homepage Welcome to WebmasterWorld Guest from 54.198.42.105
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member
Visit PubCon.com
Home / Forums Index / WebmasterWorld / Ecommerce
Forum Library, Charter, Moderators: buckworks

Ecommerce Forum

This 51 message thread spans 2 pages: 51 ( [1] 2 > >     
Site Search - Weakest function of most commerce sites
And never discussed here before
jsinger




msg:3044997
 9:37 pm on Aug 13, 2006 (gmt 0)

As commerce sites grow in complexity and product offerings, the shopper now faces the arduous task of finding the product he wants, even when the site offers it. Years ago we offered only 20 best-selling products and grouped those in a few simple categories. That worked well.

Now we offer far more products and some don't fit into neat product groupings. We've been forced, like most sites, to rely on a "google-like" Search Form that allows customers to enter their own keywords.

Sounds great, until you analyze what customers look for and what they find... Often NOTHING

Last month we installed software that reveals how poorly our search process works. Furthermore, running the same searches on competitors' sites showed their search forms were often worse than ours.

Product searches rely on words found in product descriptions, product names and key words the site owner supplies. The trick to getting the keywords right is to think like a customer (or 1,000,000 of them!).

A sporting goods site may sell "baseballs," but in most cases a search for "base balls" or "baseball" returns nothing. Certainly "base-balls" or "base bals" will come up with zip.

You think the customer will modify his search and try again? Our research indicates they rarely do that. One search and often they're gone

I can't get over how stupid some customers can be in searching. Many searches contain obvious typos or misspellings. Incredibly, one customer insisted on running together multiple words: "needtobuywidets."

Site owners and webmasters fail in the keywords they supply. Shoppers tend to search in plurals, [widgets], yet many searches only turn up the product in singular. That they fail isn't surprising. Getting the keywords right is hard work and it has to be polished thru months of research that few sites conduct. Small sites are utterly clueless, of course.

What's the answer? First avoid free form searches if possible. At best they're inadequate. At worst, they're a site killer. Second, realize their limitations and provide search by categories, too. Obtain software to see what is being searched for, and polish and repolish the keywords your site presents.

Do you have any tips to share in helping visitors find offerings?

 

FalseDawn




msg:3045031
 10:43 pm on Aug 13, 2006 (gmt 0)

Personally, I don't offer a search facility, and don't use them for the very reason you stated - it is very difficult to make it "intelligent" enough to be of any use.

Well thought out categories and easy drill-downs should reduce the need for a search anyway.

Also, I find that the best conversions happen when a potential customer lands directly on a product page (the search engine has already done the search for me, so why waste time trying to replicate their algorithms (even if i could)?)

If you are relying on a customer finding your product(s) through a search, I think you are doing something wrong, and as you said, a poorly implemented search is often worse than none at all...

Bewenched




msg:3045130
 12:53 am on Aug 14, 2006 (gmt 0)

We actually implement a sub field that has alot of mis-spellings in it for items that we sell. However we don't display them to the bots or customers on the front end. It only comes into play within the search function.

BeeDeeDubbleU




msg:3045283
 7:13 am on Aug 14, 2006 (gmt 0)

Well thought out categories and easy drill-downs should reduce the need for a search anyway.

Absolutely! And if you have too many categories of unrelated products then perhaps you should have two or more websites. Nothing is more off putting than a site that sells model trains, footwear, baseball caps and garden furniture.

If you must do this have a look at how Amazon or some of the other major players do it.

jetboy




msg:3045457
 11:15 am on Aug 14, 2006 (gmt 0)

If your site is big enough to warrant a free text search, then getting rid of it completely is going to cost you sales; probably more than its lack of functionality costs you already.

Yes, you should have a browsable product hierarchy, but assuming that it will fulfil everyone's needs is a mistake.

Bewenched is talking about a controlled vocabulary - often described as a taxonomy, thesaurus or synonym ring. Adding a controlled vocabulary can be very labour intensive, but it can be built up over time, gradually improving the usability of your search.

Your first step should be to pick up a copy of O'Reilly's Information Architecture For The World Wide Web, which explains how it all works. This book is gold dust in so many ways. You could also do a lot worse than getting hold of 37 Signals' Defensive Design For The Web, which may save you from screwing up the implementation!

jsinger




msg:3045757
 3:52 pm on Aug 14, 2006 (gmt 0)

Does anyone else have the ability to monitor site searches?

