If you want someone to give you their money, you better treat them well… So why is so many online businesses speaking to their customers like they were doing something stupid?

It’s the same thing you see over and over again – and even though I love and use a variety of open source software, I have to say that larger open source E-commerce initiatives does a terrible job at handling user input gracefully.

I’ve been there myself, as a developer, I tend to write unfriendly error messages from time to time. I realized years ago that it was time to sharpen up my skills and focus more on the user. After all, that’s what matters the most.

Some time ago, I read an excellent blog post over at Get Elastic called “Losing Customers At The Register: 12 Checkout Blunders” about how e-tailers lose revenue on their site. First point on the list you’ll see unfriendly credit card errors!

I was working with one of our pilot customers, The Wall Company, and did some research looking into their data to see that credit card errors occurred a few times a week.

We changed their error messages from the nasty “Something went wrong, please try again”, to this:

That alone, helped 49 of 126 credit card error turn into success. Data from the past 3 months shows us that 10% of all completed orders have at least one credit card error!

Wouldn’t it be great if you could just add a new product to your store front and get measurable feedback in a matter of days? You always need to get feedback on how your product details are helping you sell the product. How can you know that you’ve set the price right? What about the images? Headline? Page title? How does the product page perform in the search engines?

All these questions are very hard to answer by digging up data from your visitor and sales statistics. Using Google Website Optimizer can give you a good idea – but it’s not the easiest thing to install if you haven’t got the technical skills it requires.

I’ve been thinking about a metric that gives you an idea about overall performance of your E-commerce store. This metric (Site Experience Index) is part of what I call The E-commerce Framework, that will be part of an upcoming Milkshake Commerce release. Let me explain what The E-commerce Framework can do.

Ordinary web analytics data does a terrible job at giving you an overview of how visitors use your site. You get loads of details, but at the end of the day it’s about trends and overall performance. If you could get a site level metric that tells you how much value your visitors brought to you today, divided into areas such as category pages, product pages, shopping cart, about pages, FAQs, contact forms, checkout process – you’d easily get an overview of site performance.

So for each area of your site, you get a score. That score will change from time to time. If you just deployed a complete redesign of your site, you could see if your score decreased, or increased. The score is a relative score, meaning that it is abstracted from site traffic which will make you able to compare your own site to other sites – every day of the week with each one another, and you can compare the score after deploying a new design, adding new functionality and such.

This is more than a conversion rate metric. It’s a metric that tells you the overall experience of your site, with the ability to drill down into certain areas of your site.

Do you have the idea, that your users have trouble finding what they’re looking for on your site?

Do you find it hard using your own site’s search engine? Or, are the delivered results just not good enough?

In this blog post, I want to take a look at the state of internal site search, as it is today on average E-commerce sites. Do we leave our users and customers more frustrated than they were before?

At last, I’ll give you a checklist of what you need to get right, in order to achieve the internal site search experience you’ve always wanted your users to have.

The state of Internal Site Search

Let me first point out, that I’m not taking the likes of Amazon into consideration. If you look at Amazon’s site search, you’ll know one thing: The more money you have, the more money you can spend on a better site search. Amazon is not the average E-commerce site, which is why I won’t take a further look at their site.

So let me start out by looking at a shop that runs on Magento E-commerce Software - a somewhat popular E-commerce platform. Through their website, I found one of their customers, Homedics. They sell Massage, relaxation and wellness products. A classic mistake with site search is that you get a “no results” page when searching for category names.

Let’s say that I’m looking for Foot Spas, which is exactly what they sell. Like many users, I don’t spent time on figuring out their navigational structure and goes straight to the site search which is located at the top right hand corner.

It’s a nice and visibly clear search box which looks very promising, so I type foot spas and hits the ‘enter’ key.

Much to my disappointment, I get a useless “no results” page simply telling me the frustrating message: “Your search returns no results”.

What’s wrong here?

1. Their product categories are not searchable.

2. Their no results page is totally useless. They could at least give me suggestions to help me get on with my life when I’m stuck like now.

