.oO Tailwind

First seen by Barbd, Ex.plode.us allows you to search social spaces for people’s profiles and it draws them together.

You can run a search for Warren Hutchinson and Snowbadger (My alterego) and get different results.

This make it easier for people to ‘dig up’ the real me. Eeek.

I worry about advertising here.

I wonder what this does for the privacy, professional / social persona debate I mentioned in an earlier post.?

Here’s a list of me and my friends… It’s interesting that Snowbadger has more friends than Warren hutchinson.

Hmmm.

10 things to ring my bell this week.

1. See how much of a cliché you are
2. Cool panoramic of Paris by night.
3. New easy to read wine labels from Wine That Loves…
4. Bruce Mau’s ‘Incomplete Manifesto for Change’ - A list to make you ‘do’ differently not just ‘think’ differently (Via Brand Autopsy)
5. Sing n search, you sing it, it will find it (provided you have a decent mic built in that is)
6. Record last.fm and other stations if you have a PC
7. Gmail keyboard shortcuts Improve productivity with Gmail, some secret delete shortcuts in here.

Twitterlinks:

1. Twitthis Tell people via Twitter about a blog or posting.
2. Twitterbuzz See what ‘Twits’ are linking to.
3. Celebrity Twitter Follow the lives of your fav celbs on Twitter. Yaaaawn.

Continue reading ‘Tailwind Lazy Links 2′

CPC has been the mainstay of ad revenue for some time now. Ads are displayed alongside relevant content be-it a search results page or something else and the advertiser pays a fee every time someone clicks their ad.

Simple, but cheatable.

It’s open to mistreatment, largely by an advertisers competitor. Because they can click the ads and cost the competitor advertiser for ghost queries or click fraud as it’s been called.

Pay-Per-Action is a clever evolution of this that should prevent click fraud as it’s not so easy to spoof. It will only cost the advertiser if the user visits their site and they carry out another action. This action needs to be the completion of a key goal such as buying something, registering for something or downloading something as to avoid further click fraud.

It’s nice, it’s simple to understand but will require some complex understanding of what advertisers want users to do. It’ll also help the smaller companies track ROI and may even get them to think about ‘goal based’ design for their own websites.

PPA and CPC will live together for sometime and due to there being a new link style for PPA, the two methods should be decipherable.

But, I spotted some slightly worrying news that Google are introducing a new format for these ads. Check out the FAQs. The new format allows advertisers to display up to 90 characters inline with page editorial so that they don’t look like ads at all, they look like hyperlinks. They will only reveal themselves as ads when rolled over and will be mixed in with the page content.

From the Google AdWords site:

“Text links are hyperlinked brief text descriptions that take on the characteristics of a publisher’s page. Publishers can place them in line with other text to better blend the ad and promote your product.”

Hmmm. I wonder what this will do to the integrity of link language…?

At the moment, despite the fact that others have been using this in-line link-ad approach before, the hyperlink is a trustworthy little device for users. I can see some clever copy-writing being used to suck people into sites that they don’t want to visit. As the PPA model works on further actions or goals being completed, I think we’ll see some frustrating uses of this.

I guess I need to see one before I can work tis one out.

This is a little bit weird.

I was conducting some image searches today on Google, one for some fruit pastels and one for a badger (don’t ask) and this image of a badger revealed himself twice to me!

What are the odds on that?

And this one:

Browsing Results

In this post we’ll discuss the importance of information design in building faith in your search tool. In Malcolm Gladwell’s Blink he describes the subconscious art of snap judgement. This concept is directly relevant to the art of designing search as it’s on the results page where thou shall be judged and users are quick to draw conclusions!

We’ll cover:

• Information Design
• Snap Judgement
• Relevance Transparency
• Shaping Intent & Results Manipulation

Generally the target page of any user experience is the most critical and in the case of search, sound information design of the search results page, or SERP as it’s sometimes known, is vital in more ways than you’d first expect. Firstly it’s the target page and therefore the most important page in this experience, but interestingly it is a target page that is adapted and recast by the user as they shape their intent based on the tools and information you present them.

