What Did Google Do In 2009?

It’s often said that bad news travels fast, and so it is with Google. The company that uses the informal motto “Don’t Be Evil” has found itself under a lot a scrutiny in the past year especially from privacy groups. Recently there have been reports that the Google Toolbar continues to send browser information even after it’s been disabled. The company has also faced a lot of criticism from publishers and copyright holders for its Google Books venture.

At the same time though, they do continue to roll out useful – and usually free – new tools for people to use. The video above is from their Webmaster Central video series and contains a good recap of some of the positive things that Google has done for users, developers and webmasters over the last year, some of which might have passed you by.

Unfortunately, the one that is of most interest to me right now, Social Search, is a Google Labs experiment that I can’t get access to yet.

[On a side note, it's interesting that the video is available in 360P, 480P, 720P and 1080P, so YouTube are definitely going for TV integration at long last.]

So, where do you stand on the Google Are Good / Google Are Evil debate? Do you mind that one company knows so much about you and your habits, or do you just enjoy the benefits of their services? Let us know in the comments below.

Tagged with:  

Labelling the World

Composite image of the Earth at night.

I’ve been looking a lot at the way we locate things, through maps and through the naming of things. Our addresses are weird things to begin with. We usually start by naming the place we are in and then move outwards in increments into the world around us. It shows some of the evolution of our idea of position.

In the past we wrote our address and most of the people who read it would know within the first few words who and where we were: “Oh that’s Joe who lives in the next village over” they would say, and that would be an end of it. They wouldn’t have to think about Counties, States, or Countries. Who talked to people in other countries anyway, except kings and their like?

Nowadays the world has grown smaller, and in many ways it would be preferable to have an address that narrowed with each phrase, each increment bringing the sender closer to the addressee. In the online world, our lovely folk-evolved addresses seem more trouble than they’re worth. Wouldn’t it be nice to have a little code we could parse?

Well, of course, some such mechanisms exist, but many of them are old things that evolved out of archaic postal systems and are still maintained for those systems. What should it matter to me how a postman is routed from his supply depot?

Something like Geohash on the other hand provides a code for all places on the earth, much like latitudes and longitudes, and to a high degree of accuracy – it’s a sphere, it’s not too hard. So, we can easily choose one system for everywhere on earth, and not have different systems for the UK, USA, China, India, etc. Just one common system.

We can then leave it up to each postal service to translate that system to their old postal code thingy, and we have no barrier to entry to the finding places game. The resulting addresses are nice and simple too, but even though the code-to-place procedure becomes more open and accessible, it would still be nice not to have to type a code into a phone to find out you are going just round the corner. Well, a Wikipedia type thing (not Wikipedia though, we’re not all worth an encyclopedia entry) could easily map locations to names, and using semantic web structures we could get to know a lot about each place we bother to name.

My own country (Ireland) has always been blessed by the ability of putting things off for a long time. Sometimes this enables it, when it finally gets round to doing things, to adopt systems unencumbered by legacy. So, if we choose to have an encoding for places, why not use Geohash or an equivalent?

If the Irish postal system wants to connect their own system to this it should be relatively simple, and we won’t be left with some debacle like those poor people in the UK, whose once public, now private(-ish) postal system had it’s postcode system adopted as a de facto identifier standard, even though it’s not a particularly good identifier, except for delivering mail. And just to make matters worse, their government, who are supposedly interested in unlocking innovation through free access to public sector data and information, continue to allow the postal system to charge for access to the postcode system!

Thank god my government is bound to:

  • Learn from this, and not spend millions on some new weird standard.
  • Choose, or make, a freely available system.
  • Quickly reap the benefits.

That’s what we’re going to do, right?…

Tim.

Tagged with:  

How To Improve Data.gov.uk

OK, so the Data.gov.uk stuff hit a raw nerve with me yesterday. In itself, it was pretty disappointing and the reporting also touched on one of my pet hates: Why do a lot of journalists not ask questions any more? They just seem to repost what they’re told. But, I’m past all that now, so here are a few thoughts on how a government data website could be better implemented.

The W3 put a bit of effort into a data browser extension for Firefox called Tabulator. It’s nice and deserves to be better. If I look at a page on Wikipedia with Firefox, say for example

http://en.wikipedia.org/wiki/Ystrad_Meurig

Then if there is data I am interested in as a starting point, I can go to Dbpedia and at http://dbpedia.org/resource/Ystrad_Meurig my data browser extension will find a graph of data that it can understand.

