Findability is a constant theme in content strategy and technical communications, yet it seems to me that people often treat findability as a problem existing outside the content. Findability is addressed using SEO tactics and by devising sophisticated top-down navigational aids, such as taxonomies and faceted navigation, but it is seldom seen as issue to be addressed in the content itself.
I believe this focus on top-down findability is wrong. Top-down finding aids have their place, but the majority of the focus should be bottom up, and it should start with the content itself.
To begin with what should be the most obvious point, but which seems to be often forgotten: readers are not looking for content, they are looking for knowledge. If I go shopping on Amazon, I am not there to find a page describing a book, I am there to get the book. If I find the page, but then cannot get the book, my quest has failed and I leave disappointed. If I go to a used car site like autocatch.com, I am not looking for a dealer listing of a car, I am looking for a car. If I find the page, but the information on the car is inaccurate or incomplete, I may not recognize it as the car I am looking for, and thus I won’t find the car, and will leave disappointed.
Similarly, readers do not search for content because they want pages of text (or graphics, or video, or any other kind of content). They search because they want to know something and if they don’t leave knowing what they want to know, they leave disappointed. If you conceive of your findability strategy as being about getting readers to topics or pages, you are missing the point. The reader may well find the page and still not find the information. They may find the information and still not get the knowledge. Finding is not complete until the reader acquires the knowledge they seek.
Trying to solve findability from the top down runs into a number of difficulties, some of which I have talked about before:
- Web Organization is not Like Book Organization
- Readers Don’t Classify their Experience
- Sometimes, Readers Do Classify their Experience
Jakob Nielsen finds that people are increasingly search-dominant in their information seeking behavior, even though they are not very good at searching. Advanced search and navigation features generally don’t help:
In general, we almost never see people use advanced search. And when they do, they typically use it incorrectly — partly because they use it so rarely that they never really learn how it works.
The combination of reliance on search and poor search skills can only mean one thing: people are landing deep inside your content, and often in the wrong place. Expecting them to fix this by going back to your search and browse mechanisms (expecting them, for instance, to use the TOC of your online help system) are doomed to failure, because people are both search dominant and bad a searching.
The real findability problem, then, is how to get readers from the wrong place deep within your content to the right place deep within your content.
Information foraging theory, tells us that information seeking behavior is often like the foraging behavior of wild animals. Readers do not follow a logical path, but sniff around for the scent of good information. Whether the reader will stay on the page they landed on or quickly abandon it, therefore, will depend on the information scent of the page they have landed on, and the pages it can lead them to.
Pages that have a strong information scent make the information they contain more findable. The reader may easily be directed to a page by search or by a TOC or faceted navigation system, but not detect in the smell of the information they need. They may give up on the page before they get as far as the information they are looking for.
How do you give your pages a strong information scent? One of the key characteristics of an Every Page is Page One topic is that it establishes its context. By establishing its context quickly, a topic lets the reader know what they have landed on. You can’t assume that the reader is already correctly contextualized by the finding tools they used to reach the page. People just aren’t that systematic or skillful in their research methods. They get to the page and they sniff — does this place smell like information?
If the page smells good at first sniff, the reader continues. The smell is a good start, but they are still looking for the real nutrition. Finding isn’t over yet. Another key characteristic is that EPPO topics conform to type. Why is this important? If it is context that gives the content its smell, it is type that makes sure the content actually contains all the nutrients the smell promises. If the topic does not conform to its type, if it does not contain all the information that a topic of this type is supposed to contain, it is at best a Twinkie — empty calories with no nutritional value, and at worst a perfumed turd. If an appetizing smell is followed by a sour taste, the foraging reader will no longer trust this information patch and will wander off. (By conforms to type, I don’t mean concept/task/reference, which are far too general for this purpose. See The Tyranny of the Terrible Troika.)
