Last week it was announced that Google’s search suggestion feature is being updated. The major changes are:

1. You will see results and ads in the suggestions, along with the keyword suggestions (it looks like only one result and ad at any time)
2. The suggestion feature will be rolled out to the standard search results page – at the moment it is only on Google’s home page & toolbar. (I guessed they would do this in my post about Google Suggest last August)
3. The suggestions shown will incorporate your search history

I haven’t seen the new features live yet – but I wonder what impact this will have on the referrals you see in your web logs. If someone searches from Google’s homepage – and clicks on a result from the Google suggest list – then the referrer will presumably be google.com. This will mean that the referral URL won’t contain the search term. I recently saw google.com in our own web logs but didn’t think anything of it at the time. I wonder if was a referral from Google Suggest?

If someone searches from an existing search page then the referrer will be that page and won’t contain search term they were actually using. That means the referral URL will contain a different keyword – possibly completely irrelevant. For example if I search for Microsoft then type in Apple and choose the Apple.com homepage from Google Suggest then the referral URL in Apple’s web logs would be something like: http://www.google.com/search?q=microsoft and analytics software may mistakenly assume that they were ranking for the term microsoft.

There may be similar problems with the reporting for paid search. Presumably the folks at Google have thought of this and have some sort of solution.

Has anyone seen evidence of this happening?  Is there anything Google could do to help us accurately see the search term that was actually used (for example sending people through a redirect that contained the keyword in the URL)?

When Google added a color picker to their image search last week I thought it would be worthwhile sharing a few ideas about the use of color refinements in site search. First – let’s take a look at what Google have done. When you do an image search a drop down menu is available at the top. When you click on it you see a selection of 12 standard colors that you can filter the results by.

Goolge will be doing some analysis of the images to work out which are the main colors in each images. For the few test searches I have done this works well.  Google shows the same 12 colors in the drop down even if there are no results. I think it could be improved slightly if they only showed the colors in the drop down for which there were results. Nevertheless, I imagine this will be a popular feature.

I recommend having a similar feature in your site search if you have results where color is important to your visitors. We normally recommend that you don’t use drop downs for the refinements because the drop downs make it slightly more difficult to use (one extra click) and you have to click on them to see what refinements/facets are available. The trade off is the refinements take up more screen real estate. By putting the refinements on the side and using a grid view you can still get plenty of results above the fold.When displaying a color filtering option – you should only show the colors for which there are results (this is the same for other refinement options as well).

Some problems

Color filtering can get a little tricky if you have products that come in several colors – if you don’t have a photo for every color of every product then the results may look a little strange. For example if you have a shirt that comes in blue and green but you only have an image of the blue shirt – then when you filter the results to show those that come in green – the blue shirt will show in the results. This is strange for the user because they’ve asked to see only the green products yet the image shows a blue shirt. It’s a valid result because the shirt does come in green but you need to tell the user somehow.

One solution is to take an image of every product in every color available and show the appropriate product when a color filter is applied. This can be very expensive and time consuming and if you’re custom making the products (for example sofas) then you may not have every combination available to photograph. Another solution is to show the available colors in the search result. Your Blinds do this in the example below. When you mouse over a color a larger image of the color pops up along with it’s name.

Choosing multiple colors

The example below from Jelly Belly shows an implementation that allows you to select more than one color. In this case blue and pink are selected. Jason Marrone, Ecommerce Marketing Manager at Jelly Belly mentioned in the ecommerce podcast that they have many visitors who are choosing candy to match a color theme for a function. These filtering options definitely help those people. I like this implementation – with the ticks in the colors that are selected.

Automatically pulling color from images

We did some experiments a few years ago where we analyzed the product images on a customer’s site and automatically tagged them with the major colors that appeared in the images. This worked reasonably well – but we didn’t have a technique for excluding background colors – which meant it didn’t work perfectly. I’m sure there are techniques for removing background colours – such as ignoring any colors that are on the edge of the image – or something more sophisticated. In any case we haven’t had much demand for this. If you don’t have your products tagged with their colors then let us know if you’d be interested in having a service like this. It will help us work out whether this is something we should be putting some more effort into.

I was reading Jacob Neilsen’s Alertbox article on Interaction Elasticity today. In the newsletter he mentions that shorter navigation paths are usually better. “4 easy clicks are more usable then 5 easy clicks, because the extra click is more work for users”.  This sounds obvious and it got me thinking about how it relates to site search (I pretty much do that for everything I read).