This is really nice.

The Data.gov.uk site has a query page for looking at its semantic data, but no clues as to what I can ask for. Dbpedia has a query page and from my Ystrad Meurig example I already know a lot about how Dbpedia might be storing its data. I know that I can ask for latitudes and longitudes using the W3 positioning predicates (using Tabulator I can browse from the data page to those predicates and find out more about them – it’s all linked). I know Dbpedia has a thing called ‘distance to Cardiff‘ so I could query Dbpedia for all things with a distance to Cardiff that have latitudes and longitudes and then I could plot them on a map.

This is properly linked data. This is what a government data site should be like.

I mentioned the Ordinance Survey ontology in yesterday’s rant. I like it, but it could be better. It has a solid structure for an administrative geography of the UK (not including Northern Ireland). However, the current version 2, is already out of date. A number of the unitary authorities were merged last year. This information is already reflected on the corresponding pages in Dbpedia, along with nice tagging to link the new authorities to the data on the authorities they have replaced.

The OS version 2 ontology replaced version 1 in a fairly unhelpful way, but that was OK because they said they were still playing around with how to work with ontologies. Will the next version play well with the current one?

The Dbpedia way of doing things means that not only do we end up with an up to date administrative structure but we also maintain a history of that structure. That history can be useful if we have to consider people and not administrations. A person might get around to acknowledging a change in the administrative make up of his area – eventually – but it wont happen immediately. The online structures need to be able to link them from old knowledge to new concepts. Here is the advantage in all these new ideas: Nothing leads to a dead end, everything is given more meaning by its connections.

The other Tim

Tagged with:  

It’s The Data, Stupid.

Data.gov.uk

Data.gov.uk

Oh, the excitement of it! Tim Berners-Lee is getting governments to listen to his cry, to set data free. Oh, the disappointment of our first look at the UK’s efforts! Where is the semantic data? Where are the ontologies to link concepts across datasets?

[For those of you not interested in the technical side of things skip over the next paragraph if you like - it's just technical ranting...]

This being a first pass the semantic data and the ontologies may be in there, but if they are they’re well hidden. There is a sparql page but no indication of which values are searchable. All the data sets I looked for were available in CSV and XSL; hardly linked. Turning one of the CSV data sets into RDF using well known namespaces took me about 30 minutes, so it shouldn’t be too hard for the site to get better, and quickly. Will it?

OK, that bit aside, the point is that the launch of this site seems to have been a deadline achieving exercise rather than an announcement of anything actually being ready. That being the case, somebody needs to put up their hand and say “That’s rubbish”.

It’s especially hard for me because I’m as excited about the possibilities of the semantic web as my more illustrious name-sake, but this ain’t those dreams, not even close. I was hoping to be able to complain loudly in the pub about my own useless government here in Ireland and how they weren’t doing anything to make their data available. “Look at how good the UK is”, I could have said. Oh well, another day.

So why am I so disappointed? What is this stuff anyway?
Well the semantic web can be a way to connect… well everything.

When I talk about a thing, say Sutton, I can link it to a description and eliminate any possibility that I am talking about a different Sutton. “Which Sutton?” you ask. “Sutton, Peterborough; Sutton, Craven etc…” and I answer “Exactly my point”. I can link it though:

http://www.ordnancesurvey.co.uk/ontology/AdministrativeGeography/v2.0/AdministrativeGeography.rdf#osr7000000000001643

[N.B. Don't go to the above URL unless you want to download a massive file!]

which is a link to the Ordinance Survey UK’s ontology for Administrative places in the UK. If this excellent data (if a little limited, and a little out of date) is on the new data.gov site, then it’s hard to find. It would be a good way to tie geographic data sets together. An arbitrarily named field in an xls file is, on the other hand, not a good way to link data together.

Nice idea Tim, now you need them to actually do the work.

The other Tim

Tagged with:  

When Advanced Segments in Google Analytics were introduced just over a year ago I was delighted. Finally there was an easy way to slice our website data in more than one dimension at a time. By setting up an advanced segment for just “Organic Visitors from the United States” for instance I could really drill down into what a specific set of users were doing on the site and see how their behaviour differed from the norm.

