As we’ve worked with different e-retailers to create a more effective shopping experience, the question inevitably comes up “How Many Facets do You Need?” The answer of course is, “it depends.” When you have few products in a category you may not need any facets, but when you have many products facets become essential. For 30 products, 2 or 3 facets is usually plenty. But when you have 500 products it might take 6 or 8 facets to enable the visitor to filter the list to a manageable number. Ideally we like to get a list filtered down to 1-5 products. If there are many more than that you start to get too many choices, and decision making becomes more difficult. So the number of facets you need really depends on how well they divide your product listing. It may be that you have 3 facets, but they are not structured in such a way that you can ever get the list down to a manageable number. This is why implementing parametric search is so challenging. In most cases it will not be sufficient to just add filters for Brand, Price, and Customer Rating. What we try to do at FindWAtt is extract as many attributes/parameters as possible, and then calculate which ones are most effective in reducing the product selection per click. This means that it will take fewer clicks for the visitor to narrow down the product list, and get the visitor to a purchase decision more quickly and effectively.

While assisting a friend of mine in shopping for a new car, I recently stumbled across an impressive implementation of faceted search on Kelly Blue Book. It's called the Perfect Car Finder, and allows you to search or browse, and refine your selection based on a multitude of different attributes.

According to the press release in 2007

"...nearly 80 percent of shoppers who visit kbb.com have not yet decided on which model they plan to purchase and less than half know which make they are interested in. With an ever-growing need among vehicle researchers for help in sifting through hundreds of vehicles available today, kbb.com has launched two enhanced online shopping and decision tools to assist car buyers early in their shopping process.

The first of these enhanced tools is Kelley Blue Book's 'Perfect Car Finder®,' which lets consumers search through more than 400 new vehicles and more than 1,100 trim levels of those vehicles by the features and optional equipment that matter most to shoppers. The second is an all-new comparison tool allowing consumers to view an all-encompassing side-by-side comparison of vehicles."

There are basic options for people to shop by Make, Model, Body Style, Size, Price and Gas Mileage. There are also advanced options for people to shop by Engine Type, Horsepower, and Various Interior/Exterior/Safety Features. It's a shining example of how I'd like to be able to shop for many of the products I buy.

Kelly Blue Book is just one example of the strides made in faceted search. Particularly in the Automobile Industry, there are several examples of faceted search. Yahoo Autos Car Finder is another example of a very impressive set of facets on which to refine your search.

It is not surprising that the Automobile Industry is adopting these techniques. Cars are very expensive and complex products, which have many of different variables that need to be considered. Like computers, they also have a highly structured set of attributes and attribute values, making it a natural fit for high quality faceted search. Add to that the market research by KBB that 80% of shoppers don't know which make or model they want to buy, and guiding the purchase decision becomes critical to success.

Have you seen any other Automobile related sites that employ impressive faceted search? Do you agree that it is a good fit? I'd be curious to see any research into how much this type of shopping interface improves conversion rates in Automobiles.

E-commerce websites contain lots of product information that is underutilized. I know this professionally because structured product data is our business. Macys.com, is a favorite shopping site of mine, and was the obvious place to go when I needed to get a blazer in a hurry for our trip to the IRWD conference.

Normally when I shop online, I have some familiarity with the product I'm buying. But in the case of blazers, I was clueless. I didn’t own a blazer and didn’t know any of the relevant attributes. All I knew is they needed to match the various slacks I own, and of course I wanted something with style.

Unfortunately, after searching for "Mens Blazer" and refining by Blazer > Mens, I became stuck. I was presented with a list of 134 blazers and limited options for navigation.

I could narrow results by Brand, Special Size, Mens Waist Size (which seems wrong for a jacket), and Color, but I wasn't ready to make decisions based on these attributes. I wanted to get a sense of the Materials, Styles, and Patterns that were available.

Being in the product data and attribute business, I knew this information was probably available to me, just not easily accessible and usable. To get at this unstructured information I would have to click on the product details for each item.

I found one blazer that looked pretty clean, and looked at the attributes:

- Material: Linen; cotton

- Style: Two button front; Flap pockets; Side vents

- Lapel: Notch

It has some features that I like. But are there any others like this one? What other kinds of materials do they have? Do I have to buy a blazer that is Dry Clean Only? What about patterns?

I had recently read a Smashing Magazine article about How Hard It Can Be to Shop at Macy’s. They were shopping for bed sheets, but experienced the same limited filter options. I could have left the site, but I was confident that Macy's had a good blazer, and I was determined to buy one. Macy’s faceted navigation was failing me, so I took matters into my own hands.

I grabbed all of their product data for blazers, and ran it through our system, to bring out all of the unemployed attributes and attribute values that were present. Look at the rich structure that was produced:

By adding important attributes to help filter results, I turned a potential hour long shopping session into a quick, focused buying session. I gained confidence that Macy’s had what I wanted, and spent time purchasing rather than exploring. I ended up going to the store to make sure the blazers looked good on me (decided to buy 2), but that is the subject of another post.

If Macy’s were to make these attributes available to everyone how much of an improvement in key shopping metrics (bounce rate, conversion rate) do you think they’d get? And how much would the clickstream of the attributes chosen by consumers to refine their search be worth to Macy’s in terms of tailoring their website to meet customers needs?

When you’re running a startup, there is never enough time to do everything you need to do. Blogging can easily get overlooked. But what better way is there of sharing your experience and expertise with the world. One thing that doesn’t help facilitate regular blogging, is the sloppy interfaces presented in Word Press and Drupal. Especially since the people who can contribute the most to a blog are not necessarily the system administrative type.

Desktop Blogging to the rescue! There are a couple very nice programs that allow you to create clean, elegant blog entries, and make the blogging experience much more enjoyable.

Windows Live Writer (FREE) – this is a very good desktop client for blogging. Once you have set it up, it is very easy to compose blogs, insert images, and post them to your site. You can even go back and edit them without ever having to login to your website.

BlogJet (30–day Trial; $40 to buy) – this appears very similar to Windows Live Writer but has some extra added power. We think this program may function well not only for blogging but also for creating content for our website (e.g. pages).

With that said, I’d also like to share my experience in setting one of these programs up to work with a blog. For blogging novices like me, it can be very difficult and frustrating to find information. Especially with something as dynamic as Drupal (an attribute-based content management system), with much power comes much responsibility. So for anyone who is having the same problem I had, here is a lead on how to set up a desktop blogging client. They are all similar, but I chose the free Windows Live Writer.

Setting Up Drupal for Windows Live Writer

For Administrator (i.e. if you haven’t set anything up)

1. Download the Windows Live Writer BlogAPI from Drupal.org and extract it into a folder (make sure the version you download is compatible with version of Drupal you have, and make sure you remember where you put the folder)

2. Install the module using a FTP Client like FileZilla.

a. Once you have installed FileZilla, run the program, and click File -> Site Manager, click New Site.

b. Add your website/blog URL under Host (e.g. findwatt.com), Servertype=FTP, then add the username and password given to you by your website hosting company.

c. Once you have connected to the FTP, navigate the Remote Site folder structure (on the right) to find the modules folder (/html/modules). Then navigate the local site (your computer) and find the folder named after your module (i.e. wlw_blogAPI). Select both folders (so they’re highlighted grey).

d. Right click on the wlw_blogapi folder and select Upload. Now you’ve uploaded the module to drupal.

e. Now go into your Drupal admin and enable the module. And your’e done with the setup.

 For User (i.e. once admin has set it up, here’s how to connect)

1. Download and Install Windows Live Writer (you may need to sign up for a Windows Live account first)

2. I can’t give any better instructions than this post (from the guy who wrote the module)

The 3 key processes that determine whether your customers can find what they want

Have you ever been shopping in a grocery store, and one of the store employees politely asks, “Are you finding everything ok?” If you are like me and don’t know the layout of the store like the back of your hand, you take the opportunity to ask the clerk exactly where everything on your list is located. Otherwise you end up wandering up and down each aisle, looking at the shelves, hoping to find stuff. This is the basic concept behind findability: How easy is it to find what you are looking for? And in most grocery stores, I have found the findability to be very poor.

In his book “Ambient Findability” Peter Morville defines Findability, dictionary style:

find-a-bil-i-ty n

a. The quality of being locatable or navigable

b. The degree to which a particular object is easy to discover or locate

c. The degree to which a system or environment supports navigation and retrieval

On the web, and specifically for e-commerce sites, findability means how easy is it for a shopper to find the product they are looking for.

This is extremely well put, and I can’t say it more clearly. For e-retailers, having a website where the products are findable is critical to their ability to compete. If findability is poor, sales and customers are lost because…You Can’t Buy What You Can’t Find!

But how do you judge findability? There are three basic processes to be evaluated:

1. If you know what you want, can you find it? – This is called known-item-search, and appropriately, the search box is most often used. If I am looking for an 8 GB iPhone 3G, and I search using that exact phrase, do I get those results right away? Or are the first results something unexpected, like accessories for the phone? Or nothing at all?

2. If you have some idea of what you want, but do not know exactly, can you find it? –This is where findability is actually most difficult to achieve. You know you want a coffee maker, but which kind? Ideally you have some knowledge of coffee makers and which kind you prefer. But it may be you don’t have a clue. How well can you find the right coffee maker for your situation?

3. If you’re just browsing, will you easily discover something you want to buy? – In this situation, you are not focused on specific product type. But can you navigate through all the available choices? And to what extent can you refine your search.

This evaluation is far from academic. Findability has real business consequences. If you can find something easily, you have more time to decide whether to buy, and continue shopping for more. For the e-retailer and shopper alike, there is nothing worse than not being able to find the product they might buy. And worse, frustration in finding a product causes disillusion, preventing a transaction, and possibly losing a customer.

Will Evans, founder and Principal User Architect for Semantic Foundry, sums it up nicely, saying:

Increased findability leads to increased business results

• More people find what they’re looking for – faster – thus improving conversion rate
• Increased customer satisfaction and loyalty
• Decreased customer service cost

Opportunities for targeted merchandising

• Up-selling based on selected facets (similar attributes might mean affinities in the customers mind
• Cross-selling based on selected facets
• Each facet selected is valuable customer data (these are attributes customers are saying they are interested in!)

Have you shopped on your own site lately? Try each of the 3 findability processes. Even with known-item search, we find that lots of sites have problems. If you start with only a one or two keyword search, are you presenting your customer with pages and pages of products with no way to filter other than brand and price? Are you forcing your customer to keep adding keywords to the search box before they can get down to a manageable number? Or worse, does your search end up with fewer results than customers would get if they navigated to the relevant subcategory (a common problem)?

This will give you a rough sense of your site’s findability. To start managing your site’s findability, use the Findability Evaluation Framework available in our Findability Review section to identify where you can get the biggest Findability boost with the least effort. Or sign up for a FREE Findability Review, and we’ll do the work for you.

Here @FindWAtt, we want to be a resource for the entire e-commerce community. In addition to being an authoritative source on the fuel that drives success for online retailers—high quality product information—we also connect retailers and vendors (e.g. e-commerce platforms, data feeds, CSEs, etc.) and bring retailers and their customers closer (see @cantfindcantbuy, where we provide a focused look into the frustrating world of online shopping from the customer’s perspective).

In that vein, we have found Twitter to be a fantastic tool for staying aware of and in touch will the groups that are relevant to you, especially with Twitter’s new Lists feature. Here are some thoughtfully researched and organized lists of the top 500 Internet Retailers on Twitter (since 500 is the max), various other internet retailers, and the top players in various e-commerce domains. Please follow these Twitter lists and review them for quality assurance. These are not just our lists, they are everyone’s lists.

Note: These lists will be continuously maintained and improved with more research and your input. Tweet me @danbarbata for anyone I should be adding to these lists! Thanks.

• Top 500 Internet Retailers (on Twitter)
• Top 100 Internet Retailers (on Twitter)
• E-Commerce Platforms
• Multichannel Data Feeds
• Site Search & Navigation Companies
• Comparison Search Engines (CSEs)
• E-Commerce Merchandising (Rich Media) Companies
• Product & Customer Reviews

iStockphoto is an incredible source of images – over 4 million to choose from! Unfortunately, finding exactly the right image is worse than looking for a needle in a haystack. My first experience on iStockphoto was an amazing and extreme example of the desperate need for structured and attributed product data to aid customers in shopping.

