E-Commerce site - but no categories - so bad SEO?

7 Feb 2009 - 9:03pm
5 years ago
15 replies
1410 reads
Den Serras
2009

I'm creating a very large e-commerce site that's expected to
eventually have tens of thousands, perhaps more, entries. It is for
an industry that is heavily categorized, but every supplier has
totally different categories, so they effectively become meaningless.
So we're going to put it all together without categories, or at least
without a category tree model. It's search engine only, baby. All
those former categories are now keywords, effectively allowing one
item to be in dozens of categories.

I'm worried about how sites like Google will index ours. Even if
Google is a tree-free structure, they seem to depend on trees to find
their way around a site. Without any hard links, are the spiders going
to say "oops, there's only one page to this site" and move on? I'm
assuming they're not smart enough to do searches within the site's
engine. Do we create a few hundred likely searches and submit that to
Google as a map? Do we develop some kind of bridge like eBay does so a
Google search turns up keyword items? The client ain't rich and I
ain't an SEO expert... yet... So any help is appreciated.

Thanks!
Den

Comments

7 Feb 2009 - 10:11pm
david grubman
2008

Dennis

I challenge the lack of categories. Something about that bothers me. While
you may not want to push the category as the primary discovery method, I
strongly feel the need to include some browse functionality for customers to
discover what you offer. by using a search only mechanism customers will
never discover products. They will only find the products they are looking
for at the moment.

If you insist on not having a browse mechanism, then I would suggest
creating a site map page that is basically a flat index list of all of the
products on the site and is automatically kept in sync with the products on
the site.

good luck.

- david

--
david grubman
--------------------
Buy your Jeep stuff from www.californiajeepauthority.com

7 Feb 2009 - 10:47pm
Den Serras
2009

We spent a lot of time talking about this. There'll be so many
products, and so many that can fall into multiple categories; and in
the end, our categories will be arbitrary anyway. If you're buying
an art deco couch, that could be under furniture-couches-large or
styles-art deco-furniture or interior-living room-furniture-large or
items-large-rectangular-wide or lifestyle-upscale-expensive or
style-modern-prewar-art deco or colored-two colors-black and white or
lord knows how many others.

Forcing categories means forcing the audience into a particular
mental model, and in the case of our store that has so many users of
different professions, we inhibit as many as we help.

For this, using keywords and allowing powerful filtering will get the
user where they want while taking advantage of their own mental model
(assuming the manufacturer puts the right combination of keywords
in).

Yes, we're limiting the user. But I think most people will prefer
coming to the problem with their own language. I'm just worried that
without Google indexing a big site, we won't even show up on their
radar.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

7 Feb 2009 - 11:02pm
Robert Hoekman, Jr.
2005

>
> Forcing categories means forcing the audience into a particular
> mental model, and in the case of our store that has so many users of
> different professions, we inhibit as many as we help.
>

Sorry to be frank, but this is a cop-out. It's a easy-out answer (that I've
heard before) for companies afraid of making difficult IA decisions. Do you
think Target.com, Amazon, BN, and other major retailers out there have only
users of a single profession with a single mental model? Do you think a
product on one of those sites only shows up under a single category?

Run some card sorts. Study other sites with faceted navigation. Study what
successful sites structured like your own (likely any commerce site) are
doing and how they work.

-r-

7 Feb 2009 - 11:25pm
Den Serras
2009

The point of a category tree is to facilitate a very limited search
based on an initial assumption or mental model. But you have to agree
with that breakdown, and the manufacturer has to have the exact same
definition, or you're actually driving the user away.

Category have the *illusion* of usefulness when dealing with large
sets. Imagine a Venn diagram with multiple fuzzy-bordered bubbles -
which is what the real world is like. If an item is even 1% outside a
bubble it won't show up in a category. Yet the user is forced to make
an arbitrary distinction - is it a comedy or a drama or a family movie
or a children's movie or an animated movie (all could apply to
Finding Nemo). You pick the one that you think is most likely and
hope that you're right. And if you're not? SOL. But if there's no
categories? You skip the whole farce and go right for your OWN logic.

