Selling on the web has gotten increasingly complex with more channels and partners than ever before. For one thing, you have to make the tricky choice between using cloud-based e-commerce tools (easy, cheap, and fun, but prone to outages) or your own data center servers. You have to enable customers to complete transactions on not only their desktop computers, but also on tablets, mobile devices, TVs, digital out-of-home media, and more. In addition, you need to add a layer of targeting, be it business to consumer, business to business, or consumer to consumer. The list of e-commerce concerns goes on and on.
Let's look at e-commerce from a higher level and investigate how to vary the user experience depending on what is being sold. Too often, companies look at e-commerce as a single, monolithic concept -- either you're hawking products or you're not -- but this kind of thinking can result in an experience that's incomplete, complicated, and frustrating.
While there are many combinations, e-commerce sites generally break down into three basic categories: large online catalogs, commodity products and services, and freemium/paywall content sites. I’ll start with the first category, since that’s what marketers see the most of.
Large online catalogs
The online catalog is the poster child of e-commerce, as it's what most of us think of when we imagine a commerce site. The most obvious example is the "massively multi-SKU," Amazon.com. The hallmarks of such sites are their vast numbers of products.
The toughest job on these sites is figuring out what the buyers are coming for and how to get them to where they need to go. The big issue for the user-interface on these properties is cognitive load (how many items the users can hold in their short-term memory at any given time). People become overwhelmed when viewing crowded pages that contain no hierarchy and too many options. So, when you have a site containing a variety of products like Target.com -- which sells everything from potato chips to flat-screen TVs -- you'll need to implement user-friendly tools to get your shopper to the right aisle, so to speak. These tools include:
Let’s be honest: -Search is the crutch of a bad interface. When you can't easily find what you want on a site, you start using the "search" field. So yes, it must be there, and it must work well, but let's call it table stakes.
While it’s the most important feature on an e-commerce site, global navigation often gets short shrift. Or, on the flip side, retailers may go too far in the other direction, creating a bloated global nav that confuses more than it helps. We recently conducted user-testing for a large retailer to choose between "global nav" draft options. The main categories were set, but, as with any catalog site, there were far more categories than could possibly be displayed in the navigation bar. This test was to decide what should go in the nav's "see more" area. It was a great lesson on reported behavior versus observed behavior. Every user in the test said that they wanted the option with more items in it. But, when it actually came to finding an item, each person was much quicker with the list that had fewer items in it -- cognitive load in action.
Parametric browse and filter
How can you make it easy to shave down a huge number of results into a manageable few? Use parameters to filter the results. Zappos.com is the master of this: "men's shoes" yields 6,484 results while "men's boots casual size 10 moccasin" yields 12. Now, that's a number one can work with.
Upsell and cross-sell
Part of the difficulty with large catalog sites is unearthing the many thousands of products. Create ways for intelligent pairings to surface after a user has looked at the details of one product. Point them to additional products that make sense -- like batteries for a flashlight or a spatula for a frying pan. Or give your site the ability to cross-sell. For example, if users bought a flashlight and a frying pan, they might also be in the market for a tent or a sleeping bag.
A little information can be a wonderful -- or it can be a dangerous thing. While "product recommendations" have in some cases gotten better -- thanks, Netflix -- a recommendations engine can be tricky. In essence, the site is predicting what you might want to buy next. The rule of this one is simple: Don't do it if you can't do it well. There's nothing more annoying than buying your grandmother a present and then -- for the rest of your life -- having the site assume you're deeply interested in cardigans and orthopedic shoes.