In his EContent column of July 2005, Bob Doyle wrote about A-Z indexes (“Your Site–from A to Z”). He suggests using such an index as a less expensive, pragmatic alternative to taxonomies and thesauri. Building and implementing classification systems takes a lot of resources while the payback for the investment is unclear. A well-done index, on the other hand, is a modest investment with clear findability benefits.
Of course, I full-heartily agree with Bob. A good, structured, well organised index can be excellent help in finding your way around a website. But as always in content management, the problem isn’t just in the visitor-facing front-end of the site. There’s also a matter of maintainability: the CM conflict of interest of having to please both visitors and providing ease-of-use to the producers of the site.
Suppose you had someone do an excellent index of your site. It’s up there, shiny and new, leading the way through previously uncharted content. And then… the content is re-written. The structure is subtly remodelled. Sites have a nasty habit of constantly changing. As a matter of fact, any good website is in more or less constant flux. Where does this leave your A-Z index?
Often, the index won’t hold up too long. New content isn’t indexed. Old content vanishes or is shuffled around and the index points the wrong way, leaving visitors confused or worse, with a 404. The index starts out as a one-off investment with a clear return, and ends up being both widowed and orphaned. Visitors aren’t forgiving of slight mistakes in indexes – they’ll never use or trust it again.
This could be solved by either employing an indexer, automating the indexing process or just being very self-disciplined. But since A-Z indexes tend to be the quick fix for failing navigation they often turn out to be static band-aids. Too little consideration is paid to the fact the index, too, has to be constantly updated and optimised.
In my opinion A-Z indexes seldom provide a real advantage to a website. Not because they’re not a good tool per se – but because they’re used for the wrong reasons and the effort they take is always underestimated. The time and money spent might more effectively be invested in clear navigation. After all, wouldn’t you rather follow an intuitive path to the information you need than browse through the alphabet, hoping a term representing the answer to your question will come up before Z?
This post was originally published on the blog of Content Management Professionals Benelux. I have added the comments received on the original post as one, below.
COMMENT BY BOB DOYLE, 11 AUGUST 06
Adriaan,
Interesting criticisms. Some observations.
The Site Index is never principal navigation. That should be the standard global/banner and then local/left contextual navigation.
Those are primary and powerful because they are a de facto taxonomy, organizing the content into site sections and sub-sections. They reflect the fundamental information architecture of the site
The greatest benefit of an A-Z index is that we are all familiar with indexes from the book experience. We have learned to scan an index and thereby quickly develop an overview of a book’s content.
I believe we can do the same with a website index.
We get a quick understanding of what the site contains – better than we can from reading simple short navigation labels or examining a site map (which usually just present the site navigation on one page). Don’t you agree?
But you raise the excellent point that where a book is static and its index is fixed, with a website the content is variable so the index must be maintained.
Indexes are not easy to maintain, but then neither is good content easy to manage. Even CM Pros are heard wildly dreaming that a CMS should be so easy to use that absolutely no training or learning of any kind is required.
So my advice on the site index.
1) Do not even try to make it comprehensive and correpond one-to-one with every page. Just make it sure it is representative of the content. Include important pages that may have their contents updated, but not changed fundamentally. Include every landing page for important sub-sections, unlikely to ever produce a 404. Etc.
2) Tag all pages that are indexed so when they are changed the content owner is notified that the index entry needs to be reviewed.
3) Read the excellent advice of Heather Hedden and her Web Indexing SIG.
http://www.web-indexing.org/ Get a professional to craft your site index so it is designed for minimum maintenance.
4) Have one person in charge of the index doing her/his best to maintain it if/when the site evolves into new content areas. Heather did this herself for the first year of the CM Pros index, now Jack Shasha does it.
The principal navigation reflects the information architecture, as does the site map. The site index does not. It samples the contents to provide a very easily scanned overview of the website substance.
To sum up:
Navigation, IA, Site Map show us formal structure.
Site Index gives us a snapshot of the content itself.
Hope this helps,
Bob.
COMMENT BY ADRIAAN M. BLOEM, 15 AUGUST 06
“Indexes are not easy to maintain, but then neither is good content easy to manage.”—that, indeed, is my main point here: the effort a good index takes is often underestimated and it’s often thought to be something that might fit in between lunch and dinner.
If one isn’t prepared to commit to points 1 through 4 you describe, and/or the index is a patch for failing navigation, then rethink the strategy. A good index is a great tool, but its creation is not a trivial effort and without real commitment will only lead to alphabet soup.
Your advice should be mandatory reading for anyone contemplating adding an index just for the sake of having an index!
COMMENT BY HEATHER HEDDEN, 15 AUGUST 06
Bob Doyle is correct in stating that an A-Z index is not a substitute for navigation. It’s really a form of search, that is human-indexer crafted, rather than automatic.
Maintaining an A-Z is definitely an issue, and for this reason A-Z indexes make more sense on sites, sub-sites, or sections of sites that are relatively static. This is often the case of portions of academic institution sites.
One tool for creating site indexes, HTML Indexer, has a feature of embedding some code in the indexed web pages that when a page is deleted, the linked index entries are automatically removed from the index. So, there is some automatic updating, but of course, when new pages are added to the site they must be indexed, too.
While you might want to contract a professional indexer to create your A-Z index, you ultimately want someone within your organization sufficiently trained to feel comfortable in maintaining the index. I offer online training in website indexing.