I find such software incredibly useful in sniffing out new products to carry and in fine tuning the keywords we provide for products we already offer.

--
Isn't it interesting that there is so much talk about cart abandonment (on WebmasterWorld and elsewhere), but nothing about visitors who bail out at the search function?

haggul




msg:3045782
 4:03 pm on Aug 14, 2006 (gmt 0)

We monitor and log all searches - and have reports based on that info that help us place products further up the pecking order.

Also our search engine takes into account spelling issues, random word ordering etc etc - it does help. You have to cater for all - some like drilling down knowing that they are refining their selection as they go - others want to bang in a keyword and click a button.

jsinger




msg:3045825
 4:42 pm on Aug 14, 2006 (gmt 0)

Have you also also noticed that visitors often give up after just one unsuccessful search? That's the frustrating thing I see. Perhaps they switch to a category drill down, but we don't have the ability to determine that.

Our customers aren't techie types, btw.

digitalghost




msg:3045875
 5:20 pm on Aug 14, 2006 (gmt 0)

This issue has come up on the board a few times. Cases have been made for removing the search function entirely from thr site, to moving the search feature to make it less prominent.

>>one search and they give up

Often true if the search function is poor. If the person is at a widget site for Alabama and a search for 'Dacatur Alabama Widget Agency' returns no results, the perception is that 1. The site has no information on the subject. 2. The search function sucks so why bother?

When I first started looking at drop-down menus, (quite a few years ago), there was a marked drop-off in traffic that correlated almost perfectly with the depth of the page within the menu structure. It's also a mistake to think that people like to drill down and refine their search. Some simply have the tenacity required to continue their search.

Surfers are more willing to use drop-down menus now, but in all the tests I've seen or conducted, it's better to go broad rather than deep.

A clear and concise static navigation structure yields the best traffic and conversion results, but 'clear and concise' sometimes becomes a challenge when there are thousands of product offerings.

There are a couple of things to try.

1. Move the search box so that it isn't the first thing people see when they arrive.

2. If you find people are arriving at your pages from search engines and quickly navigating to other pages or using the search function, it's time to assess those pages and their placement to see why those pages aren't providing the sought after information. Sometimes it is as easy as removing some language from the page to ensure that it doesn't rank for a term that it wasn't intended to rank for.

3. Parse the query data and use that data to provide alternative searches within the navigation structure of the page. Similar to Google's 'did you mean' function.

4. Evaluate your categorization structure. This is sometimes the reason for pages ranking for unintended terms. Witness the poor 'toy' site that sold toys for adults and for children. Not those toys, but things like 3k MP3 jukeboxes. Unfortunately, the cat structure looked like:

Adult>Toys>Electronics>Gadgets

People searching for 'adult toys' were not amused to find a jukebox.

5. Take a hard look at other directory and categorization structures out there and ask yourself if you would like to navigate them to find a specific product. Try them out. How easy or hard was it to find what you were looking for? Compare those results to your site and make the necessary changes.

6. Don't look at a search box as an easy solution to finding products or information on your site. It seldom is. People gravitate toward the search function if it is at the top of the page and tend to ignore the navigation elements of the page.

7. Use your log and query data to refine your page elements.

8. Give serious thought to your taxonomy and ontology. Use those structures to provide intuitive navigation.

ispy




msg:3046157
 9:15 pm on Aug 14, 2006 (gmt 0)

Early childhood education, and reduce risk of exposure to lead and other toxic substances.

lorax




msg:3046753
 10:42 am on Aug 15, 2006 (gmt 0)

Have you also also noticed that visitors often give up after just one unsuccessful search? ... Perhaps they switch to a category drill down, but we don't have the ability to determine that.

Questions:
1 How do you know they've given up?
2 What exactly are you tracking if you're not tracking category level clicks?

>> I can't get over how stupid some customers can be in searching.

Are you sure someone isn't testing your site search tool?

jsinger




msg:3046903
 1:29 pm on Aug 15, 2006 (gmt 0)

Questions:
1 How do you know they've given up?
2 What exactly are you tracking if you're not tracking category level clicks?

