": "Computing", "price": "$739", "name": "Awesome Product 0", "id": "1" } ]

Overriding factories

We can override some or more of the values provided by a factory by explicitly passing them in like so:

server.create("product", {name: "Yet Another Product", rating: 5, category: "Fashion" })

The resulting record would be similar to:

{ "products": [ { "rating": 5, "category": "Fashion", "price": "$782", "name": "Yet Another Product", "id": "1" } ]


With a factory in place, we can use another method on the server object called createList. This method allows for the creation of multiple records of a particular model by passing in the model name and the number of records you want to be created. Below is it’s usage:

server.createList("product", 10)


server.createList("product", 1000)

As you’ll observe, the createList method above takes two arguments: the model name as a string and a non-zero positive integer representing the number of records to create. So from the above, we just created 500 records of products! This pattern is useful for UI testing as you’ll see in a future article of this series.


In software testing, a test fixture or fixture is a state of a set or collection of objects that serve as a baseline for running tests. The main purpose of a fixture is to ensure that the test environment is well known in order to make results repeatable.

Mirage allows you to create fixtures and use them to seed your database with initial data.

Note: It is recommended you use factories 9 out of 10 times though as they make your mocks more maintainable.

Creating A Fixture

Let’s create a simple fixture to load data onto our database:

 fixtures: { products: [ { id: 1, name: 'T-shirts' }, { id: 2, name: 'Work Jeans' }, ], },

The above data is automatically loaded into the database as Mirage’s initial data. However, if you have a seeds function defined, Mirage would ignore your fixture with the assumptions that you meant for it to be overridden and instead use factories to seed your data.

Fixtures In Conjunction With Factories

Mirage makes provision for you to use Fixtures alongside Factories. You can achieve this by calling server.loadFixtures(). For example:

 fixtures: { products: [ { id: 1, name: "iPhone 7" }, { id: 2, name: "Smart TV" }, { id: 3, name: "Pressing Iron" }, ], }, seeds(server) { // Permits both fixtures and factories to live side by side server.loadFixtures() server.create("product") },

Fixture files

Ideally, you would want to create your fixtures in a separate file from server.js and import it. For example you can create a directory called fixtures and in it create products.js. In products.js add:

// <PROJECT-ROOT>/fixtures/products.js
export default [ { id: 1, name: 'iPhone 7' }, { id: 2, name: 'Smart TV' }, { id: 3, name: 'Pressing Iron' },

Then in server.js import and use the products fixture like so:

import products from './fixtures/products'; fixtures: { products, },

I am using ES6 property shorthand in order to assign the products array imported to the products property of the fixtures object.

It is worthy of mention that fixtures would be ignored by Mirage JS during tests except you explicitly tell it not to by using server.loadFixtures()

Factories vs. Fixtures

In my opinion, you should abstain from using fixtures except you have a particular use case where they are more suitable than factories. Fixtures tend to be more verbose while factories are quicker and involve fewer keystrokes.


It’s important to return a JSON payload that is expected to the frontend hence serializers.

A serializer is an object that is responsible for transforming a **Model** or **Collection** that’s returned from your route handlers into a JSON payload that’s formatted the way your frontend app expects.

Mirage Docs

Let’s take this route handler for example:

this.get('products/:id', (schema, request) => { return schema.products.find(request.params.id); });

A Serializer is responsible for transforming the response to something like this:

{ "product": { "rating": 0, "category": "Baby Products", "price": "$654", "name": "Awesome Product 1", "id": "2" }

Mirage JS Built-in Serializers

To work with Mirage JS serializers, you’d have to choose which built-in serializer to start with. This decision would be influenced by the type of JSON your backend would eventually send to your front-end application. Mirage comes included with the following serializers:

  • JSONAPISerializer
    This serializer follows the JSON:API spec.
  • ActiveModelSerializer
    This serializer is intended to mimic APIs that resemble Rails APIs built with the active_model_serializer gem.
  • RestSerializer
    The RestSerializer is Mirage JS “catch all” serializer for other common APIs.

Serializer Definition

To define a serialize, import the appropriate serializer e.g RestSerializer from miragejs like so:

import { Server, RestSerializer } from "miragejs"

Then in the Server instance:

new Server({ serializers: { application: RestSerializer, },

The RestSerializer is used by Mirage JS by default. So it’s redundant to explicitly set it. The above snippet is for exemplary purposes.