3. Much to my surprise, they have a nice auto complete feature that suggests a search for “foot spa” (not spas). Now this phrase gives me two products as a result. They should’ve included this phrase in my original phrase without even telling me.

Sadly, this is not all.

When I finally got some results, I noticed that I couldn’t filter, sort or search within the current results.

Filtering gives me the ability to trim my results down to as few products as possible, that I can make a decision about later on. Again, much to my surprise they offer filtering on their product category pages. Although it is a very shy filtering feature, it’s useful if you want to narrow down your search results by price.

Sorting is absolutely vital. Sorting is a very useful personalization metric for you as a show owner to use. This gives you a great opportunity to get to know your users. Sorting by price tells you that a user is looking for something cheap. So don’t recommend the most expensive items! Sorting by popularity tells you that the user needs some sort of 3^rd party recommendation of the product, and price is maybe not that important.

Searching within results is needed when you’re trying to find a specific product that you don’t know the exact name of. Searching within lets you forget about price, brand, and popularity and so on.

Sadly enough, we can find even more things that are wrong about this one.

Users come from different countries and regions with different perceptions of the world, different ways of looking at the same things, and different ways of spelling, or should I say that a lot of users don’t spell correctly.

You can’t just ignore misspelled search queries. You have to deal with them, and make the best of them. This work has to be done by your site search engine. Trivial typos, such as “foot saps” instead of “foot spas” should just be silently dealt with.

Spelling this phrase as “food spas”, “fot spas”, “foot sbas” or “foot spaas” must also be dealt with silently. At most, ask the user gently if they meant “foot spas” – like Google does.

Even when I get some results from this site search, I’m frustrated due to the lack of filtering, sorting and searching within results. Especially sorting, which I find very important on a daily basis.

Let’s take a look at another example. This time, I found a site running on another quite popular E-commerce platform, osCommerce. The site I found is Beauty by Nature, an Australian Aromatherapy, Skin Care and Perfume shop.

I’ll do exactly the same thing as I did with Homedics. So I found the product category “ANTI AGEING SKIN CARE”. To increase the odds of getting some results, I’ll type “anti ageing” in the search box. Which was very difficult to find, located at the bottom left of the page layout.

This search returned the, almost expected, no results page. As I mentioned, this is a classic mistake with site search engines, and it will turn out to be very costly for the shop. I think most of the users are searching for anti ageing products on this site leave immediately.

As the last test, I’ve found a Shopify based shop, Rebekka Guðleifsdóttir Photography. I can’t do exactly the same as I did with the other two examples, because they don’t have product categories. So I just did a search for a product, and of course I got a result. But selling photos and not displaying images on their search result page is a huge mistake. And I still haven’t got any filtering, sorting or search within capabilities.

Misspelled words are not dealt with here either, as I actually expected.

The state of site search on average E-commerce sites is just not good enough. They don’t include enough results, and they haven’t got enough features for the user to narrow down and sort search results. This makes the user experience very bad, and it will cost sales.

The Important Features of Site Search Checklist

Sorting is very important for the user, because it is the fastest and easiest way to manipulate the result set without exclusion. Being able to sort, in both directions, by price, name, customer rating, most sold, discount etc. is very useful. You will quickly see that a lot of users will sort by customer rating, simply because most users want another person’s opinion on something before buying, or by price because they’re looking for something cheap.

Filtering is another very useful feature of site search, and product lists in general. Whenever you perform a site search for a very broad keyword, like monitor you will get a huge result set. In order to find a monitor you want, filtering lets you narrow down by a set of properties like brand, type, size, VGA, DVI, HDMI, display resolution, energy efficiency, contrast level and what else describes a monitor.

Filtering is often a set of links on the right side of the page with a title like ‘Narrow down your search’.

Search within to further narrow down your result set. So instead of narrowing down your result set through filtering, you want to narrow down by search. When searching within, all sorting and filtering settings should be kept.

Respect personalization settings: This is one of the most underestimated things on websites. What personalization settings, like sorting reveals about a customer  is a lot more than you think. As a user, you expect your settings to be respected throughout your entire visit to a given site and even beyond visits. So if you’re on the hunt for something cheap, all lists and search result pages will always sort by cheapest first.