Information design is the order of the day here as the aim is to rapidly address the users satisfaction criteria and converge them on to a single, successful result.

There are many potential paths the user can take from the SERP including:

• View a result
• Expand the result set
• Contract the result set
• Navigate through the result set
• Search within
• Search again
• Change indicie
• Bail

/state the obvious.

The page needs to be designed to deal with all these concepts sufficiently in mind to guide the user to the right decision based on some clever work around the information hierarchy and design. However, there is a fine art in getting this right despite the fact that there are several conventions employed and an obvious pattern language established.

First and foremost when users arrive at this page they are seeking instant gratification, they want confirmation that they are on the right path, that the information scent is high. SERPs employ a range of strategies to meet this need such as:

• Reiterating the input terms
• Highlighting the matched words in the excerpt
• Showing the number of matches
• Providing suitable result manipulation tools

Reiterating the terms indicates that the user has ‘been heard’ and that they spelled things correctly and that the query formulation was accurate. Too high a number of matches suggests a lack of focus whereas too few matched keywords indicates a more general results set than desired.

Varying responses to these pieces of information start guiding the user onto distinct paths; shall I search again? Shall I investigate some of these results? Can I manipulate these results?

Note these examples whereby results, ads, controls and asides are well placed and presented:

Yahoo Search for ‘Samsonite Suitcases’

Google Search for ‘Samsonite Suitcases’

Note how Yahoo! provides a richer set of options, links to shopping, ‘Also try’s’, Sponsored results, ’shared by Yahoos’ etc.

(Hint: this is why I prefer to use Yahoo rather than Google).

In those first few nanoseconds where users are assessing the quality of response and simultaneously validating both their query and the quality of the search tool alot of this ‘conversation’ takes place. However, in these micromoments they still haven’t finished. The user is yet to zone-in on a particular search result and SERPs are best designed to meet this scan-reading behaviour.

Users need to be able to separate controls, ads and results clearly and easily.

Designing well for this situation means having a good understanding of your users goals. I personally find that the new image search on Google has flouted these basic principles of good information design as they have hidden key criteria behind a roll-over. It inhibits a users ability to quickly asses the results set and make a choice. With the new design they have to roll-over each image to ascertain detail that helps decide which result to focus on.

Well designed SERPs pay attention to the information and typographic hierarchy of the page and can guide users eyes through the layers being displayed, reducing cognitive effort and shaping intent.

• Users scan the page
• They identify a valid result set
• They may identify an invalid result set
• They assess the source of each result
• They assess the synopsis/extract
• They zone-in on a particular result due to a combination of synopsis, source and keyword match

It’s for these reasons that sites such as Yahoo! and Google are so well loved, in addition to the obvious ones about accuracy. Through the use of sound information design they reduce cognitive effort in understanding the information, its relevance and what could be achieved with it. They utilise a recognised (née established) design pattern so familiarity breeds trust through reduced cognitive load.

So, in these very few seconds there is a significant impact in the users perception of the brand - does it deliver on the proposition? Is it accurate? Is it easy to use? Do I get why it’s returned things in this particular order.

The order of the results is HUGELY important. Users want accuracy they don’t want cynicism. it should be clear why results are there and there should be a clear separation between actual results and paid for adverts. Perceivably the results need to present themselves as impartial, it has to be clear why elements are where they are and in many ways you should see diminishing importance in well designed SERPS.

Of course all of this is well and good, but what we really need is to address the users satisfaction criteria. This means present them with the right information so that they can make an informed choice about which result in the set to focus on.

Users will all have varying satisfaction criteria and its this criteria that defines their choice in result. For example one user may trus one source more than the other because they have heard of it, another user may find that the synopsis accurately summarises what they areafter, on image search a user might ust ant something of a certyain size (like when searching for album art).

For example, when conducting a web search, displaying the source URL for a result is vitally important in terms of the credibility of information. This information shouldn’t be first in the information hierarchy and it shouldn’t be buried either. It should nestle nicely within the second or third information layer (it’s where a typographic hierarchy is quite important) and it’s where Google have gone wrong in the redesign of there image search.