What if, as is not unlikely, the reader has not landed on the page that contains the information they need, but on a related page. If the topic they have landed on establishes its context clearly, then they should recognize pretty quickly that they are in the wrong place. But chances are the right place is nearby — nearby in the sense that it is on a subject closely related to the current topic — and that it has a strong subject affinity with the current page. If the page fully states its context, and links along the lines of subject affinity within its context, there is a good chance that one of those links will take the reader to, or closer to, the information they need. And the very fact that there are links tells the foraging reader that it will require minimal energy to continue foraging this patch for a while.
This approach will certainly be more fruitful than searching the site. A recent study by Jared Spool found:
Using an on-site search engine actually reduced the chances of success, and the difference was significant. Overall, users found the correct answer in 42% of the tests. When they used an on-site search engine (we did not study Internet search engines), their success rate was only 30%. In tasks where they used only links, however, users succeeded 53% of the time.
Providing links along the lines of subject affinity from the current topic is much more likely to lead the reader to the correct topic than leaving them to search. As I have noted before, links are the last mile of findability. Robust and accurate linking can allow the reader to follow their nose to the content that contains the information they need to acquire the knowledge they seek.
Once the reader has found the right page, and been able to recognize and digest its information, we are making real progress. The reader finds the page nutritious. That is good, but it is not necessarily the end of findability. The reader may have eaten, but they may still be hungry. Because findability does not end until the reader acquires the knowledge they seek, it does not necessarily end with a single page. If you think of your findability strategy as a creating a single downward vector ending on a single page, you are missing the fact that a single page is often not enough to give the reader all the knowledge they seek. Findability is not complete until the reader has found and understood all the pages they need to acquire the knowledge they seek.
If the reader’s meal is incomplete, how do they find the additional nutrition they need. Sending them back to top down navigation is clearly not a good strategy. But chances are that the additional information they are seeking is related to the topic they have just read. The topics that contain that information have subject affinities with the topic they have just read. If those subject affinities are linked, the reader can follow an easy scent trail to the rest of the information they need.
Findability, then, begins at the bottom up, with the content. It starts with content that fully serves the purpose for which it is intended, which is best achieved by defining specific types for the different types of content you create (based on actual studies of user information needs) and making sure all topics fully adhere to those types.
Secondly it requires that all topics fully establish their context so that the reader knows where they have landed, and that they make that context fully navigable along all significant lines of subject affinity so that the reader can move to the correct page if they are not there yet, or find additional information if they need it.
Thirdly, it requires rich and systematic linking along all lines of subject affinity throughout the topic so that readers can keep foraging your content set until their need for knowledge is satisfied. A systematic approach to linking along lines of subject affinity is critical here. As Sean Carmichael notes, summing up Jared Spool’s The Secret Lives of Links:
Websites are full of links. How useful these links are in helping users complete tasks is another story. Links have to guide users as they follow the scent of information. A vague or confusing link often leads users down a wrong path and in turn increases their rate of failure.
Links should help the user follow the scent of information. In other words, they should follow the lines of subject affinity between topics.
These things — topics written to a defined purpose, conforming to a defined type, establishing context, and linking richly along lines of subject affinity — are the bedrock of findability. SEO is a garnish on top of this, but great content appropriately linked is the foundation of good SEO anyway. Site level search and browsing tools may play some part, but research shows they don’t work all that well, and if the content they lead to is not good, and does not let the reader complete their finding, they accomplish nothing.
Findability begins at the bottom, with the content. It can’t be added on afterwards or from above. Findability is job one from day one for everyone.
Yes! And the key issue is to know what purpose and type a page must conform to. We can create great EPPO pages that deal with knowledge nobody is searching for. Findability begins in my world by knowing what information-seeking goals users have. That is what my research is all about.
But you make me a little confused here. We have previously talked about “include it all, filter it afterwards” and that a key to findability is user friendly filters, which can be social media filters etc. But classification filters such as in faceted search (which is different from faceted navigation) can play a role to help the user filter a list of pages. Users search for pages including a key word and are then provided with filters to narrow down the result list. Do you call faceted search bottom-up or top-down?
I see classification using taxonomies as a way to provide the context of a page. This way, we can add labels to the content that actually do not appear in the content per se. I have elaborated on this here: http://www.excosoft.se/index.php/about-us/blog/item/13-how-is-a-topic-signaling-its-content?