Telling the manufacturers that there are no categories will at least
make them put the categories in the keywords, so folk are no worse
off, or, in my hope, they'll look at other likely ways a customer
might find an item and put in more.

Categories are fine for computer components. It stinks for books,
movies, clothes, people, or anything else that doesn't have only one
possible interpretation.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

7 Feb 2009 - 11:47pm
david grubman
2008

There is no suggestion here that you remove the ability to search products.

All of this can certainly be verified in user testing. Include variations of
only a search, primarily search with secondary categories, and then flip
them. No matter what is said on this board, proof comes from the user.

- david

On Sat, Feb 7, 2009 at 8:25 PM, Dennis Serras <dennitzio at yahoo.com> wrote:

> The point of a category tree is to facilitate a very limited search
> based on an initial assumption or mental model. But you have to agree
> with that breakdown, and the manufacturer has to have the exact same
> definition, or you're actually driving the user away.
>
> Category have the *illusion* of usefulness when dealing with large
> sets. Imagine a Venn diagram with multiple fuzzy-bordered bubbles -
> which is what the real world is like. If an item is even 1% outside a
> bubble it won't show up in a category. Yet the user is forced to make
> an arbitrary distinction - is it a comedy or a drama or a family movie
> or a children's movie or an animated movie (all could apply to
> Finding Nemo). You pick the one that you think is most likely and
> hope that you're right. And if you're not? SOL. But if there's no
> categories? You skip the whole farce and go right for your OWN logic.
>
> Telling the manufacturers that there are no categories will at least
> make them put the categories in the keywords, so folk are no worse
> off, or, in my hope, they'll look at other likely ways a customer
> might find an item and put in more.
>
> Categories are fine for computer components. It stinks for books,
> movies, clothes, people, or anything else that doesn't have only one
> possible interpretation.
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=38306
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

--
david grubman
--------------------
Buy your Jeep stuff from www.californiajeepauthority.com

8 Feb 2009 - 12:40am
Josiah Johnson
2009

I'm going to also push again your decision to give up on browsing
simply because of a complicated or multi-faceted hierarchy. I don't
think anyone here has been telling you to put every single item into a
particular place in a single category tree.

Having a browse function that allows your users to, for example,
"Browse by Size" or "Browse by Style" or "Browse by Function"
etc... will not limit you or your users' browsing ability if you
comprehensively study the different facets and categories in which
users perceive the items in the catalog to belong. Place items in as
many places in the tree as they belong.

Not that this is the best way of doing it, but notice how Amazon has
done it for this item (find the section "Look for Similar Items by
Category"):

http://www.amazon.com/Beck-Childrens-Wooden-Bookshelf-Shelves/dp/B000SOOEC8

Best of luck,
Josiah

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

8 Feb 2009 - 2:23am
Den Serras
2009

Guilty: I'm totally talking out of my butt. I don't have the money
for studies on this, and my client has a very limited budget.

Maybe someone whose client is actually willing to spend money on
testing will try this. Get a few hundred manufacturers together of
soft-category items (not stuff like vacuum parts and computers, more
like music, clothes, furniture, food).

Ask 1/2 of them to pick from a list of categories (or suggest more)
and then come up with keywords.

Take the other half and then ask them to just come up with keywords.
I bet you dollars to dimes that the keyword-only people will come up
with twice as many, maybe more.

Now get two sets of equally brilliant programmers together with the
same budget.

Tell one group they need to come up with the best dang category
browser they can and the best search engine they can. Tell the second
group to put all the money into the search engine. Tell both that the
search engine needs to work like people think. Multi-path, positive
and negative searches, visually assisted (like tag clouds), easy
product removal by keyword. Get two sets of designers, one to fit in
categories and search and the other search only.

