In any system you get out what you put in.
Online marketplaces are systems that bring together buyers and sellers. The sellers are varied and have different products. Bringing all of this product together on one platform is how a marketplace provides choice for consumers.
One often overlooked yet important driver of marketplace success is what I call the standardisation of the presentation of supply.
Range of choice and tools to navigate that choice
By supply I mean is whatever is being sold on the marketplace. You could call it the product catalogue or the inventory. Whatever you call it, a good marketplace helps buyers to search, filter and compare the choices available to them.
For example, if I search for hotels in Central London on Booking.com I get 1,064 hotels returned. I can filter my choices down to ones that might suit me better.
- Say I only want 3 star hotels with Wi-fi and air conditioning, I find that there are 64 hotels.
- Say I also want to ensure that there are family rooms and where the reviews are better than 8/10, I get 25 properties.
- 25 properties is a more manageable amount of hotels to browse than 1,064.
Being able to search and filter is what customers need to make sense of so much choice.
Good UI is a growth driver
The UI (User interface) design determines how easy it is for buyers to find what they are looking for.
In a real world store such as a supermarket, goods are categorised arranged into sections (dairy, bakery, fruit and vegetables etc). Likewise, online marketplaces need to organise their goods.
Search and filter is the cornerstone user interface design for online marketplaces. (By the way, there are other ways to find goods, it's not always search and filter. e.g. Stylect helps women discover shoes in their mobile app through a Tinder-esque swipe left vs right to like / dislike suggestions).
And - when you've found a couple of things that interest you, you might want to compare choices. If product listings have some consistency in their presentation, you can compare options more easily. It's not just about how you find products, it's about how you present them. UK online estate agent Rightmove does this well. Each listing has a consistent layout giving some predictability to the user experience.
If search, filter and compare are done well, the conversion rate from visitor to buyer is improved. The better the conversion rate, the more the marketplace can afford to pay to acquire customers (i.e. a bigger marketing budget). A bigger marketing budget drives growth and growth drives market share.
As a marketplace you want to be the number one marketplace in your sector. (After all, buyers search eBay because there is so much choice, sellers list on eBay because there are so many buyers).
One of the foundations of growth therefore is making it super easy for customers to find what they want.
Good UI needs standardisation of data-sets
Search, filter and compare requires that software can compare like for like data to find matches and mis-matches. If booking.com want to offer a filter choice such as Wi-fi, every hotel needs a Y/N flag in the database on whether it has Wi-fi.
This means that every product listing needs to store as much information as possible in a standard data format. It's difficult to write code to compare two hotel descriptions if they are just text. If some of the attributes of the hotel are stored as data points which the code can compare, then you can use these comparisons in search and filter.
A key component therefore of marketplace success is the ability to create a standard data structure to normalise what would otherwise be a disparate set of product listings.
Good data sets need good data frameworks and collection processes
Standardisation is important for most marketplaces. For marketplaces where sellers can list their own products directly (e.g. eBay), it's particularly important to build input forms that collect the data in a structured way. If you use the eBay iPhone app to list an item for sale you'll notice just how much standardisation has been designed into the data structure - right down to the postage choices.
The information architecture of a product listing is therefore a critical part of the product/service design and is often overlooked. Getting this right early can help a marketplace give a better consumer experience.
In building a marketplace, a key priority in the product design process is mapping out the data structures for each listing type. The question to ask is, "does this information have to be in words?"
Other data formats that are more comparable include;
- Yes / No
- choices from a pre-defined list
This type of data is structured data. Build listings on structured data and you can build great product discovery tools and you can build consistent high quality presentation of listings.
Curated versus non-curated
A curated marketplace is where the marketplace chooses which merchants / suppliers to feature. A non-curated marketplace is where anyone can list. Booking.com is an example of a curated marketplace. eBay is a non-curated marketplace.
The non-curated marketplace faces the greatest challenge in maintaining standardisation of the presentation of supply. It is normally done by allowing suppliers relatively little freedom in how listings are added. A specific format needs to be adhered to. Product Managers in such marketplaces spend a good deal of attention on validating the inputs from suppliers into their systems. Form design, validation scripts and API and CSV import structures if all done well will have a strong positive influence on the quality and consistency of inventory data.
Curated marketplaces can afford to have a fewer systems to control the data input process. That is, provided that they have human beings to do the quality checking or content loading. Sometimes the same rigour is needed with the input forms but these might be internal systems rather than externally facing systems.
If data quality is important, how do we improve it?
An important operational activity for a marketplace is quality control. A person or team are tasked with analysing the integrity of the data and making plans to improve it.
I've found that a good starting place to analyse the quality of the listings is to produce exception reports. Exception reports tell you which of your listings/products doesn't match the criteria/standards you are looking for. If you require at least 3 photos per product you would need a simple report that shows you any product that has less than 3 photos. It doesn't need to be a fancy web reporting system, it could simply be a SQL query against your database and a text file report.
By finding exceptions and then systematically correcting them you can improve the overall data quality. Then it's important to feedback to the product team so that they can improve the data entry points to prevent the exceptions occuring again.
Good data only gets you so far
It's not all about structured data though - there need to be a "wow" in the product that brings people back, this could include for example...
- Great inspirational design
- Curated showrooms that help visitors discover exciting goods
- Amazing price promotions
- Superb photography
- Expert reviews
- User generated reviews
- Simple payment and delivery options
However, my point remains this: Structured data-sets are essential to get right when building great marketplaces.