Our monitoring software records text searches by keyword, IP and time. We can see how many results were returned. (often none, or sometimes far too many to be very useful) "Giving up" means they've stopped entering keywords in the search form. Then they either leave our site or drill down thru category links which probably works for most customers. Our site is quite targeted with only a few hundred products.

"I can't get over how stupid some customers can be in searching."
Are you sure someone isn't testing your site search tool?

Well I've been busy testing other sites' search tools, so I guess that happens. :) But I don't just enter one keyword. I often test in singular and plural to see how searches handle those. The ones at our site are real customer searches with maybe an occasional supplier or competitor.

lorax




msg:3047285
 5:38 pm on Aug 15, 2006 (gmt 0)

So when the search results are returned I assume they're a combination of category and product pages. This may seem painfully obvious but just so I'm sure - are you tracking the click thrus on the search results?

Re: testing. I've gone to sites and tested their search tool using a variety of different words and miss-spellings to see how well their sites handle them. I've yet to find one site that has a search tool that works well. I don't think that search should be given lower prominence though. It's a viable tool and in the hands of someone who knows how to use it - it's essential. I'd suggest you don't punish those who can use it wisely just because you have a few jugheads. ;)

jsinger




msg:3047539
 8:09 pm on Aug 15, 2006 (gmt 0)

So when the search results are returned I assume they're a combination of category and product pages. ... are you tracking the click thrus on the search results?

Search results show customers individual products with a picture, brief description and a link for more info and ordering. Our software tells us the search words used and the number of products returned. We don't have any way to see whether that customer actually clicks the link or orders the product. I guess I could compare search IPs and order IPs to see what is ultimately purchased, but that would be a ton of work.

Re: testing. I've gone to sites and tested their search tool... I've yet to find one site that has a search tool that works well. I don't think that search should be given lower prominence though. It's a viable tool and in the hands of someone who knows how to use it - it's essential. I'd suggest you don't punish those who can use it wisely just because you have a few jugheads. ;)

In building a site, the jughead vs. WebmasterWorld genius ratio needs to be considered. Our site is pretty dumbed down. And I think that's the way 99% of commerce sites should go. Consumers are used to shopping in categories:
Go to mall>>
Find appropriate store>>
Locate widget dept>>
Select product>>
check out.

For a very small site, well planned category drill down is all that's needed. A search form might do more harm than good. And for a huge Amazon type site a good text search is essential. But even Amazon offers both. I'll guess that most book customers use the form, while apparel customers will drill down while looking at pictures.

Because major sites offer search forms and tiny sites often don't, many e-commerce retailers probably conclude that search forms are the better way to navigate. Actually they're just a necessary evil even for large sites. I don't think many webmasters understand how deficient they usually are. If they did, there would be far more discussion about this topic. (compare, cart abandonment)

Where to put the search box? Unfortunately the trend has been to stick it in the customers face, right at the top of each page. Amazon and Wal-Mart do it that way. But I think that with small focused sites, text search should appear in a less prominent spot. Indeed, only a small number of our visitors use the text form. I presume the rest drill down or use search engines to land on the desired product at the outset.

sniffer




msg:3047853
 12:17 am on Aug 16, 2006 (gmt 0)

so what about the search options some carts have,eg 'All words', 'Any word', 'Exact Phrase', etc?

Any interesting findings there? Also, where can you get this software (or what can i search for to find it? lol )

jbgilbert




msg:3047887
 12:42 am on Aug 16, 2006 (gmt 0)

Right On!

This was my #2 gripe about most web sites

#1 is the ones that makes me scroll right and left!

hannamyluv




msg:3047908
 1:04 am on Aug 16, 2006 (gmt 0)

The real problem with search is the user. And that is just not an option for blame with ecomm.

Most people are too impatient to use categories. That works fine is you carry a 100 or so products but the more you have, the less likely a customer will spend the time scolling through categories.

Search engines (the big ones, not site ones) have trained people to think that search is easy and quick. So they assume that all are easy and quick.

Take away search or try to hide it, and I bet you will lose sales.

Your best bet, if you have a lot of products is to start to build a database. Get a programer to write up an add on to your search that allows you to redirect searches. Have someone go through the analytics and once a week add in the words that had no results. Using the example above, write your search so that if they search for "baseballs", it looks for baseball. If your seach checker sees that someone searched for "base bal", add that to the mix. Someone types "white ball with red stiches" redirect them to baseball.