Let’s see the output of both JSONAPISerializer and ActiveModelSerializer on the same route handler as we defined above


import { Server, JSONAPISerializer } from "miragejs"
new Server({ serializers: { application: JSONAPISerializer, },

The output:

{ "data": { "type": "products", "id": "2", "attributes": { "rating": 3, "category": "Electronics", "price": "$1711", "name": "Awesome Product 1" } }


To see the ActiveModelSerializer at work, I would modify the declaration of category in the products factory to:

productCategory() { let categories = [ 'Electronics', 'Computing', 'Fashion', 'Gaming', 'Baby Products', ]; let randomCategoryIndex = Math.floor( Math.random() * categories.length ); let randomCategory = categories[randomCategoryIndex]; return randomCategory; },

All I did was to change the name of the property to productCategory to show how the serializer would handle it.

Then, we define the ActiveModelSerializer serializer like so:

import { Server, ActiveModelSerializer } from "miragejs"
new Server({ serializers: { application: ActiveModelSerializer, },

The serializer transforms the JSON returned as:

{ "rating": 2, "product_category": "Computing", "price": "$64", "name": "Awesome Product 4", "id": "5"

You’ll notice that productCategory has been transformed to product_category which conforms to the active_model_serializer gem of the Ruby ecosystem.

Customizing Serializers

Mirage provides the ability to customize a serializer. Let’s say your application requires your attribute names to be camelcased, you can override RestSerializer to achieve that. We would be utilizing the lodash utility library:

import { RestSerializer } from 'miragejs';
import { camelCase, upperFirst } from 'lodash';
serializers: { application: RestSerializer.extend({ keyForAttribute(attr) { return upperFirst(camelCase(attr)); }, }), },

This should produce JSON of the form:

 { "Rating": 5, "ProductCategory": "Fashion", "Price": "$1386", "Name": "Awesome Product 4", "Id": "5" }

Wrapping Up

You made it! Hopefully, you’ve got a deeper understanding of Mirage via this article and you’ve also seen how utilizing factories, fixtures, and serializers would enable you to create more production-like API mocks with Mirage. Look out for the next part of this series.

Smashing Editorial (ra, il)

How To Run The Right Kind Of Research Study With The Double-Diamond Model

How To Run The Right Kind Of Research Study With The Double-Diamond Model

How To Run The Right Kind Of Research Study With The Double-Diamond Model

Steve Bromley

2020-05-29T10:00:00+00:00 2020-05-29T21:08:33+00:00

Product and design teams make a lot of decisions. Early on in the development of a product, they will be thinking about features — such as what the product should do, and how each feature should work. Later on, those decisions become more nuanced — such as ‘what should this button say? Each decision introduces an element of risk — if a bad decision is made, it will reduce the chance for the product to be successful.

The people making these decisions rely on a variety of information sources to improve the quality of their decision This includes intuition, an understanding of the market, as well as an understanding of user behavior. Of these, the most valuable source of information to put evidence behind decisions is understanding our users.

Being armed with an understanding of the appropriate user research methods can be very valuable when developing new products. This article will cover some appropriate methods and advice on when to deploy them.

A Process For Developing Successful Products

The double diamond is a model created by the UK’s Design Council which describes a process for making successful products. It describes taking time to understand a domain, picking the right problem to solve, and then exploring potential ideas in that space. This should prove that the product is solving real problems for users and that the implementation of the product works for users.

Diagram showing the design council’s Double Diamond model
The design council’s Double Diamond model (Large preview)

To succeed at each stage of the process requires understanding some information about your users. Some of the information we might want to understand from users when going through the process is as follows:

The double diamond image with user research questions linked to each phase
Some research questions appropriate for each stage (Large preview)

Each stage has some user research methods that are best suited to uncovering that information. In this article, we’ll refer to the double diamond to highlight the appropriate research method throughout product development.

Diamond 1: Exploring The Problem And Deciding What To Fix

The first diamond describes how to come up with a suitable problem that a new product or feature should fix. It requires understanding what problems users have, and prioritizing them to focus on a high-value area. This avoids the risk of building something that no-one is going to use.

The most effective way of understanding the problem is to get true first-hand experience of users performing real tasks in context. This is best done by applying ethnographic and observational methods to identify the range of problems that exist, then prioritizing them using methods such as surveys.