One of the nice things about search is it is a very efficient way of navigating on a site. You click on the search box, you type in your search term and press enter (or click on the search button). Then if the search is relevant the page you are looking for will be above the fold and you can click on it. If the results are not relevant you may need to click on a facet, a sort option, a related search, scroll down, go to the next page of results – or do another search. This is why relevancy is the single most important feature in a search. If the results are relevant it reduces the amount of work you need to do.

The auto complete functionality that we released a few months ago shortens this path further by reducing the number of keystrokes you need to make – by 8 on average. I see Google is testing enhancements to Google Suggest to not only show search terms but also results and ads as you type. Showing search results as you are typing reduces the path even more – you don’t need to see the search results page. For this reason I think  this type of functionality will become more widespread – both in web search and for site search.

Another idea that we are starting to encourage our customers to investigate is to put the focus on their search box on most of the pages on their site – essentially any page that doesn’t have a more important form on it. This again reduces the navigational path – you don’t have to do that first click on the search box before you start typing. You also don’t have to do the work of finding the search box (which on some sites can be surprisingly difficult) – you just start typing after the page is loaded.

You don’t see many sites that put the focus on the search box. Google does it on their homepage. They’ve done a lot of usability testing – so it must be a good idea. Amazon doesn’t do it. They’ve done a lot of usability testing – so it can’t be a good idea. Mmmm. Personally, I really like it. I think Amazon should do it - let’s see if it catches on elsewhere.

I saw an article by John Ferrara discussing how Alan Turing’s ideas apply to search (thanks to Daniel Tunkelang from Endeca for pointing this out on his blog). John observed that people often expect search engines to understand them in the same way that a person would. While search engines aren’t able to do this, one tactic he suggested was to help people formulate their queries by using a suggest function. I’ve blogged before that I am a big fan of this functionality. Yahoo has had their Search Assist for over a year, Google recently added suggestions to their home page and we’re offering our Auto Complete to all of our customers (now over 300 sites and growing). John predicted that this functionality will be ubiquitous in a couple of years and your site will look behind the times if it doesn’t include suggestions. I agree.

Another approach we take to query formulation is to show related searches on the search result page itself, both at the top and/or bottom of the results and with each search result. These enable people to execute a new query just by clicking on one of the search suggestions.

John went on to talk about how to improve the search assuming the user has done a good enough job of phrasing the query. He quite rightly pointed out that the best result is often not at the top and suggested reviewing your search logs to identify the most popular queries and tuning the results for those queries. I agree that it is important to  review and tweak the results for the most popular terms – but this approach isn’t scalable because search has such a long tail. The approach we take at SLI Systems is to watch which results the users are clicking on and rank those results higher. This lets the users do the tuning and means the search continually gets better without a huge effort on your behalf.

I see Google has finally put the Google Suggest functionality on their home page. I’m a big fan of this feature and use it all the time in their toolbar. We’re recently started offering our equivalent functionality (that we call Auto Complete) to all of our site search customers. This is being well received – so you’ll start to see this on a lot more sites. I think before too long people will come to expect this feature from a search box. I expect people will use the feature more as they see it appearing in more and more places.

I noticed Google doesn’t have the Suggest functionality on their search results page. I expect it will come eventually.

Aaron Wall spotted Google showing Related Searches below search results. This is his image.

I really like these – we’ve been generating them for almost a decade. They made their first appearance on the original snap.com (which later became NBCi.com) using our GlobalBrain technology. We now show them on most of our customer’s site searches. Here they are on the Travel Channel:

Barry Schwartz on Search Engine Land asks how Google generates these. I’m guessing that they are popular search terms used to find that particular URL. As Barry pointed out, if you search for one of the suggestions, get more links, you’ll see the URL from SEObook ranking top.

This type of related search is reasonably popular on the site searches we host for our customers. Typically we will see about 5-10% of people who search will click on one of these links.

Last week You Tube announced that they will optionally show you search suggestions as you type. I really like this feature – it helps people formulate queries in an intuitive and familiar way. Most browsers have a similar functionality that will show previous queries you have entered – so people don’t have to learn anything new to use this. 

My main complaint is that it is not on by default. You have to click on the advanced link beside the search box and select the option. So very few people will use it. I’ve also see some suggestions that don’t have any results – they need to make sure this doesn’t happen. However these are minor complaints. I’m sure this is a trend we will see more of.