Now ask a group of users from very diverse industries to search for a
number of nameless products on those two sites using three or four
descriptors and three or four exceptors. Example: blue button-down
shirt, large, dress-casual; no cotton, no white buttons, no pockets,
not made in China.

If I'm right (a VERY big if) some of the category group will try the
category browse, and I'll bet a chunk of those get frustrated enough
to actually spend less time once they switch to keyword.

Now that they're all keyword searching, the keyword-only crowd has
twice as many chances to hit a product as the others, and a site that
is designed to make that searching more intuitive to use, easier to
adjust, and with more keywords to hit on.

I bet the search-only crowd has more success, in less time, with less
frustration.

If categories are that useful, why did all the major search engines
drop them years ago? I think it's because they simply aren't
helpful with real-world, hard-to-qualify stuff, and the manpower
required to categorize wasn't worth it.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

8 Feb 2009 - 2:25am
Den Serras
2009

Oh, and if anyone has any SEO suggestions...

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

8 Feb 2009 - 3:28am
mcaskey
2008

Interesting! But still sort of unclear in my mind...

So it's not necesarily going to be free of entry point design, such as
suggesting paths, and making-known some of the general groups of things
available, but it will be highly keyword-search-focused?

Almost sounds like bash with a cheat sheet.

Will the site adapt the paths highlighted or keywords suggested, based
on historical user facts at individual or group levels?

</rambling questions>

;)

Dennis Serras wrote:
> Oh, and if anyone has any SEO suggestions...
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=38306
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>
>

8 Feb 2009 - 3:36am
mcaskey
2008

Oh, and as for SEO, if you take a look at Google Analytics, you can get
a report on search terms used on your site, and pages that became
involved there, and what was linked to where visitors went to and what
they did from there.

I don't know everything about seo, but I do know that it does help to
have indexed pages within your site linking to each other in love
triangles. This is a proven tactic. The folks over at StomperNet lay it
out pretty nicely in terms of SEO friendliness.

I suppose you could kill two birds with one stone by adding some
cross-selling, up-selling, group-selling, flock-selling features to the
product pages. Will you feature anything more than pure product
content? Advertorials maybe?

So, will you ever have pages of several items that are anything more
than search-generated?

Wow, I am really interested in hearing more about this! :)

Mike

Dennis Serras wrote:
> Oh, and if anyone has any SEO suggestions...
>
>
> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
> Posted from the new ixda.org
> http://www.ixda.org/discuss?post=38306
>
>
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>
>

8 Feb 2009 - 3:41am
Scott McDaniel
2007

Google and similar situations will use meta-data (if available) in the
code, how instances
and entries are linked throughout the internets, how the content is
stored, and a whole host of
associations with the company, their industry and so on.

In short, even if you don't create the method of organization from a
SEO point of view,
the search engines and other methods of discovery will, in a way,
create them for and in spite
of how your company chooses to implement the site.

When it comes to meta-data, it becomes a matter of association and
working knowledge of
how to fit categories together - different terminology between
different vendors can be bridged
both explicitly in the site's code and through a consideration of how
you conceptually linked
and organized: if an item fits categories A through M, give that item
or group of items the
attributes "A-M", inclusive of each company's categorization and
simply what the item is and
why users would want to find (and hopefully purchase) that item from your site.

Scott

On Sat, Feb 7, 2009 at 9:03 PM, Dennis Serras <dennitzio at yahoo.com> wrote:
> I'm creating a very large e-commerce site that's expected to
> eventually have tens of thousands, perhaps more, entries. It is for
> an industry that is heavily categorized, but every supplier has
> totally different categories, so they effectively become meaningless.
> So we're going to put it all together without categories, or at least
> without a category tree model. It's search engine only, baby. All
> those former categories are now keywords, effectively allowing one
> item to be in dozens of categories.
>
> I'm worried about how sites like Google will index ours. Even if
> Google is a tree-free structure, they seem to depend on trees to find
> their way around a site. Without any hard links, are the spiders going
> to say "oops, there's only one page to this site" and move on? I'm
> assuming they're not smart enough to do searches within the site's
> engine. Do we create a few hundred likely searches and submit that to
> Google as a map? Do we develop some kind of bridge like eBay does so a
> Google search turns up keyword items? The client ain't rich and I
> ain't an SEO expert... yet... So any help is appreciated.
>
> Thanks!
> Den
> ________________________________________________________________
> Welcome to the Interaction Design Association (IxDA)!
> To post to this list ....... discuss at ixda.org
> Unsubscribe ................ http://www.ixda.org/unsubscribe
> List Guidelines ............ http://www.ixda.org/guidelines
> List Help .................. http://www.ixda.org/help
>