Double Diamond PhaseAppropriate MethodWhy?
Explore the problemEthnographic and Observational studiesGives deep insight into what problems people have that can inspire product decisions
Decide what to fixSurveysDiscovers how representative problems are, and helps prioritise them

We’ll review each method, in turn, to describe why it’s appropriate.

Explore The Problem With Ethnographic And Observational Methods

The first phase of the double diamond is to ‘explore the problem’. User research can build up an understanding of how people act in the real world and the problems they face. This allows the problem space to be fully explored.

The double diamond image with ‘Explore the problem’ highlighted
Explore the problem (Large preview)

This valuable behavioral information is only uncovered only by watching people do real tasks and asking them questions to uncover their motivations and issues. Doing early qualitative research will help identify the problems that people have. These problems can inspire product ideas, features, and help teams understand how to successfully solve user’s problems. This information will also help disregard poor product ideas by revealing that there is no real need for it. This leads to a more useful product being developed and increasing the chance of success.

The most appropriate methods for doing this are ethnographic. This can include diary studies, where a user’s interaction with the subject matter is captured over a number of weeks. This reveals issues that wouldn’t turn up in a single session or that people wouldn’t remember to talk about in a lab-based interview.

This isn’t the only way of uncovering this kind of in-depth information though. Other suitable observational methods include watching people use existing software or products, both in the lab or in the wild. These are quicker and easier to run than diary studies, but are limited to only capturing a single interaction or what the participant will remember when prompted. For some subject matters, this might be enough (e.g. navigating an online shop can be done and explored in a single session). More complex interactions over time, such as behavior with fitness trackers, would be more sensible to track as a diary study.

Decide What To Fix With Surveys

The second half of the first diamond comes next. Having understood real user’s contexts and what problems they have, these can then be documented and prioritized to ‘decide what to fix’.

The double diamond image with ‘Decide what to fix’ highlighted
Decide what to fix (Large preview)

This prioritization will be done by product managers who take into account many factors, such as “what do we have the technical ability to do” and “what meets our business goals”. However, user research can also add valuable information by uncovering the size of the issues users have. Surveys are a sensible approach for this, informed by the true understanding of user behavior uncovered in the earlier studies. This allows teams to size the uncovered issues and reveal how representative the behaviors discovered are.

Combining quantitative methods with generative user research studies help inspire early decisions about what a product should do. For example, Spotify’s discovery work on how people consume music analyzed primary research fieldwork to create personas and inform their development work. This allows a team to complete the first diamond with a clear understanding of what problem their product is trying to solve.

Diamond 2: Test And Refine Potential Solutions

The second diamond describes how to end up on a successful implementation of a product to fix the problem. Having decided which problem to fix, research can then explore different ways of fixing that problem, and help refine the best method.

Double Diamond PhaseAppropriate MethodWhy?
Test potential solutionsModerated usability testingCreates a deep understanding of why the solution works, to inform iteration
Refine final solutionUnmoderated usability testingCan get quick results on small questions, such as with the UI

Test Potential Solutions With Moderated Usability Testing

The second diamond in the double diamond design process starts with evaluating a variety of solutions in order to decide the best possible implementation of a product. To achieve this with rigor requires usability testing — creating representative prototypes and then observing if users can successfully complete tasks using them.

The double diamond image with ‘Test Potential Solutions’ highlighted
Test potential solutions (Large preview)

This kind of study takes time to do properly, and attention on each individual’s user experience to understand what causes the behavior that is observed during usability testing. A moderated session, with the researcher present, can ask probing questions to uncover things that participants won’t articulate unprompted such as “what are you thinking currently” or “ why did you decide to do that?”. These kinds of studies reveal more data when a moderator is able to ask participants these questions, and avoids missing the opportunity to gather more data from each study, which can be used to evaluate and iterate the product. A single moderated research session potentially reveals more useful information than a series of unmoderated tests.

This kind of in-depth exploration of the problem has been a key part of AirBNB’s early success. In 2009 the company was close to bankruptcy and desperate to understand why people were not booking rooms. By spending time with users reviewing the ads on their website, they were able to uncover that the pictures were the problem. This then allowed them to focus their iteration on the process for gathering photos of rooms, which put them on the path for changing hotel booking forever. As the global pandemic changes people’s behavior with holidays in the future, in-depth qualitative research will be essential as they continue to adapt to new challenges.