You don't catch the first one, but you just took care of anyone later.

Yes, it is time consuming, but eventually you can build up a decent database that is custom designed for your site and customers. Which is something that no off the shelf system could give you.

tedster




msg:3047923
 1:29 am on Aug 16, 2006 (gmt 0)

I think that with small focused sites, text search should appear in a less prominent spot.

I'm quite partial to having one dedicated site search page, and then linking to it on the main pages, rather than putting a big old empty text box on every page -- that just begs for some filling in. When I've changed site to this approach, their stats usually go up significantly, unless their Information Architecure really sucks, but that's a whole other thread.

Using site search logs is SUCH a big help for a site. Hannamyluv has nailed it -- and it only hurts in for the first few months or so. Also, you can get some great ideas for new products, articles and even Information Architecture changes.

I started a thread 4 years ago on this topic -- it still has some good stuff in it for today:
[webmasterworld.com...]

graywolf




msg:3047934
 1:39 am on Aug 16, 2006 (gmt 0)

When I "worked for the man" I wrote some code that did that logged all searches with zero results. After building up bit of data we did one of two things. Some searches we automatically redirected to the right terms for example [garbage cans] got redirected to [waste receptacles]. The merchants refused to change the name to [garbage cans] insisting they didn't sell "garbage". The second was having a database of likely possible other matches. For example customers searched for [dishes] and the merchant called it [formal dinnerware] and [casual dinnerware]. So the customer got something like

Your search for "dishes" returned 0 results

Did you mean to search for any of the following terms:
Casual Dinnerware
Formal Dinnerware

It takes a lot of time and is really going to "hurt" in the beginning but you can improve things and hopefully learn a bit in the process.

tedster




msg:3047940
 2:00 am on Aug 16, 2006 (gmt 0)

Your examples are superb, graywolf. This process can really hurt those "image oriented" folks in the corporate marketing office who think they can control the way the pubic speaks. Site Search logs can be a big wake-up call.

In the wisest of all Marketing Depts, they would throw caution to the winds and also try to rank for "garbage cans" on the web search engines, but things don't always go that way. Redirecting Site Search results is a decent first step.

jsinger




msg:3047946
 2:09 am on Aug 16, 2006 (gmt 0)

so what about the search options some carts have,eg 'All words', 'Any word', 'Exact Phrase', etc?
Any interesting findings there? Also, where can you get this software (or what can i search for to find it? lol )

We use the Shopsite shopping cart. Its built-in search function is pretty rudimentary, at least the way we have it set up.

The search analysis module is a freebie--for the moment anyway-- that our host, Lexiconn, offers to their Shopsite users. They also offer a paid Cart Abandonment Module that we bought a few months ago. I find the search module to be the far more interesting and useful tool. I guess Lexiconn might offer it to outsiders at the right price. I have no idea how compatible theirs would be with other cart/search systems.

tedster




msg:3047957
 2:30 am on Aug 16, 2006 (gmt 0)

one customer insisted on running together multiple words: "needtobuywidets."

This can be a really tough one to deal with. For one client we had a custom routine written for "zero results" that pealed off the characters one at a time until there was a match. That's brute force and not so precise, but it does give some results.

From regular search keywords in the server logs, I can tell that many people do run-on typing. I'd say you're lucky of only one person did it on Site Search.

jsinger




msg:3047969
 2:47 am on Aug 16, 2006 (gmt 0)

Great discussion on this topic you had here in 2002 Tedster:
[webmasterworld.com...]

I thought I was breaking new ground with this subject. LOL! Interesting that most posters were very leery of search forms even then.

Funniest comment was from Brett Tabke about the word "Webmaster" being searched for here several times a day.

rogerd




msg:3047971
 2:52 am on Aug 16, 2006 (gmt 0)

Too many search results is probably a more common problem - like searching here for "web" or "google".

I think the research value of site search is worth whatever inconvenience it creates.

Plus, a good site search function will let you run keyword-targeted ads along with the search results. This gives you a chance to promote high margin or important products to related searches. For example, if someone searches for "tops widgets" but your best-selling, highest margin widget products are Acme Widgets, you can show ads for them along with the "tops" results.

ronburk




msg:3048147
 6:02 am on Aug 16, 2006 (gmt 0)

it is very difficult to make it "intelligent" enough to be of any use.