Every now and again though I used to see some strange results come back for reports I was doing. At the start I put this down to the feature being in Beta, but the anomalies didn’t go away. Having come back to use advanced segments in recent days to dig into some very specific user behaviour I noticed the problem was still there so I decided to find out what was causing it.

As an example, when I did a search (filter) in my top content report for all the URLs containing “/dentists/” I got a figure of 313,000 page views for the period of time I was looking at. However, if I swapped out the filter and used an advanced segment I had set up to do the same thing (i.e. only include data for pages that contained “/dentists/” in the URL) I got a figure of 434,000 page views, and I could see pages that clearly didn’t match the segment I thought I had setup.

At this stage the difference was far too big to ignore, and I assumed that lots of other users would be aware of it too. I was right. The Google Analytics help forum was full of questions about why data sliced with filters wasn’t matching data sliced by advanced segments.

The answer came in one line from a guy called MikeOstrowskiASU. He said simply that “Advanced segments are based on visitor sessions.”

Suddenly it was all clear. The advanced segments when used on the top content report were bringing back all the pages visited by anyone who had visited a page with “/dentists/” in the URL during their session rather than just the pages with “/dentists/” in the URL.

Just to confuse matters, advanced segments on visitor reports do match up with filtering because they are based on user sessions. For instance, using an advanced segment for “Visitors from the United Kingdom” on the visitors report will match up with going into the Map Overlay and clicking on the UK.

So, the moral of the story is be careful how you use advanced segments and what you infer from the data you get back. They might not be doing what you think they are.

Have you had a similar problem with Google Analytics, and have you found any workaround for it?

Breadcrumbs

Website Breadcrumbs

Bank in September we made several changes to RevaHealth.com to improve visitor engagement. One of these changes was to add breadcrumbs into the header of the page for visitors that want to broaden their search criteria. We weren’t 100% sure about this feature and we have been keeping an eye on our metrics to see what their effect is.

Overall we were very happy with the results of the experiment. Bounce rate was down by about 4% and the average number of pages viewed was up by 15%. Best of all conversion levels also increased, and not just because less people were bouncing. So we were happy, right?

Well no actually. Since we had made several changes at the same time and only done A/B split testing and not multivariate testing we did not know which changes had contributed to the positive effect. This is a frustrating state of affairs because although the overall effect was positive certain features may actually have been having a negative effect but were being compensated for by the other positive ones.

So in order to get a better picture of what had worked and what hadn’t we dug through our logs and metrics looking for anything out of the ordinary and we discovered some startling information.  Despite the breadcrumbs occupying some of the best real estate on the page almost no one was clicking on them.

Over the two month period of November and December the breadcrumbs were clicked on 18,000 times by 15,000 visitors. Over these two months we had 700,000 visitors (note December is a seasonally low traffic month) visiting 2.9 million pages.  Just 2.1% of visitors were clicking on the crumbs resulting in less than 0.7% of all pages viewed.

So The Breadcrumbs Were A Failure?

Unfortunately life isn’t that simple. The breadcrumbs may have had other benefits that their poor usage was masking. It could be that their presence on the page put the information being displayed into better context for the user resulting in our reduced in bounce rate or it could be that they were the catalyst that increased the conversion rate.  However, since we didn’t do multivariate testing there was no way for us to tell with our data.

So What Can We Do?

It’s simple really – we’re going to do more testing.  We are going to roll out a version of site without the breadcrumbs to see if their presence has any affect on our bounce or conversion rates. Once again this is going to be a straight A/B test rather than multivariate testing. If by having the breadcrumbs on the page we don’t observe a noticeable reduction in our bounce rate or an increase in our conversion rate, then we will have to ask ourselves whether we could be using that prime real estate more effectively.

Do you use breadcrumbs on your site? Do your visitors use them at all or are they there purely for information purposes?

Tagged with:  

The Cheapest Advert You’ll Ever Get

Every time a search engine results page displays a snippet of one of your web pages it is an advertisement for your website. In one way it is a free advert, but another way of looking at it is that you have paid for it through the cost of creating, publishing and marketing your content.

The search engines provide you with two lines to get your unique selling proposition across to the user and it is amazing the number of sites that rank well but fail to take advantage of this. The two lines I’m talking about are the content below the title and above the URL in the search results.

RevaHealth.com - Google Search Result

Companies agonize over their Google AdWords text, endlessly tweaking and A/B testing, determined to egg out every last cent of value. These same companies don’t expend the same effort in optimizing the flavour text that the search engines display in their search results.  This makes no sense! Increasing the click through rate of a well ranking organic search result, one that appears in the top five results say, will always have a greater effect on your traffic than any increases you can bring about using CPC traffic for the same keyword.

So How Do You Control The Flavour Text That Google Displays?

The bad news is that you can’t completely. Google will typically take a section of text off your page that is most relevant to the user’s query.

The good news is that while you can’t control it completely you can control it for your main keywords. This is done through the Meta Description field on your page. If Google trusts your site then there is a very good chance that it will display your Meta Description as the flavour text in its search results.

If Google isn’t displaying your Meta Description for your main keywords then look at the section that it is displaying. You should now modify this section so that you get the most from your listing in Google.

What Should Your Advert Look Like?

Make It Eye-catching. When you buy a newspaper ad in your local newspaper they charge for two things: the size of the advert and the position in the paper. Size matters. The bigger your advert is the better, so make sure you use up all of the available space that the search engines allow. After all they’ve given it to you – use it. Just make sure you don’t use more space that they have given you and have half your killer sales message cut off.

Get Your Keywords In. Even though it doesn’t matter from a ranking perspective Google will bold any words that match the keywords that the user has searched for.  Take a look at the flowers example below and notice in the flavour text that the words buy and flowers are highlighted. Google does this so users will easily be able to see which results are relevant to them.  The bold words stand out and attract the eye. Don’t forget that your search result is an advert for your site – you want it to stand out as much as possible.

Include A Call to Action. Whatever your site does – tell the user to do it. If you sell books say ‘Buy books’, if you process tax refunds say ‘Get your tax back’, if you do restaurant bookings say ‘Book a table at your favourite restaurant’.

Include Your Main Selling Points.

Online flower retailing is a highly competitive business, and it is easy to see which companies get it right when it comes to their search results. Take a look at the example below; the three companies shown have clear unique selling points in their search results.

Make The Advert Unique And Relevant

Each page on your site should have a unique Meta Description.  If you don’t make them unique you are reducing the chances that Google will use them. This is because if they aren’t unique Google knows that they don’t accurately reflect the actual content on your page (Google Webmaster Tools will tell you which of your pages have duplicate descriptions).

Worked Example – Online Flower Retailers

The online flower business is highly competitive and as an industry SEO is absolutely vital so we should expect these guys to get it right.

Eye-Catching – All three get the main keyword ‘buy flowers’ in their results text and therefore get good bolding. Similarly all three sites use close to all of the available space without any loss of meaning.

Call To Action – Interestingly 1800flowers.com doesn’t start with a call to action. Instead they focus on a sales message. This wouldn’t be what I would do. That said, I have to assume that they do it because it works.  Both of the other two start with strong calls to action: ‘Order Flowers’ and ‘Buy Flowers’. If I was in the market to buy some flowers I think I would have a strong positive reaction to these calls to action.

Main Selling Points – Buyflowersonline.com gets three selling points in addition to their call to action. Proflowers (which incidentally has one of the highest conversion rate of any website at 42%) gets four selling points in.

In my opinion 1800flowers.com is getting fewer clicks than its position would merit because they aren’t using the space Google has given them as effectively as their competition.

As an interesting aside you may like to know that the letters ‘e’ and ‘r’ are commonly mistyped when they appear beside each other in a word. If you look at the second listing above you will see that buyflowersonline.com is targeting these errors by putting ‘flowres’ at the end of their page title (the first line of the search results) and when you search for this in Google they jump straight to the top.

Have you any tips for what to include in your Page Title, Meta Description or URL to make your search result more clickable?

About Face Clinic

Kambiz Golchin on TV3's Morning Show

For people looking into having cosmetic surgery it can often be hard to get other people to talk openly about their own experiences. TV3′s Morning Show recently ran a piece that featured a 20 year old woman who had a breast enlargement and 42 year old woman who had Botox treatment. They also ran some questions by Kambiz Golchin, a surgeon at the About Face Clinic in Dublin.

You can watch the 40 minute video of the discussion of cosmetic surgery over on TV3′s website, but be warned, it does show graphic scenes from the breast enlargement surgery.

Tagged with:  
© 2010 WhatClinic.com Blog