This doesn’t mean that the moderator has to be in the same room as the participant. Often it can be very valuable to find participants who are geographically remote, and avoid over-sampling people who live in major cities, which is often where research teams are based. Screen sharing software, such as google hangouts or zoom can make remote sessions possible, while still having the session run live with a moderator.

Refine Final Solution With Unmoderated Usability Testing

The final stage of the double diamond describes refining the final solution, which can require a lot of small iterative tests. A shortcut to the deep insight from moderated testing is remote unmoderated research. This includes tools like usertesting.com which allow teams to put their software in front of users with little effort. By sending a website URL to their panel of users, they send back videos of their participants using the website and commenting on their experience.

The double diamond image with ‘Refine final solution’ highlighted
Refine final solution (Large preview)

This method can be popular because it is quick (multiple sessions can run simultaneously without a moderator present) and cheap (participants aren’t paid a huge amount to take part). Because of this, it is often considered an appropriate tool by companies looking to start doing user research, however, care needs to be taken.

This method has constraints which means that it’s only sensible for later on in the design process. Because the participants on these kinds of websites are all people who test multiple websites regularly, they become increasingly different to a normal user. This changes their behavior while using websites and makes it dangerous to draw conclusions from their behavior about what other users would understand or do. This is called a sampling bias — creating a difference between ‘real’ users, and the users being tested.

Because of these risks, these studies may be most appropriate late in development, testing content or UI changes, when the risks of getting decisions wrong are much lower. Iterative studies ensure that users understand what has been created, and are able to use it in the way the designer intended. An example of this is the iterative usability testing the UK’s Government Digital Service ran to ensure citizens could successfully identify themselves and access government services.

After The Double Diamond

As we’ve covered, it is important to select the right method to de-risk product decisions before launch. When a product is launched, it will be immediately obvious whether there is an audience for it, and whether people understand and can use the product — both through how well the product sells, and through reviews and customer feedback.

The double diamond image with ‘Solution delivered’ highlighted
After the double diamond (Large preview)

Nevertheless, launching the right product doesn’t mean that the opportunity for research is over. New opportunities to explore real user behavior will continue to inspire adding or removing features, or changes to how the product works.

Double Diamond PhaseAppropriate MethodWhy?
Solution deliveredAnalytics + moderated usability testing combinedInform future updates post-launch with qualitative and quantitative insight.

Combining some of the methods we’ve described previously with new data from analytics will continue to drive high-quality decision making.

Research After The Solution Is Delivered With Analytics

Post-launch analytics are an important part of building a complete understanding of the behavior of users.

Analytics will reveal what people are doing on a website. However, this information is most valuable when combined with understanding why that behavior is occurring. It is also important to be aware that analytics are only seeing a short section of a user’s experience, the part that happens on your website and their whole end-to-end journey also includes a lot of things that happened off the site, or in the real world. Building a research strategy that combines insight from analytics with an understanding of motivations from qualitative studies is a powerful combination to inform decision making.

This requires close collaboration between the analytics team and the user research team — regular community events, skills sharing and project updates will create awareness of the priorities of each team, the type of research questions they can support one another with and identify opportunities to work together, leading to a stronger combined team.

Optimize Your Research Process

In this article, we’ve covered some appropriate methods to use to inform product development. However, there can still be resistance to running the right kind of study.

New research teams may be asked to cut corners. This can include suggesting participants who are convenient, such as friends, without taking the time to screen them to ensure they represent real users. This can be suggested by colleagues who are unaware of the risks caused by taking decisions based on unrepresentative research.

In addition to running research studies, a researcher has to be an educator and advocate for running the right kind of studies and help their colleagues understand the differences in quality between the type of information gathered from different research methods. Presentations, roadshows, and creating posters are some techniques that can help achieve this.

Incorporating user research into decision making can be quite radical at some organizations, particularly those with a history of deferring to client wishes or listening to the highest-paid person in the room. A lot of hard work and creativity are needed to bring about change in how people work. This requires understanding the decision maker’s current incentives, and describing the benefits of research in a way that shows how it makes their life easier.

If an organization understands and accepts why running studies using appropriate methods it shows a real desire for improving the quality of decision making within the organization. This is an encouraging sign that a new research team has the potential to be successful.