Is it? Haven't tried it, but I'm not convinced this is true for modestly-sized websites and a programmer reasonably interested in the problem domain. In those cases, there's a good chance the searches cluster heavily in a manageable number of "roughly the same" queries. Likewise, for a mid-sized website, there's a good chance that the best results are not dozens of "hits" ranked by "relevance", but custom-designed text that lists the only one or two plausible destinations for that topic.

Seems like a fairly simple framework built around a decision tree that you tune (and retune from time to time) based on actual searches and contents might do OK. Results might stink for the visitor needing to use Boolean operators, or trying to match some obscure phrase, but such folks are probably savvy enough to just switch to Google site: for that kind of search. When you've only got a few hundred pages, I bet it doesn't take a terribly big tree to match most of the queries with the most likely destination URLs.

I suspect that search algorithms designed to adapt to "any" content are the wrong way to go for a modest-sized company website.

smells so good




msg:3048181
 6:26 am on Aug 16, 2006 (gmt 0)


...one unsuccessful search? That's the frustrating thing I see. Perhaps they switch to a category drill down, but we don't have the ability to determine that.

We don't have any way to see whether that customer actually clicks the link or orders the product.

Don't you have access to your server log data? IP tracking through the site can give you a very clear picture.. notwithstanding the AOL users.

I haven't implemented site search. Suddenly I can see more benefit to a decent search than ever before. Great discussion.

pageoneresults




msg:3048191
 6:50 am on Aug 16, 2006 (gmt 0)

We've been using Google Web APIs for Site Search with pretty good success. No need to worry about all the usual stuff being discussed. It's worked like a charm and the clients use it themselves to search their own sites. Yes, there is the 1,000 daily query limitation, but it serves it purpose for the smaller client.

I've seen some very unique implementations of site search. I know of a few that actually generate search-term.htm pages from the query. Neat little setup and very impressive end results page. :)

Large scale ecommerce sites can always purchase a search appliance. It's an investment that will pay for itself rather quickly.

Oliver Henniges




msg:3048192
 6:53 am on Aug 16, 2006 (gmt 0)

Basically I'd support in the first place what FalseDawn said in msg #2. Nevertheless
For one client we had a custom routine written for "zero results" that pealed off the characters one at a time until there was a match. That's brute force and not so precise, but it does give some results.

is a very interesting idea. Beginning with php I thought the levenshtein function would provide some help, but in fact it doesn't really.

I was just thinking whether some "brute force lexical similarity" might work. Not as sophisticated as current state of LSI but perhaps simply some a-priori-calculated (and stored) value resulting from a discrete cosine transformation over the ascii-strings of the product description on the one hand and that of the searchphrase on the other might reveal significant matches. Don't forget that the lexical basis of the pool is already narrowed down heavily by the customer being on a specific site. Just an idea, my mathematical competence doesn't suffice to elaborate that in detail.

lorax




msg:3048348
 10:54 am on Aug 16, 2006 (gmt 0)

I'm quite partial to having one dedicated site search page, and then linking to it on the main pages, rather than putting a big old empty text box on every page

That's a good idea Tedster. Force them to choose between clicking on the link to make a search or on something else that looks more likely. And yes, there are a number of other issues that we're assuming have been dealt with that would make this approach safe to use.

It also strikes me that perhaps we've made it too easy for the customer to shoot themselves in the foot. graywolf's idea is a good quickfix but it needs constant attention to make sure you capture all of the crazy ways people search. Then you need to determine what they were actually searching for and assign it to one or more responses. But making the customer think for a moment about whether to use the search tool versus using the navigation provided buys you a more qualified click rather than a shot from the hip.

neesheeth




msg:3048355
 11:04 am on Aug 16, 2006 (gmt 0)

Pageoneresults: "We've been using Google Web APIs for Site Search with pretty good success. No need to worry about all the usual stuff being discussed. It's worked like a charm and the clients use it themselves to search their own sites. Yes, there is the 1,000 daily query limitation, but it serves it purpose for the smaller client."

That's an interesting application of Google Web APIs. However when I went and checked their faqs, it says that the APIs cannot be used for commercial applications.

This 51 message thread spans 2 pages: 51 ( [1] 2 > >
Global Options:
 top home search open messages active posts  
 

Home / Forums Index / WebmasterWorld / Ecommerce
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved