It’s the little thingsā€¦ like thumbnails!

A while ago I wrote about this topic, adding thumbnails to your listings in search results, not sure if anyone made any updates based on it, but I finally swallowed some of my own medicine and tried it out on maps.unimelb.

Each page now has a reference to a thumbnail meta tag, which is retrieved as a static image from the same source (open.mapquest) as our current tile background. The thumbnail uses the latitude and longitude coordinates to display a tiny map of the location.

Map thumbnail example

Try it out with this query:

Oops, you weren’t meant to see that!

It happens from time to time, things change, someone forgets to place the correct permissions on a page, doesn’t really matter how it happens, the problem is the same. Google is showing results that you don’t want shown. How do you stop it?

Once again, Google have pretty good documentation for this, but lets have a bit of a review of the situation.
Continue reading “Oops, you weren’t meant to see that!”

Promoted Results

Our upgraded search tool allows us to configure Promoted Results. Google uses Promoted Results for advertising, however, in our case, they’re the results that we believe are most appropriate to the keywords the user has typed.

Promoted Results have the potential to be a great help to users. There are about 20 things that people search for which account for 90% of queries. Many of these things are known by different names. Scholarships, for example, is probably what people are searching for when they type in ‘bursary’, a term which might be more common overseas. By creating a Promoted Result we can present to the user a ‘best guess’, without excluding the other organic search results.

How to request a Promoted Result

At first, we’re asking you to use the helpdesk. We don’t really know how much demand there’s going to be and what resources we’ll need to service that demand, so the helpdesk will provide us with some valuable reporting. Eventually, we intend that the process will streamline to a simple web form which results in a review process.

The Review Process is going to be important. We’ll be moderating requests to ensure they aren’t too general and are focussed on delivering the right message to users. If there are any problems or changes we need to make, we’ll contact you.

What will they look like?

Promoted Results will be visibly different from organic results. They will display a title, the URL and an optional description, wrapped in a blue box.
A Promoted Result for the Contact Directory
Not visible to the user, Promoted Results also contain a list of keywords, or ‘Triggers’, and an optional date range during which the Promoted Result should be available. A list of the currently configured promoted results is available to view.

Regular Expressions

Selecting keywords for a promoted result is often a bit more complicated than it seems – people can type all sorts of variations which google may or may not realise are related to your promoted result. We can use regular expressions to match patterns instead of specific words. So, for example any search in the form of a question like “how do I…” will trigger a promotion for Ask.unimelb.

Anyway, get to it. Request a promoted result via the ITS helpdesk. We’d love to see more sites using the feature. More general information about the Web Search service on the ITS website.

Synonyms in search

Synonyms are a feature of Google Search that we’re not expecting using a great deal, but worth discussing so that people are aware of the pros and cons.

Synonyms have two parts, a Term, and one or more Variants. When a user’s search query matches a configured Term, the search which is conducted silently adds the Variants, hopefully making the search more relevant.

This is great if you have terms which have a particular meaning in the context of the search engine.

At the moment, we only have a couple of Synonyms configured. One of them is alloc8, which was the name of the old timetabling system. It was included because, even though it’s no longer in use, the logs showed that people were still searching for it. Rather than have these searches go to a dead end, the Term alloc8 now invokes the Variant Timetable, which returns results that are at least relevant to the user.

The only problem with Synonyms is that they are silent. It may confuse a user to see all these results – none of which contain the actual term alloc8 anywhere. So, with that perspective, we’re going to be super careful about adding new synonyms, as they do have a potential for unexpected results.

We’ll soon be releasing a process that allows requests for new Synonyms to be included, so keep it in mind.

Provide meaningful results

Users search for a variety of reasons. They may be vague about what it is they really want, they may have failed to find what they want by navigating, they may just be in the habit of using search as a first port of call.

Site owners can help users who search, find what they’re looking for, by providing relevant, quality information in the results returned by searches. It’s not that hard, and your users will thank you!

There are four parts to a search result listing, and each of these can help the user make a decision about which result to click on.

The components of a search result

  1. Title: The title is the most important by far. It’s taken from the <title> tag of your page. Make sure that it clearly says what the page is about. It may get truncated, so ensure that the unique points about the page are in the first few words of the title.
  2. URL: Most users won’t really look at a URL, but if they’re logical and meaningful, they can help a user make a decision. For example, if there are two pages about an annual event, and one has /2012/ while the other has /2013/, that would help a user make a decision about which one to click.
  3. Description: Very important part of a search result. Google makes its own decision about what to use here; sometimes using the <description> tag, sometimes a snippet from the page, sometimes a combination of both. A well written description tag, which includes relevant keywords, is most likely to be used. If Google has to fall back to finding a snippet of text in your page, there’s a significant risk it will get something out of context and skew the apparent meaning of the result. In some cases, Google may even source the description from the Open Directory Project – not even your page!
  4. Thumbnail: A relevant image can support your message. This is most useful for an online shop, where an image of the product is displayed, but it can also be useful if there are images relevant to the page content. The problem is that Google will add any image it thinks is relevant, and it’s often not. You can avoid this problem by specifying the image you want used.

As usual, Google provides some useful information about this topic. There’s a page about Metadata, and some detailed information about writing good titles and descriptions.

Rich Snippets give you Star Power

Bacon is wonderful, isn’t it? And Bacon Roses are even more wonderful, right? And if you were to search for bacon roses in google, and you saw a recipe with a 5 star rating, you would click on it, wouldn’t you? Try it!, see if you can resist!

Of course, there is a point to this exercise (even if you’re a vegetarian), it’s those stars! How did they get into the Google search results? The answer is what Google call Rich Snippets; also known as Metadata, MicroData MicroFormats, Inline Markup, RDFa. Yes, you can add data to your pages that Google can display in search results. If you have useful data, it’s well worth doing, as you’ve just seen the power of a few gold stars on a recipe.

By default Google recognises Rich Snippets that relate to a range of different items (including, as in the example, Recipes). If you can make use of data you’re already serving in your pages to describe the nature of the data to google, then it has the potential to make your results much more attractive to users. It’s well worth doing, but how?

I’m not going to tell you how, simply because there are hundreds of different ways, and the best way will depend on the data you’re describing. So, instead, I’ll list some starting points.

I would love to hear of people who are already using this stuff around the Uni, so please, let us know.

Thumbnails on search results

Anyone who’s used Google will have noticed the little thumbnail images in the search results. Not all results have them, and even those that do sometimes have images that aren’t really relevant (apologies to the library!).

A not terribly useful thumbnail in a search result.

If you manage a site, however, you can change this by specifying a thumbnail to associate with your result.

The simplest method is just to use a meta tag in the head of your page:
<meta name="thumbnail" content="http://example/foo.jpg" />
That’s it! Just upload an image somewhere, and link to it. Google will resize the image to fit in a 60 pixel square, so if you want it to look it’s best, you should probably optimise the image to that size already.

The possibilities are endless. If you were listing publications, the thumbnail could be dynamically drawn from an image of the front page. If your site is in the CMS, you could generate the meta tag to automatically use the asset thumbnail.

There is another method of specifying these images using a structured content snippet that Google calls a ‘PageMap’. PageMaps are more complicated, so if you are only wanting to specify thumbnails, then there’s probably not much point using them, however, PageMaps can also influence a bunch of other result display options which I plan to cover in the coming weeks. As usual, Google has some pretty good documentation about Thumbnails, as well as the other aspects of PageMaps.

Robots, keep out!

Search engines use software ‘robots’ to examine pages, and follow links. These tools are designed to discover stuff, and they’re very good at it! If you leave a link anywhere Google can find it, then the content will find its way into the index, sometimes with embarrassing results! A robots.txt file is simply a way of telling search engines that you don’t want certain content indexed.

Why bother?

Good examples of content you might not want indexed might be your template files, test pages, archived content, and out of date content that still has to be publicly accessible. It’s worth keeping all of this out of the index, because the less irrelevant content you have in the index, the easier it is for visitors to find the good content. Google may even penalise a site that has old, poor quality, or duplicate content.

Setting it up

Setting up a robots.txt file is dead simple. It’s just a plain, unformatted text file that sits at the root of the website. Try typing /robots.txt after a few of your favourite domains, and see what comes up: (1,2,3,4,5) you can learn a lot from seeing what others do.

In most cases, however, you’ll be keeping it pretty simple. First, you’ll specify who you’re giving directions to, and most often, it’s everyone, so you’ll use a wild card asterisk character:

User-agent: *

and next, you simply list the things (individual files, or whole directories) you don’t want indexed, so your complete file is going to look something like:

User-agent: *
Disallow: /mytest/
Disallow: /template-assets/
Disallow: /old-home.html

That’s it! If you want to test your robots.txt file, there are any number of free tools on the web. This one seemed to work quite smoothly, and there’s a good one in Google’s Webmaster Tools.

Auto Completions

One great feature we’ve enabled in the relaunched search engine is Auto Completion.

Auto Completion is triggered as soon as you start typing in the search input field – little javascript functions fire off queries to Google’s servers, which return ‘best guess’ matches for what you’ve started typing, and present them as a list of suggestions.

As we use our search engine, Google will take note of the text people type and the Auto Completions they follow, and the suggestions will become more and more relevant and accurate to our needs. It’s truly amazing stuff.

Of course, as smart as all this is, Google’s algorithm won’t always get it right, so we have the option of managing our Auto Completions to include or exclude patterns and results that we think are relevant.

A custom Auto Completion simply consists of a string of text. So, if we were to insert an auto completion of “The Graduate School of Foo”, it would appear whenever someone typed a query which matched any part of that string. When a user follows that Auto Completion by clicking, a search is performed on the whole string.

We’ve also enabled Promotions to appear in Auto Completions, which may offer greater impact, and we have the option of adding Promotions which only appear in Auto Completions, so there’s plenty of flexibility with these features.

If you have a problem which may be resolved by using Auto Completions or Promotions in the search engine, ask us about it via UserVoice, then log a Help Request via the IT website to make it happen.

Getting your site noticed

Search engines are smart, and we all use Google because it’s one of the smartest. However, even Google can’t give useful results if the content supplied to it is not properly presented. Now that we’ve improved our Google search tool, site owners will probably want to see their sites perform better.

The investment in improving your site for search engines is paid back two fold, because as well as improving the visibility of your site for internal searches, it will also improve in public searches via

SEO (Search Engine Optimisation) is a bit of a loaded term. In the past, SEO techniques have been abused to bring sites to the top of results for search terms that may not have been related to their content. These days, however, the search engines continually evolve their ranking engine to nullify the effect of false techniques, and reward sites that return relevant results to users. Google never disclose exactly how all this works, but they do provide some excellent documentation for web site owners that, if followed, will help make a site more visible in Google.

The main points are simple, just ensure that your:

  • Site is logically structured
  • Pages are Accessible, with headings that relate properly to the content and logical, semantically structured information.
  • Pages have unique, logical titles and well written description meta tags
  • Ensure your URLs are easily understood

If you just do these things, Google, and other search engines should love your site. There’s more that you can do, and if you really want to get tricky, make sure you read more of Google’s documentation, but just these things will cover the 80/20 rule and should see your site ranking well in the results.

Worth noting that changes may take some weeks to be reflected in the Google index, but if you think your site is having problems, then there are a few options for requesting assistance.

Number of posts found: 19