The next step for new researchers will be to establish the logistics of running research, including creating a research process, building out the tools and software needed, and identifying the highest priority research questions for your organization. There is a lot of great guidance from the research community on techniques to do this, for example, the work being done by the research ops community.

Smashing Editorial (ah, ra, il)

Japan Is a Enticing Market for Cross-border Ecommerce

Japan is the world’s third-largest ecommerce market following China and the United States. However, it is often overlooked by foreign online merchants because of an unusually low cross-border purchase rate and a somewhat insular approach to online shopping.

Only 10 percent of Japanese consumers buy goods on foreign websites. Nevertheless, with a population of 126.5 million, a 93-percent internet penetration rate, and 75 percent of those making purchases online, Japan is an enticing market for cross-border online sellers.

Research firm Statista anticipates that Japanese ecommerce sales in 2020 will amount to approximately $100 billion, a diminished estimate due to global economic changes. An unpopular sales tax increase in 2019 is also dampening consumer spending.

After a contraction in the last quarter of 2019, the Japanese economy continued to shrink in the first quarter of 2020. Japanese consumers, once considered extremely brand-loyal, are now more cost-conscious and showing an increasing willingness to try new brands in their quest for bargains.

Japanese Online Consumers

  • Most Japanese do not read English or other foreign languages and buy only from Japanese language sites.
  • Japanese consumers prefer detailed information about the product they are considering.
  • Forty-four percent use price comparison sites.
  • Japanese consumers are avid mobile users. More than half of online purchases are on a mobile device.
  • While credit cards are the payment method for most online sales, the Konbini cash convenience-store-payment system is also popular, with 50,000 locations across the country. Locally-issued credit cards are much more popular than Western-brand cards.

Japan is home to a rapidly aging population, but seniors use the internet at a level similar to younger persons. Japan is losing people because of a low fertility rate, so targeting the older population — which comprises 28 percent of the total population — makes sense.

Revenue opportunities exist for those merchants who can address the needs of older consumers. For instance, prepared meals are a $20 million and rapidly growing market with growth attributable mostly to seniors, according to Statista. The Covid-19 pandemic has contributed to a spurt in online sales by this vulnerable segment.


Three marketplace platforms — Rakuten, Amazon, and Yahoo Shopping — account for over one-third of online transactions in Japan. Rakuten and Yahoo Shopping require non-American foreign companies that wish to create accounts to start a local company or work with a local intermediary on their behalf.

Rakuten. Japan’s largest marketplace in revenue has more than 100 million members and is strongest in fashion and food. American brands that sell on Rakuten’s platform have their own storefront and product listings. They can also advertise via sponsored product listings and leverage other advertising features on Rakuten, such as coupons, newsletter ads, and banner ads. Rakuten has its own fulfillment service called Rakuten Super Logistics.

Amazon Japan ranks first in site visits and sales. It provides a relatively easy way for foreign merchants with an Amazon seller’s account to market products directly to Japanese consumers. Amazon requires that all sellers provide customer support in Japanese. Amazon handles fulfillment from Japan, but companies must have a Japanese company as an importer of record to deal with customs.

Yahoo Japan Shopping is the ecommerce arm of the diversified web portal of Yahoo Japan, which is owned by Softbank and Alibaba and is separate from its American parent. Yahoo Japan Shopping provides a wide range of products, including cosmetics, electronics, apparel and footwear, home equipment, and furniture.

Zozotown is an offshoot of a mail-order company, Start Today. It is the largest specialized fashion and footwear ecommerce platform and offers products from many local and international brands. Yahoo Japan acquired a majority stake in Zozotown in 2019.

Qoo10. This online marketplace has a broad portfolio of products in fashion, beauty, sports, digital, mobile, home living, life, food, baby, entertainment, and books, Operated as a joint venture with eBay, Qoo10 also sells in other Asian markets and is especially strong in Korea and Singapore.

Wowma was founded by KDDI, one of Japan’s largest mobile phone providers. The site offers more than 20 million products, including cosmetics, decorations, clothing, electronics, and personal care services.

C2C marketplaces. Japan has a thriving consumer-to-consumer market that has significantly expanded in the last few years. It consists of second-hand marketplaces and online auction sites. The two major players are Mercari, which operates via a downloadable application, and Yahoo Auctions. Rakuma is a C2C flea market application owned by Rakuten.