Yahoo! released a search tool called ‘Mindset‘ which aimed to create more accurate search result sets based on shaping the user intent. What does this mean? Well a user could type in ‘Volkwagen Golf’ into the search entry field and then hit search to see the results, then they could drag a slider from ‘Shopping’ to ‘Researching’ and the results below would resort themselves in real time.

On this page note the information design, the separation of ads and the clear ranking.

In essence this allows the user to further define what it is that they are trying to do. Do I want to travel to Paris or am I researching places to go when I’m there?

Post-query manipulation tools are critical in shaping intent and they are the subject of our next posting Refining Results

Then of course there is advertising, but that in itself warrants a whole posting and I sense your eyes are getting heavy reading this one, we’ll leave it there.

As a reminder, here are the 7 Stages of Search Experience:

1. Intorduction to framework
2. Search Proposition
3. Personalisation
4. Query Formulation
5. Error Prevention and Correction
6. Browsing Results
7. Refining Results
8. Discovering More

Readers that have dipped into some of my earlier posts will know that designing user experiences around site search and search engines has been one of my career pet topics. I’ve worked on numerous types of search tool and as such have long ago developed a framework for driving the interaction design of such an experience.

I’ve written these posts because on a number of projects that I’ve been involved in I have seen teams think of search as ‘a page’ within the system accesses via an omn-present text entry box, top-right. They are failing to realise the fine tuning required to created a well integrated tool for users to discover content and I wanted to provide a basic to medium level crib-sheet for creating search experiences. Consider it a checklist, though it’s not that list like I’m afraid.

Please see the following posts for the earlier parts in the series where I talk about things to consider when defining search user experience:

1. Introduction - 7 Stages of Search Experience
2. Part 1 - The Search Proposition
3. Part 2 - Implicit / Explicit Personalisation
4. Part 3 - Query Formulation

Continuing the series in Search Experience Design (which I have to say I completely forgot I’d started due to a new family addition - forgive me), we are now going to look at Error Prevention and Detection.

There are many different types of error prevention and correction that can be employed by search tools and they typically play out mid-query once the user has submitted something to the system. Some of them happen automatically, others require the user to clarify their intentions whilst others operate post-query on the search results page.

There can be a problem with keyword searching in general, I say referring to the earlier 3 mental models of search employed by users when devising a find strategy, I can again use the example of ‘thai green curry’. If the user simply enters ‘thai green curry’ then the search tool won’t be able to determine whether they want recipes for thai green curry or whether they to buy ingredients for it.

When users ‘talk’ to search engines they like to use this kind of truncated language but unfortunately it lacks the emphasis that would help them get better results, the error prevention and correction strategies employed by search tools aim to correct this.

These tools can range from mid-query devices (between the point of submitting a search request to the display of the results set) through to post-query devices such as Google’s infamous ‘Did you mean this instead?’

The purpose of these devices is to remove input error, reduce ambiguity and enable convergence to a more accurate results set. I’ve seen many a site specification delivered without these simple business rules included, some of the devices may include:

• Error tolerance (spelling mistakes Qwerry Vs Qwerty)
• Case tolerance (QWERTY Vs qwerty)
• Truncation tolerance (Qw… Qwer… Qwerty)
• Format tolerance (Qwerty01 Vs Qwerty 01)
• Colloquialisms (Localised labels / slang i.e. Bairn = Baby)
• Synonyms / Thesauri (similar terms, i.e. dog and hound)
• Near matches – Being able to match items that are thematically near rather than
  linguistically near. (Wedding Dresses – Photographers – Catering)
• Ambiguous queries – Dealing with a lack of clarity or trying to converge to a
  smaller result set (London, UK or London, Canada)

Post-query correction/convergence takes place within the search results page:

• ‘No Results’ sets – Handling situations where there is no matching information –
  pushing the user on to potentially relevant information
• ‘Did you mean?’ – Suggested alternative spelling

Interesting and progressive ‘No results’ handlers will always offer the user something for example a food site that didn’t have a recipe for ‘thai green curry’ should offer the user a range of other thai style dishes it has, or indeed at least some other relevant content.

Of course, the relevancy of alternative content and its ability to ‘reveal’ itself to users under error contexts is wholly reliant upno the content loading process and the metadata created with each record. However, our fictitious search tool doesn’t have a record so how would it know what to present?

I’m afraid this is where human intervention (and with it some emotional, rational and non-systemic logic) helps.

I once worked for a search company whereby 4 people sat and watched how people searched for things and plugged the gaps by building synonyms, relational data and a local vernacular. Of course this is okay if you govern your own content, but not if you are crawling others sites.

Back to error prevention and correction, it is also worth investigating any implicit corrections that take place on the query. As sometimes the search tool will identify a common spelling mistake and automatically correct it. Or it may be that due to a lack of results, the search tool has amended the search criteria and expanded or contracted the query, for example ‘there are no results for ‘Paintballing’ in London but there are for Surrey, Berkshire, Essex and Kent.’

Search help
I’ll quickly write something about help pages as they are used and can provide a variety of information to guide users in the finer points of your search engine. Topics that the basic level of help should cover include:

• The mechanics of search - how it works
• Detailed descriptions of each feature on the page
• Tips on how to conduct a successful search
• Explanations of al the search result manipulation controls on the search results page
• Advanced search techniques
• How to customise search

That’s it for now. there is obviously more detail but this post has already supersized and if you’ve read this far, the post is either interesting or you have serious staying power!

As a reminder, here are the 7 Stages of Search Experience:

1. Intorduction to framework
2. Search Proposition
3. Personalisation
4. Query Formulation
5. Error Prevention and Correction
6. Browsing Results
7. Refining Results
8. Discovering More

So, in a previous life I was a bit of a search specialist. I still am when it comes to the user experience of said search, but I have waned on what’s what on the tech front etc..

So with a view to upping my acumen in this area ,and yours dear readers, I’m going to trawl search related news and offer you the best of the rest.

Caveat here - if you are a search specialist/guru/swame, you will be picking this stuff up form the usual suspects like SEW and SER. So just expect similar news, but with my commentary.

Click to Call in Google Maps
When I worked at Yell.com it was always clear to me that Maps were going to eat the directory lunch at some point. Location based services lend themselves particularly well to this visual form, and there’s something inherently engaging and intuitive about maps.

Why is this?

People love maps and there is so much ambient information held within them. They perform on a primary, secondary and tertiary task level. You can see the geography of a place, it’s proximity to others and routes in and out. Of course there’s a lot more, but n information terms maps offer lots of layers in one hit.

I guess that’s why I love mindmaps so much as a note taking form.

How many times have you ever opened up a map just to look at it?

No?

Come on!

Anyway, Google have finally launched Click-to-Call on their maps. Eeek everyone else, this is a great thing.

Search for Pizza in your locale and then, when they are highlighted, click-to-call and you’ll be connected.

Oh the seamlessness.

Consistent Protocol for submitting to Search Engines
Anyone who’s ever worried about their effectiveness on key search engines such as Google, Yahoo! et al (yes okay, I’ll say it) and MS Live will be all to familiar with the curiosity of how the spiders and their algowotsits work.

Well, try this for size; the major players are coming together to support a standardised protocol called ‘Sitemaps‘. This means that site owners can now tell search engines about their content rather than expect those lil’ digital arachnids to do the work.

The black art of submission may have just gotten easier. But hang on, there is a whole industry built on this, so maybe it’s not that good.

If like me you fancy digging a bit more, go to the official site.

That’ll do for now, I’m on the train and have work to do.

Stage 3 of the 7 that we’ll cover in this series actually sees the user start to interact with the search tool.

Hooray.

By the time they have reached this stage some quick fire assumptions have been made subconsciously about what is on offer and how it all works and some implicit and possibly explicit preferences have kicked in. These soft preferences are driven by the default state of search options or user profile (Safe Search on/off, 50 results displayed, etc).

But we now arrive at the moment where users start to create a search query using the form elements presented to them on the page, but as we do so, please bare in mind the mental models I told you about in the Introduction to this series.
This is where we see just how ‘intuitive’ the design really is.

Some search devices are good at helping their users to structure a search query because they break
up the search string into specific requests for information. These are generally known as ‘fielded
searches’ or ’structured searches’ and they control the user input to such an extent, that in theory at least, chances of a successful search result set are improved. You may also see these searches called ‘parametric searches’ where there are a distinct set of parameters from which to choose and you’ll find these more convoluted offerings on car sites and the like. Here is a couple for you to play with:

• Autotrader UK
• Edmunds/Autotrader US

Yell.com (the UK Yellow Pages business directory company (who own Yellowbook USA btw and is the largest YP in the world)) has a structured search which requests any combination of up to three pieces of information; business type (i.e. plumbers), business name (i.e. Acme Plumbing) and Location (i.e. postcode or any place name).

Of course, this approach works well if your data is structured, but of course, not all of it is and this is why the user experience varies from search to search.

More common among traditional search engines is a single search entry field that spans the central area of the page, more commonly nowadays (following Google’s lead) you’ll see these fields along with a variety of indicies across the top from which to choose; Web, Images, News, Blogs, Groups, Whatever.

This is an age old approach that pretty much acts as the blueprint for search engine design patterns but I can’t help thinking that people still feel like they’re shooting in the dark a bit.

Users often make claims that they have to ‘learn how it works’ - which as a search strategy and a user experience isn’t particularly wonderful. This may also be why people tend to be fairly loyal to any given search tool – they’ve ‘learnt it’.

A9 has a search entry field that spans right across the page, but in trying to be different, probably offers up more complications and cognitive effort than ease of use as it displays some search indicies that simply aren’t clear.

Where I’m getting to is this, when users interact with a search tool, they can either enter a lot of information upfront and have targetted results, or they can enter a small amount of information and then ‘be in the ball park’.

Google operates in the second way, and if you have read the introduction to this series you will realise that this can be a poor user experience and is why I don’t think Google is a great experience.

Oooo. Did I just say that? Yes, and I’ll fill you in on the detail as to why later.

But essentially, if you have a number of inter-related search queries, a la mental model 2 or 3, say searching for a good recipe for thai green curry or planning a holiday to Malaysia, then Google isn’t very good at tying these things together. Instead each query is singular and the adjustments are made by the user. It just so happens that most adjustments can be made by the user within the first search results page (SERP) as they are so well designed.

Have a look at Yahooo! and Google when searching fo rthe term ‘Malaysia’.

See how Yahoo! offers up, above the first result, an entry marked with ‘Y!’? That’s a Yahoo shortcut that rather helpfully second guesses what it is the user is after. Google, is rather hapless and unhelpful with the same query as they don’t have the content to display in the way Yahoo! does.

Other stages yet to cover:

1. Error Prevention and Correction
2. Browsing Results
3. Refining Results
4. Discovering More

Stage 2 of the search user experience is another one that takes place without the user really doing anything. In this stage we look at the ways in which both implicit and explicit personalisation occurs to shape the user experience in a meaningful way.

This stage is of particular topical interest as Google come under fire for a personalisation patent infringement.

Some are obviously cookie or IP sniff based and infer some level of personalisation (i.e. Google automatically re-routing you to the local country domain), whilst some allow you to customise the display, personalise input preferences, whilst others go as far as to allow you to create custom search resources (I mentioned in an earlier post the service from Rollyo. This is of course enabled through a registration.

Of course there are a whole range of tools that come off the back of explicit personalisation, including services such as Yahoo! My Web where users can save, find and share favourite web pages. However, I’ll deal with these more later as the aspect of personalisation I wish to highlight currently are those features that facilitate query submission or the default disaply of results.

The aim of such ‘tailoring’ is to short-cut certain aspects of the UX for the user in order to speed-up, appropriate or improve the accuracy/relevance of their result set or indeed to customise the display of those results.