This is good user experience, but way too many sites don’t care about you in regard to this.

Misspelled search terms, autocorrect and synonyms. You probably have both non-educated, educated, and over-educated users and I can assure you that they’ll spell even the simplest of things differently. Well, how many different ways do you think users will spell “Britney Spears”, or is her name “Britanny Spears”? Take a look at Google’s report on Britney Spears spelling correction, and think for yourself how many different ways users might spell your product names.

What are those users going to do when they see a stupid no results page that doesn’t help them get on with their life? Yes – they hit the home, exit or search button of their browser and leaves your site!

Logging all search terms is a must have feature for you as an e-tailer. You have to be able to use search terms as Business Intelligence tool that helps you understand your customers better and better. Logging should also include result set size so you can see how many results are returned for each search term, and also a date and time.

Suggest alternate products when a no results page is found for specific search terms. If you do a site search on a website for “Sony TV” but the site doesn’t have any Sony product, but loads of Philips – wouldn’t you be happy to take a look at the Philips TVs?

This is where your search term database comes in handy. As an e-tailer you need a feature that lets you see all search term that gives your users a no results page. Then you need to take the Sony search term, and link it to a Philips search term so the suggestion is made to your users for future searches.

Conclusion

Internal site search on your E-commerce website is extremely important. Yet, a lot of popular E-commerce software platforms don’t give you the quality you need.

Users are left even more frustrated than before, and it’s a very costly not to have a good site search engine for your business.

So go out there and find a 3^rd party site search engine if you are not able to change your E-commerce software. Make sure that you get the features outlined above, and that your results page is shown on your own site, with your site’s look and feel, which is not always the case with 3^rd party search engines. A 3^rd party search engine is not free, but I’m sure it will pay its own price very fast.

First feature video is here. I just wanted to show how smooth and easy browsing your products in Milkshake will be. When dealing with large datasets, it is important to have an easy an right out accessible way of moving back and forward. Paging data should also be fast, which is why we’re paging data using AJAX calls to prevent the screen from refreshing. Your screen’s scroll position should not change at all, and you should be able to click the paging buttons multiple times, quite fast.

Another variant of the earlier explained product tagging featureis also shown in the video. The one showed here is not a replacement – but just another variation. We like the tagging method that is used all over the Internet – but for some things a Parent-Child hierarchy serves the case better. So as a shop-owner on Milkshake, you can chose between ordinary tagging, and Parent-Child tagging.

During our development process of Milkshake – an E-commerce platform for startups and established companies who wants to sell stuff online – we’ve spent a lot of time thinking about the best way to organize a list of products in a store-front. Not only the store-front, though. But also the shop’s management interface.

It is extremely important to our users that they are able to:

1. Get an overview of how a product is organized
2. Re-organize the product in as few clicks as possible

In the past, our custom E-commerce solutions used the concept of a tree structure for product categories. This is a classic way of organizing products in a store-front, but it’s also very hard to achieve the 1st goal that I pointed out above. When a visitor is presented a 4 level tree-view, products seem too many clicks away and it’s difficult to browse.

Re-organizing categories in a tree structure is also quite hard. It requires a lot of screen reel estate, and a lot of clicks.

The tag-cloud is very easy to use for the visitor of the store-front. It’s a single-click experience, and you can instantly see which ones are the popular tags – in terms of number of products contained inside that tag.

As for the management interface, I did a search for a UI Pattern for tags, and found the following:

I liked the small size of it and its clean design, but missed one thing: A way to click tags instead of typing them.

Typing the tags gives us the following issues:

1. It is easy to do a typo
2. It’s hard to remember which tags is already present in the store-front
3. When more than one person manages an online store-front, they most probably have a different definition of which tags a product should have, and they could end up with close similarities which will confuse a visitor.

We decided to extend the UI Pattern above, so the first prototype looks like this:

 

You can click the links to select / deselect a tag. If a tag is not in the list you can simply type it in the textbox, just like the UI Pattern we discussed before.

In general I like the idea behind the latter, but I still think it needs some final adjustments…