Thanks for the comment, Jonatan.
The problem with classification filters in an “include it all, filter it afterward” world are twofold.
* First, they don’t include it all. They include only a tiny fraction. So they are not attractive to use who wants to include it all before they filter. I’ve described the Web as a big information filter, and I think that is the key idea here: it is the Web that filters, not the content owner. There are exceptions, of course. If you can corner the market, the way Amazon has, the user can include it all by searching within a single site. Or if you can provide a multifaceted selection mechanism that the Web alone can’t provide well, and that already matches the way the user things about the problem, like a use car site such as autocatch.com, you give users a reason to come to you. But most sites don’t meet those criteria.
* Classification filters only work where the user is already thoroughly familiar with the classification scheme. That is seldom true for tech comm, or for many other things. People don’t classify their experiences, they name or describe them.
But I agree with you about using classification to give context, as long as it is done on the page, and is specific to the item on the page. (A TOC pane does nothing to establish context). In fact, I see subject-affinity links as faceted navigation from the bottom up. This approach has three great virtues:
* Top down classification looks more like a wall than a door. It is overwhelming and confusing. But locating a topic in a classification scheme from the bottom up does not overwhelm, because the user is already located in the right place in the classification, and is looking only at the immediate locale in the classification scheme.
* Bottom up you have the opportunity to begin to teach the reader about the classification scheme, without making it a precondition of access to the information.
* Bottom-up supports positioning in as many classifications as are relevant, including irregular ones that belong only to the instance, not the type.
But people are not contextualized by the classification schema before they arrive at the topic (most won’t get there that way anyway). The topic must contextualize itself.
Readers do not follow a logical path, but sniff around for the scent of good information.
Now that picture made my day 🙂
Mark,
Thanks for this reminder that SEO isn’t the findability grail. Jared Spool’s research on the “scent of information,” which you point to here, has helped me over the years. So has his (and Gerry McGovern’s) research into “trigger words,” which help to create, in an odd mix of metaphors, the scent.
The premise is simple: a link’s scent comes not only from the “subject affinity between topics,” as you say, but also from the link’s wording. Synonyms may convey the subject affinity equally (“deal” and “bargain,” say), but if the information seeker has a specific word in mind (“deal” instead of “bargain”), that word–a trigger word–has stronger scent.
Spool says it well in “The Right Trigger Words”: http://www.uie.com/articles/trigger_words
Marcia
Thanks for the comment, Marcia
Indeed, trigger words can be very important. A subject affinity cannot be represented as an abstraction, it has to be presented as specific words, and the word that have the strongest “subject scent” for one person might not be the ones that have the strongest subject scent for the next person. (One of my pet refrains is, “All language is local.”) It is most definitely worth paying attention to what the trigger words are for your main subjects among your principal audience, though that will be easier to do for some subjects, and some audiences, than for others.
This is also one of the reasons that people will search the Web rather than searching individual sites. An individual site may not use their trigger words, which makes it harder for them to pick up the scent of information. The Web lets them search using their trigger words and find pages that use those trigger words. In some cases, those pages they find may teach them new trigger words, which they can then use to better detect the scent of the information they are looking for. Another reason findability does not end the first time the reader hits a page.
Hi Mark,
I agree with much of your post. Great findability requires good content. However I do think you are making the issue too black and white when looking at the search part of findability.
The truth is that a knowledgeable search expert with the right tools can do a lot to improve the findability without touching the content. This is done by working with automatic and rule based classification, linguistics, relevance, facets, query suggestions, result presentation, etc.
In the end, garbage in garbage out still applies of course. But to deliver a end user experience you need to work on both sides and their is a limit on the effort you can put on editors.
Another point is that designing good search is not a top down approach, but is about engaging end users and understanding what information they need and why.
The statistics showing that people do not succeed by using search functions on web sites I think says more about the horrible state of those search functions rather anything about search as a whole.
That said, I work with a search consultancy and we employ several people who are information quality experts and we do what we can to educate customers to increase their content quality. The reason is of course that if the content is good, our search experience will improve significantly and we will have happier customers.