One other area to consider is the ‘default states’ of search options. Whilst not strictly personalisation choices, the fact that the ‘Safe Search’ tool is on by default defends the brand and meets user expectations as to how something should behave. For example, it would be disastrous if an organisation with ‘family equity’ in its brand such as the BBC were to provide a web search tool that trawled adult content and revaled all by default.

As users take more control over their user experience, searches launched from desktop widgets, browser tools or otherwise increasingly allow users to control their information sources, prioritisation and display. So it stands to reason that the more customisable your search tool is, the better it will fair in this two-way web 2.0 kinda world.

This idea of users controlling the way in which they access SERPs is fairly topical as this post at Search Engine Watch supports as it describes Google Base features allowing users access to result sets via RSS.

Interesting…More later

Other stages yet to cover:

1. 3) Query Formulation
2. 4) Error Prevention and Correction
3. 5) Browsing Results
4. 6) Refining Results
5. 7) Discovering More

The first stage opens with the proposition of functionality and purpose to users.

‘What is it?’, ‘what are you offering?’ and ‘how do I play with it?’ are question running through the subconscious mind.
The homepages of these different sites offer very different propositions to users in terms of what search is and how it works:

1. • Google
2. • A9
3. • BBC
4. • eBay
5. • RollYo

How they perceive the tool is key to their application and usage of it and this perception of ‘what it is’ shapes the kind of information they’ll use it to search for and assist in defining their mental model of what the site can do.

In this instance, perception is core to ease of use.

The perception of what it is that search does is affected by both the brand and by the placement and positioning of the search tool within the information hierarchy of the page.

In a Web 2.0 context, this becomes increasingly interesting as we consider the impact of executing searches away from homepages using browser extensions, desktop widgets and tools such as Rollyo.

Rollyo - or ‘Roll your own’ is a great little tool that sits in your toolbar and allows you instant access to search either the site you are on or it allows the user to constrain their search to a set of handpicked resources. It’s great for using on sites that have a crappy search tool and is a typical example of how users can experience your site on their terms.

Anyway, back to it as I think that thread of discussion warrants a post in its own right.

Placement of the search tool:
In terms of information design, the placement of search within the page has the effect of communicating how important the tool is a mode of interaction within a given site. If it is small and unassuming it could be argued that it is not the main navigation device or that it is the least preferred mode for discovering target content. Conversely, if it dominates the page it clearly suggests that search is a key way for finding information. That said, a search device does not have to dominate the page to be well placed and understood.

Yell.com has quite a distinct proposition to its users, it heroes search as the main ‘find’ strategy both in placement and in volume (It’s turned up loud in the information hierarchy). It is quite clearly a business directory and is best interrogated using a structured query.

Conversely your typical search engine such as the mighty Google offers users a natural language text field where they can put in anything they like. It does however allow you to choose from the indices across the top which starts to focus the user and build their query. We’ll discuss ‘Query Formulation’ in part 3.

(incidentally - I’ll tell you why I think Google isn’t a great search user experience, and the ways in which it could be improved significantly later - Grand claim, I know, but bear with me).

Brand:
The brand is also important when trying to understand how users build a mental model for what they can do on a site. Again as an example, Yell.com is a business directory and users visit the site to obtain business information. The Yell brand helps users to structure a query as they formulate a mental model of how they will find something (any combination of a business name, classification or location).

For a site such as Amazon, ‘a large product catalogue’ users and so search will obviously be a key device in finding specific items, many of the products searched for on Amazon are ‘uniquely identifiable’ – they can be described quite specifically; artist and album for example.

We will need to assess the affect of both brand and placement in order to try and ascertain exactly how users perceive the search tools.

This stage is a bit lame, but it does have an affect on what user think of your site. I’ve seen this time and again in search usability sessions and have come to realise that brand and placement have a significant affect on the interpretation of search.
Stages 3 - 8 are as follows:

2) Personalisation
3) Query Formulation
4) Error Prevention and Correction
5) Browsing Results
6) Refining Results
7) Discovering More