I came across a new search engine today: RedZee. Apart from the cute but slightly annoying logo the main thing that is different about them is the way they present their search results. I am always interested to see new ways of presenting search results – maybe there is something we use to improve the site search experience for our customers. Unfortunately I didn’t find that at RedZee.

The search experience is something like Apple’s coverflow which is used for browsing through albums. Apple’s coverflow has never pushed my buttons – I’ve always seen it as more of a gimmick. I definitely didn’t like the interface for searching – mainly because you have to use your mouse to scroll through the results. You only see the title and description for the result that is currently at the top. This interface would probably work better for an image search.

There are a few other issues – presumably the most relevant result is the one at the top – is the next most relevant one to the left or the right – which way should I go? It looks like they show the homepage of the site, rather than a shot of the actual page you’ll go to. So in this example for a search for cars the Apple homepage shows featuring the MacBook Air, even though the result is for Apple’s trailer for the movie Cars.

To the RedZee people I say nice try – I think you’re one idea closer to a good one.

Google’s new Search within a Search offering seems to be getting a fair amount of attention, but not necessarily for the right reasons The New York Times piece this week: A New Tool From Google Alarms Sites touches on some issues that should be of concern to retailers. A recent blog by Alan Rimm-Kaufman: Google To Vegetarians: Eat The Damn Hamburger was a little more blunt

In my mind the best outcome from this is that it gets people thinking more about site search. It highlights the difference in the quality of Google’s search within a site compared with the site’s own search. I’m always amazed at how many sites have terrible search (this is why we’re in business). For these sites the new Google offering could be a benefit by making it possible for people to use Google to search their site. Although there are concerns about competitors ads showing on these results. However for those sites that have a good quality site search it will be vastly superior to the Google site restricted search and this feature will not be very useful. In essence, the new Google search-within-a-search takes control away from the retailer (or publisher, or whatever entity’s site is being searchedHere are the drawbacks that I see

1. The Google-provided search results will not have the visual appeal that other quality site search offerings provide – e.g. they don’t include images or other visually appealing elements
2. Because Google doesn’t understand the structure of the data on a company’s site, the search won’t allow for things like merchandising, sorting, refinements or promotional copy
3. Google does not index every page of every site, and they often don’t have recently launched or added products, so a user might use the Google site search feature and be misled into thinking the site doesn’t have the item they seek, when it could just be that Google’s indexer is not up-to-date
4. Relevance of the search results is key – if the user does type a term into the search box, how relevant are the results that come back? They could actually lead the user away from the retailer’s site, and cause the company to miss out on a conversion that their real site search solution would have generated

The other major issue is that it will likely backfire with Google’s advertisers. By increasing their control over users who have done a brand query for ‘Best Buy’ or ‘REI’ and then showing ads for those companies’ competitors on subsequent Google search-within-a-search results pages, Google is now directly competing with their retailing advertisers (and other sites) and diluting the brands of these sites. And, they’re potentially leading the user to a competitor’s site via their ad. In my view, this is not good practice – it doesn’t benefit the user and it certainly doesn’t benefit the retailer.

Ideally Google will allow the site owner to control whether or not this box appears for them. They could do this through the webmaster tools. An even more useful feature would be to allow the site owners to specify the search form – so they could send people to their own site search. It would be interesting to see if Google would allow this – because it would remove the opportunity for them to make revenue from the ads on these pages.

If nothing else, I hope this prompts site owners to look at the quality of their site search – they should make sure its delivering relevant results that are well presented. Whether people search their site through Google or on the site directly, companies need to make sure people have an easy path to finding the items they seek. Usability experts back up what we’ve been saying for years – ‘if they can’t find it, they can’t buy it.’

Wikia Search launched this week to much fanfare. The new community-based search engine from the same people behind Wikipedia intends to shake things up for Google and other major search engines by “democratizing search” – that is, delivering results based on what USERS think is important instead of a mathematical algorithm.

I like the basic concept because it’s similar to our approach – we allow users to influence the relevance of the search results implicitly by watching what they click on and this makes a huge difference to the relevance of site search.

Wikia is searching the web and is very similar to the approach taken by our sister company Eurekster with their Swiki product. The difference is Swikis are focused around particular topics and they combine explicit and implicit voting.

Wikia’s biggest challenge is firstly getting people to contribute to improve the relevance of the results and secondly getting people to use the search. I think, despite all their venture funding and the profile of their founder they will struggle to overcome these challenges.

Chris Sherman was fairly scathing of Wikia in an article on Search Engine Land yesterday: “yet another crappy search service…”. I tend to agree even though I like the basic concept.