On-Site Search Expectations vs Reality
On-site search, in my mind, draws out an awkward inner vision of wet sand passing through a narrow hourglass. It’s often like a monkey sitting on the web development team’s shoulder that nobody in their right mind wants to join in conversation but, for a number of reasons, exists to remind us of our limitations.
The Crux of the Search Problem
Let’s discuss it honestly – web professional to web professional – there is a vast chasm of experience between web search and web search. We are not Google and Google are not us… never the twain shall meet. That’s often why we fudge it and just use Google on public facing search. But what about the intranet?
Jared Spool explained the crux of the problem when he said the reason on-site search isn’t as good as Google is because it can’t ever replicate Google. For one thing Google indexes billions of pages to come up with page rank. Google looks at blog posts and other web pages that hyperlink to any indexed web page providing invaluable context to its value.
He said nobody links to web pages on an intranet. Nobody writes blog posts about web pages on an intranet. Thus the on-site search function is limited by the boundaries of its own existence.
Advanced Search
At the same time there is a fallacy that Advanced Search can dig us out of the hole we’ve provided for the user. And it’s a fallacy. It’s a fallacy based on the premise the rest of the world are computer librarians comfortable with Boolean search functions. Advanced Search is the equivalent of throwing your kid the car keys to go and get their own ice-cream.
An example Jared used was the layman’s intention with a Boolean search for Cats AND Dogs. They want stuff about cats AND they want the stuff about dogs. However, Boolean Cats AND Dogs just gives them the stuff about both cats and dogs… so where does that leave them in the Advanced Search stakes?
The Solution is in Better Navigation
The answer isn’t to make search better… the answer is to make navigation better. Because ultimately the search function is our patchwork attempt to overlook the weakness in navigation that led the website visitor to the search function in the first place.
What do people do when they go to a website – look for product… can’t find product… assess the navigation options… resort to search. So search is our development team throwing our hands in the air and saying to the visitor that it’s THEIR problem. OK smartass you go find what you’re looking for… instead of rethinking our navigation scheme to make products easily findable.
At that point where they hit your search function their expectations are probably high. Their experience, on the other hand, is probably going to be low.
July 11th, 2010 at 1:03 am
With regard to Advanced Search, surely the key is to present the options so that no Boolean knowledge is required. Ms User doesn’t need to know she’s choosing “OR” when she’s choosing “stuff about cats AND stuff about dogs”.
What site search is really good for is when someone comes to a site following a link from, oooh let’s say a Google search results page, but the page linked to isn’t there and Prof User is taken to the home page or an error page. Site search is right handy then.
Google site search is great but not everyone knows about it. Personally, since Firefox 4.0 Beta took away my Google toolbar, Google Site Search is dead to me.
One of the things I like about the web is that there are often several ways to do something, meeting slightly different needs, expectations and circumstances.
Give me great site navigation by all means, and let me keep Google Site Search in my pocket, but sometimes a nice site search box is just right.
July 11th, 2010 at 9:34 am
Ricky, the Advanced Search Boolean example was only an example… it applies pretty much across the board. We as computer people might use advanced search (and librarians of course) but Spool has found zero evidence that normal users partake of it (most of my friends struggle with email, to be honest). One user comment was “if I couldn’t get ordinary search to work how could I get Advanced Search to work”? So with the small interest and usefulness of Advanced Search then it sucks a disproportionate amount of time for a mediocre result… if that makes sense. And I’d have to agree. I’d ditch Advanced Search simply on bang for your buck unless someone could come in the door with a cost-benefit analysis that showed me this was going to add to my business objectives – ROI. How many more toasters will I sell if I invest that significant money?
I’m not sure he was saying don’t have site search at all… and I’m not taking it off this site. But I have to warn anyone looking for articles here – even if you know the title – they are pretty hit and miss. Which is why I just spit out the whole lot in a huge unpaginated list… it’s for me to find my cross-references.
I think he was saying that once you accept on-site search is not Google, that it’s a very ordinary hit and miss experience that users fall back to… it’s not the first thing they jump to on your site… then search is less the beacon we imagine as developers and probably more the crutch we use to slip away from making our stuff actually findable.
I’m also not sure about the attrition rate as users go from looking for product onto the search path… given page loads… waiting… human frustration levels and expectations of search set by Google… I’d say very few people would be using the search function. And only those with a mission (unfortunately the other point is that their on-site search experience will be mediocre at best).
What he is saying in the audio is for developers not to only half do their navigation and say people will use search… because mostly they won’t… and if they do use search, mostly they will get a crappy experience because we can’t provide the level of context Google can provide.
The simple answer then is to put more effort into making the toaster model… or whatever the product… more findable through better navigation.
But you raise a strong point. A big part of that is content maintenance – logically it’s important to delete old pages / content from the system which becomes outdated or irrelevant (posing more obstacles to people if they were left there)…
… so we also need to figure out a system to handle this beyond an uninformative 404 or home page redirect. That strategy in itself appears to be another fudge.