People know there is a problem, but they don’t think they can be part of the solution. People blame the users for using the wrong keywords or using the wrong box.
Home Depot increased conversion rate over 30% in the first two weeks. Increased average order size as well. This is important for getting buy in.
Search is a learning process. For searches of any importance search is not as simple as entering keywords, hitting go and you’re done. You enter keywords, find results, and find you’re not searching for the right thing. Read “Berrypicking: Evolving Search.”
It’s an iterative process that changes along the way. What you end up with is not what you were seeking, but it is the right thing.
You cannot just optimize for usability, but you must also improve learnability.
Search is not just about software. Search has patterns of emergence and self organization. What are our users information seeking behaviour? Do they value precision? How do they use the interface. Search results are part of the interface.
Remove rot. The simplest way to increase value of information retrieval is to make the results smaller. Get rid of stuff that should be removed. Use structured data and controlled vocabularies to help our users shrink the search space.
Empower information providers to provide meta data. It’s not just an IT problem but a knowledge management problem.
Focus on the interface.
Use incremental construction, and progressive disclosure. Support immediate response. Feedback is essential to flow. Help users predict what happens when they move a slider. Design for direct manipulation. Make sure you are designing or the unique context of use. If mobile make sure it’s designed for big fat thumbs.
Different contexts
Realtime people want the absolute latest. We want to make them aware of new stuff. In Mobile we nudge people towards browsing because of the fat thumbs on tiny keyboards. Entering queries is hard. Plus people don’t know what they can search for so we use navigation icons to prompt them. Same is true for kiosks. You have strangers making noises and looking over your shoulder. Interactive TV have been browse oriented, but as we icnrease media we have to increase our reliance on search.
Search Patterns library in Twitter. — This looks really cool.
Behavior Patterns
How do we make the expert strategy (pearl growing) available to novice users? Google uses Find Similiar
Design Patterns
Repeatable solutions to common problems. Autocomplete is very common. Practically ubiquitous on the web. Saves time. Based upon helping the user as son as we have a couple characters. Yahoo has pushed this further with “Auto Suggest” which recommends suggested related searches. Uses information like post query browsing behavior. This helps people get unstuck.
Best first is the single most important pattern. This is why Google has been so successful. Find the best results as quickly as possible. We know that the vast majority of attention goes to the first few results. After the first 3 results attention falls off a cliff. Find the best and most relevant results. Google has always made a big deal about how fast they are, it takes .12 seconds. The seed enables the iterative interactive experience. It doesn’t cost very much to search. Push for sub second response times. Don’t take slow for an answer.
Flickr has an interestingness algorithm which takes into account many factors.
Federated search is a symptom of the problem it is solving. The reason we need federated search is because users don’t know where to look. The Ann Arbor district library knows to look at the catalog for books and other media. But hidden behind the research tab is hundreds of valuable databases that can help them with problems. Libraries are struggling to help users become aware of the stuff that is available to them.
Library of congress is trying to make a more integrated and aggregated search. Why are the sources seperate and in different formats Federated search has inherent problems in that you can only search the lowest common denominator o meta data. You can search everything but it’s terribly slow.
Search crumbs/facets help provide a map. It helps us see a simple next step on how to narrow and refine their query. If this is done in an iterative process it helps people construct what would have been a sophisticated boolean query. Sometimes you’ll see the facets on the top. This puts constraints on the number you can have.
NCSU has a product called summon which allows them to integrate their library into their magazines and data collections. Lands End uses a similar system. Search done well can also provide assistance for browsing. You can browse to a category but still see the faceted search display. At buzzillions you have user participation in the model. Tags and tag fields are used as facets.
Google is also using a somewhat faceted model. It is difficult to have structured metadata for the whole web. However they have experimented with the ability to narrow or refine their search. They ahve realized that a simple interface for the box and the results isn’t enough. They ned to allow a conversational search experience.
Redefining Search
There is a huge space for innovation in search, What about people being able to find their keys with their iPhone? How can we reframe search and discovery in a way that helps us breakthrough existing barriers?
One way is to use question answering, like Wolfram Alpha. We shouldn’t provide results but answers!
Maybe it’s about decision making, like hunch is working on. You start with a question and they make you answer more detailed questions to help improve decision making.
Maybe it’s about pattern recognitiona nd helping people find the unkown unknowns. Infomation architecture is geared towards directed search, but there is a lot of work that is oriented towards helping people find information they don’t know they need. Search technology really helps with pattern recognition.
We’re always amazed when a new breakthrough comes through, but then in a couple months we’re bored. Our sense of wonder has a short shelf life. What we think of as the searchable web keeps expanding. We keep increasing what is searchable.
It’s also interesting to see how we search. iPhone lets us look for stuff that looks like, or sounds like. We can sketch to find stuff. How long until we can feel like or taste or or smells like something else?
Cross Media Integration
Right now we don’t even have the right language to describe what we’re trying to do. We’re not just talking about multichannel marketing. Nor are we talking about multi channel selling. We’re talking about cross media integration. One platform or chanel acts as a controller to other platforms or devices. We then have to figure out which features belong on which platform or device.
Do we go for consistency? Having the same featureset the same with specialization? Nike+ embeds a sensor in the shoes to track the various running metrics, like speed and distance then share that inforamtion. We’re moving from conspicuous consumption to conspicuous experience.
Redbox is also a weird product service combination. Nothing you could have predicted 10 or 15 years ago. You can search for a movie at the office, then find the kiosk it’s in and get it on the way home from work. If you’re out and about you can find the nearby box on your iPhone.
Ubiquitous Computing
Tagging of unique objects with unique idetnifiers enables you to find their meta data. There is also research going back to 1992 about creating a customer journey and their touchpoints. There are activities which happen in front of the line of visibility (what your company has insights into) and stuff which happens after the line of invisibility.
This gives us a service centered view, how do we created user centered views/ How do we create maps which help our friends and colleagues understand the comple world our customers are interacting with today.
Search is a key application but it can get oppresive. You may get sick of searching for a good movie and just read a book.
Why He Likes Search
It’s so damn hard. It’s a hybrid of design, engineering and marketing. There is a lot of ambiguity and interdependencies. Stakeholders with different world views. It’s both a project and a process. To get it to a starting point you have to do a full redesign sometimes. It could be a really big project to even get to a good starting point.
{ 0 comments }