With such a huge inventory of products (if you think of each image as a product), I was sure they’d have some way to narrow my selection. After all, I didn’t want to get stuck paging through 20 pages of images…next, next, next.

But that is exactly what ended up happening. I was looking for a picture of someone who was looking lost, so I typed in “lost.” I got back 362 pages and 10,858 results. Two categories were presented to “clarify my search.”

• Lost (Descriptive Position)
• Confusion

I could have used both metaphors for the way people tend to feel when shopping at sites like this, but I chose Confusion. Now I was down to…531 pages and 15,928 results! By being “clearer” I got 5,000 more results. The only way to further narrow my search were checkboxes labeled “photos,” “illustrations,” “flash,” “video,” and “audio.” I chose “photos” and “illustrations” and was left with virtually the same number of items.

At this point I felt stuck. I had some idea of what I wanted, but didn’t know how to find it. I knew two things:

1. I didn’t want a photograph of a person

2. I wanted the image to be blue (to keep with the theme of our site).

But I didn’t know what keyword search would produce my desired results.

Furthermore, I didn’t know what additional aspects of these images I should be considering. Simply put, I had no filters on the sidebar. I didn’t have the ability to filter on color, image type, or anything else. I had to go back to the search box, type different combinations of words, and hope to hit on the right set of images.

I found a few candidate images, but didn’t pull the trigger, because I wasn’t confident I had seen all of the choices. I wasn’t being guided through the experience. I felt like saying “Just give me the 40 images that fit my criteria, and let me decide!” But because iStockphoto did not have faceted search and navigation, this was not possible.

Without faceted navigation, or the ability to refine your search, the shopping experience is frustrating and painful. In my case, I didn’t end up buying anything because you can’t buy what you can’t find!

Men’s Sandals for Cheerleading?

Zappos.com, the wildly successful online footwear retailer that Toby Hsieh has grown from $1.6M in sales in 2000 to over $1B in 2008, and which was recently bought by Amazon for around $900 million, has just unveiled a new version of their website that uses Faceted Navigation (e.g. Guided Navigation, Sidebar Navigation, etc.) instead of Category Drilldown. I am a big fan of both Zappos as a company, and Faceted Navigation as a method of shopping, so I decided to check it out.

The primary question for me is how much were they able to improve the findability of their products. My conclusion is that Zappos’ findability is much improved, but is still not all that it could and should be. They have done many good things, but have missed the boat in some areas and even gone backwards.

Here were my observations:

Primary Improvements:

1) Facets are on the Sidebar instead of at the top.

2) Multiple facet values can be selected from the same facet (e.g. you can look at all brown or black shoes at the same time instead of being limited to one color).

3) More facets are available to select on (e.g. in Men’s Sandals, they have added Theme, Heel Height, and Insoles).

Still Needs Work:

1) I still don’t know what they mean by some of these facets. e.g. what is a “Tactical” theme vs. a “Spring” theme? I need some explanation of what these facet values mean and why I should care. And several of these values seem to be meaningless in certain contexts. Look, I found men’s sandals for Wrestling, Ice Climbing, and Cheerleading!

2)  I don’t know which facets to select, even if I know what kind of product I want. E.g. I can see the kind of sandal I want - it has straps, and an open toe area - but I don’t know which facet values they’ve assigned to that type of product. Show me the attributes of each product! That’s the one I want, but what theme is it? And is theme even the distinguishing attribute?

3) I’d like to remove undesirable results during my session so I can zero in on the products I might consider buying, and have. Moreover, I’d like to filter out all products that have certain facet values, e.g. I’m not sure exactly what range I want to spend in, but I know I’m not spending over $200 for sandals!

4) They’ve removed the frequency of each facet, and the total number of products in my selection. They need to bring this back because it helps keep my bearings and understand the consequences of my next action. If I know there are only three blue sandals, and I can already see them on the page, clicking on blue will be a wasted click.

Overall I find the shopping experience is much more fluid and intuitive. I like not having to think of something to search for, and just letting the facets and facet values guide my decision. But I still feel they are leaving a lot on the table. Presenting the choices (and ensuring the data is properly structured with these attributes) is a huge challenge for many e-commerce sites, and breaking through that is significant. But until those facets can be truly understood by all, people will continue to have trouble finding what they want.