--
"In art and dream may you proceed with abandon. In life may you
proceed with balance and stealth." -Patti Smith

8 Feb 2009 - 4:21am
Den Serras
2009

Thanks for the advice (and the reinforcement), Mike! I'm really
pushing this client to try some innovative things on a tight budget.
They have a great idea for a site and I think they need a UI to
match.

There will be editorial and reference materials, and we're planning
on selling banner ads, but I don't know if that'll be enough for
Google. There are "similar items", but for now they're all preset
by the manufacturer.

I guess creating something that links the manufacturer pages together
might do it. The client doesn't want anything visible to the user -
they don't want a competitor getting a full view at the manufacturer
list - but maybe I can make that page user-invisible. That's the
closest thing to categories we have in the site. Maybe a page that
has every search keyword on it, whether a cloud or just a list?

I wish we had the money to do more with automated analysis and
adaptation so the site makes better suggestions to the user. For now,
it'll be pretty dumb, but if it does well we can start getting all
sexy with ideas like yours.

I've been designing web sites since the '90s, and studying UID for
a while, but this is my first big acorn-to-tree UI site. I'm plowing
a lot of my beefs with other sites and search engines into this, and
probably going to lose money on it because the programming will cost
more than my fee. But if I'm right, and all this stuff I'm trying
works, it's a game changer... At least for me. Maybe it'll get me
out of "graphic" design. I'm a movie director by training (with an
engineer dad) and UX is, to me, a chance to bring my two loves
together.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

7 Feb 2009 - 9:31pm
Bradley Pollard
2009

How about charting the re-occurrence of certain words/phrases (in the
title and description of products) and use these to tag your
products. And then offer either a tag-cloud to users/search engines
OR a single level category system made up of your 10 top tags.

Brad
http://twitter.com/bradpollard
http://netlife.com.au

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

8 Feb 2009 - 8:10am
bminihan
2007

To ensure every one of your product pages are indexed, you'll want to
create a sitemap.xml page for both Google and Yahoo. Basically, it's
an XML version of a text site map, created by manually crawling your
own site. You might need to create that text index page yourself
first (you could do so by setting up your search engine, then have it
return every item in the db).

The sitemap.xml format is very tricky and a pain to put together. We
use a free app called gsitecrawler (google that and you can grab it).

If you have a thorough, current site map stored on your server, and
your pages each have good keyword meta data, you should theoretically
get the same SEO benefit as if you had a category index page.

Regarding the searchers vs browsers question: The one thing a
browsable taxonomy does, that a search engine cannot do, is tell the
first-time visitor which products you have. If people don't know
what you sell, they won't waste time running dozens of searches to
find out. Probably better to at least have your own broad categories
to at least hint at the breadth of your inventory (manufacturers,
cost, indoor/outdoor, etc).

Good luck =]

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

8 Feb 2009 - 8:40am
bminihan
2007

Also, an example of a site with products in every conceivable category
(some even uncategorizable), see http://www.ebay.com

Note they are heavily search-based as well, but still provide
categories into which I'm betting 95% of their products fall neatly
and predictably. The rest are certainly attainable via search.

I think you're right that the more diverse your products, the less
meaningful your category list will be. However, I don't think
that's enough justification to leave them out entirely.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=38306

Syndicate content Get the feed