In the end, I think it would be better to get away from seeing this as a conflict and instead work with content, navigation and search in concert to fulfill the goals of the specific site.
Thanks for the comment Mattias.
I agree that search engine tuning can do a great deal to improve the quality of site searches, which is a good thing because, for users who use the site search, it means they will land closer to the information they need. The closer they land, the less energy it will require to get to the food, and therefore more people will persist to the food, and thus the higher their success rate will be.
But of course, that only applies to the people that choose to use the site search. Those who come in via Google or via socially curated links, still land only as close as those tools can take them, and thus they still need good bottom-up findability.
And even those who use the site search as not finished as soon as it lands them on a page. Their finding isn’t finished until they have the knowledge they seek, so again bottom-up findability is still key.
Finding is a multi-stage activity. The first stage is the long jump into the content that search, a TOC, or faceted navigation provides. The second is the fine tuning of location, which good links provide. The third is understanding, which good content provides. The fourth is onward navigation to additional related information, which good links provide. We need to pay attention to all the stage if we want successful finding to occur.
So, yes, my title is a bit of hyperbole in that it suggests that search is not part of the problem at all. The intent of the hyperbole is to draw attention to what is currently a neglected, and vastly important, part of the findability picture, which is bottom-up findability.
Spot on with another very good & useful article. Also with good links – Information Foraging: Why Google Makes People Leave Your Site Faster – that are still relevant today (still have that in my old Delicious tags-never delete anything!).
One question though, do you not think that metadata can also play a role in Findability?
Thanks for the comment, Mike.
I do think that metadata can play a large role in findability. In fact, I think it should play a larger role than it does now, because metadata is what you need to drive bottom-up findability. If you want to provide effective linking along all lines of subject affinity, you need to drive linking with metadata, not rely on humans to create and manage all the links.
Metadata, in the form of subject-specific content templates or schemas can also help ensure that content meets its purpose and is true to its type, which is an essential part of findability.
Metadata can also play a role in top down findability, though I think its capacity to do so is significantly overstated today. It can help search engines find content, and it can drive faceted search in the cases where faceted search works.
One of the biggest problems I see in content creation today, though, is that we are failing to capture and to exploit the metadata needed to drive effective bottom-up findability.
The current approach of adding metadata after the fact strikes me as one of the worst mistakes we make in information architecture. As I wrote a while back, metadata is not an afterthought: /2011/11/29/metadata-is-not-an-afterthought/
Hi Mark,
I agree with you that in many organisations Metadata is still treated as a poor cousin, though that is changing for the better. Whenever I am involved in content project, it’s one of the main discussion points held that I find helps people understand their own content creation in much more detail.
Also in most cases today, nearly all content creation tools/CMS’s do now have Metadata creation tools, if not then they should have. I worked with one that had them 20 years ago!
Metadata has raised its head much more in the eBook selling process as publishers and self-pub authors realise that they need to get ‘their’ products ‘found’. However, there is still a long way to go to get metadata ‘content’ added at the right place in any workflow stage and not treated, as you rightly say, as an afterthought.
Pingback: Findability and SEO – It is about the content… | Simply Minimalist
Insightful article Mark.
In their brilliant book, Designing the Search Experience, Tony Russell-Rose and Tyler Tate, spent entire chapters explaining why search is a really information seeking and it is a continuous behaviour and not just about a single action. I would like to pick out one chapter that is relevant to this discussion: Search Modes. The argument is that people look for specific kinds of knowledge, to use Mark’s term. Some look for trends and comparisons while others look for interpretations and meanings. If we know these search modes then we can design the entire “search experience” that includes top-down navigational aids and bottom-up foraging links. Google’s use of index cards to contextualise a search mode is a good example of such a system in action. See, http://www.fastcodesign.com/1672605/how-google-unified-its-products-with-a-simple-index-card
Thanks for the comment Maish. And thanks for the reference to the book. Downloading to my Kindle now.
Pingback: Обзор свежих материалов, июнь 2013 « Юрий Ветров об интерфейсах