It also so happens that I am planning to travel to the Netherlands in July or August of 2007. I will be giving a short in-person workshop on website indexing to the Netherlands Indexing Network http://www.indexers.nl.
COMMENT BY YAIR DEMBINSKY, 15 AUGUST 06
In my view, A-Z indexes lack the most important feature of all CM solutions – context.
If I know for sure what phrase I am looking for, I’d use simple search solutions rather than organizing the content in a web site or a CM system.
The advantage of CM solutions over search solutions is the ability to associatively navigate between concepts, and to find what I am looking for even if I don’t know how it is named in the web site.
COMMENT BY BRENDAN, 15 AUGUST 06
One of our Information Architects wrote a great article about our site A-Z index (the ABCs of the BBC: a Case Study and Checklist), discussing synonyms, numbers, “the” and “a”, and comprehensiveness. It was quite well received.
But of course, a good A-Z is no substitute for a good navigation structure and site search!
COMMENT BY BOB DOYLE, 15 AUGUST 06
Í disagree about indexes lacking context.
What a good index does is show you the contents of a book. Especially it shows you at a glance what the author calls everything.
This is a terrific supplement to search, because it gives you some exact terms you might search for – and find related content that is not in the Site Index (which is impossible to keep comprehensive and inclusive of every page and paragraph).
For example, if the index contains “workflow” but not “business process” you can make an educated guess about related jargon. Similarly if it contains “tags” but not “categories.” Does that make sense?
COMMENT BY SETH EARLEY, 15 AUGUST 06
It’s interesting that you suggest A-Z indexes lack context. One could argue that they have more context, than say, a taxonomy.
An index needs to take a perspective, an indexer is trying to judge what content will be relevant to a user and then put a label on a pointer to that content.
A content management system can use a number of methods to expose content to a user in the context of their task. One is a navigational hierarchy or navigational taxonomy. Another leverages metadata and controlled vocabularies to allow for faceted navigation and of course there are various flavors of search.
A set of search results is in essence a machine generated index. That in my opinion typically lacks context. A human generated index, on the other hand is based on the judgment and perspective of the indexer (not of the author).
The indexer is trying to understand the needs of a reader/user and deciding what they would find important. We all know there are bad indexes just like there are bad user interfaces. Here is an example from my car user manual:
http://sethearley.wordpress.com/2006/06/08/a-very-bad-index
I think an a-z index is an excellent addition to a web site or intranet. It allows people to find things at a glance and learn what information is contained in the system. The terms are ‘pre coordinated’ that is, they are terms that are put together into an idea, concept or thing that has a unique meaning, as opposed to consisting of individual terms that can be assembled.
For example, “consulting best practices” is pre coordinated. “Consulting”
and “best practices” are post coordinated – the terms are assembled after the fact in order to find a piece of information.
On draw back of an index is that it is typically “hand crafted” and more difficult to maintain than, say, a taxonomy. As content changes, the index changes. A taxonomy is easier to maintain, expand and reuse since it is independent of content (not an actual access structure), but a taxonomy lacks the precision of an index.
There is a middle ground of Topic Map. We are currently doing a research project on topic maps. They hold the promise of greater precision than a taxonomy, the ability to understand semantic context, yet less difficult to maintain than an index.
For those of you interested in our research project (there is a fee for participation), you can check http://sethearley.wordpress.com/2006/07/14/syndicated-research-projects
(This is a brief description – we also have a full prospectus on the
project.)
Here are some other points about indexes versus taxonomies
Index:
Content access structure
Unidimensional
Pre coordinate – for example “taxonomy best practices”
Source specific
High levels of granularity – What is this concept about?
Dependent on access structure
Significance and relevance determined by indexer Difficult to repurpose Short term view
Taxonomy:
Document access structure
Multi dimensional
Post coordinate – for example “taxonomy”, “best practices”
Source independent
Lower levels of granularity – What is this document about?
Independent of access structure
Neither significant nor relevant by itself May be repurposed Long term view
(This comparison came from one of our taxonomy conference calls with Fred Leise and Heather Hedden – archive info available at:
http://www.earley.com/CoP_Archives/Indexing_and_Taxonomies.htm )
COMMENT BY BOB BOIKO, 15 AUGUST 06
I’d like to second Seth’s comments. While both a FT search and a “keyword”
search depend on indexes, the FT search is “dumber” than the keyword search because it is simply an alpha list of terms. A good keyword index on the other hand has significant extra smarts in it.
To my mind the biggest reason to employ a keyword (or phrase) index is to map the words or concepts that a user is likely to know to those that the system prefers. So, the most important function of a keyword index may be to list words and phrases that are not in the text.
COMMENT BY HEATHER HEDDEN, 15 AUGUST 06
An A-Z index also has a lot more context than a blank search box.
Many A-Z indexes you might find out on the Web are created by people not trained in indexing, so they have their deficiencies. The usefulness of an index depends a lot on its quality.
COMMENT BY PAULA THORNTON, 29 AUGUST 06
Let’s try a very specific example where the payback is tremendous from multiple dimensions: intranet. By leveraging the data in a corporate process repository (where the goal is to align a ‘process step’ to a ‘task’), we’re going to create a task-based directory.
In this way we have the leverage to influence the naming of the process steps (by actually having them tied to daily activities) and by implementing a semantic layer to the steps, they can appear in multiple alpha sorts (see visual sample at this A-Z index). They’re automatically updated when new tasks are defined to corporate processes.
We’re leveraging this as a mechanism that most closely matches intranet ‘intents’—people having a ‘task’ term in mind.