What Yahoo! should learn from 'Gatineau'

In a bit of a change of direction for me, I'm going to post today on something that has happened in the industry and not something that can help you with your business (sorry!).  You may or may not have noticed in the press recently that Microsoft have pulled their adCenter Analytics beta project.  In a post on the adCenter blog, Microsoft announced:

The beta has been closed, but the program will remain available to current users through December 31, 2009.

You'll  note that I've talked in the past about 'Gatineau', the original (and better) name for the analytics package and indeed I have it on my blog as a measurement standard.  I have even reviewed it, compared the data output with Google Analytics and blogged (enthusiastically) about some updates they made back in September.

The questions we need to look at here are a little bit different to when HBX got subsumed into Omniture.  Why?  HBX had a base of thousands of users paying a fee.  There were limited options of the users of HBX - they were paying and the new company they were paying offered them another option.  With Adcenter Analytics, the number of websites using it was probably very small (I'm wondering if there is any data out there on this - maybe from WASP?).  Even if it wasn't very small, none of them were paying for it, few of them were actually using it as their main source of analytics and everyone knew that it was still in 'beta'.  I suppose the question is what happened to all those DeepMetrix users from when Microsoft first bought it.

Anyway, all this means that there probably won't be too many websites directly affected.  On the flipside though, this could have all sorts of effects on the Analytics industry.  When Google first released their tool for free it was seen as making waves and it would only be a matter of time before everyone did.  Then Microsoft waded in and more recently Yahoo! have picked up the baton.

Yahoo! though could learn a lot from Microsoft's beta project if they are going to succeed:

1. Yahoo! need a Unique Selling Point


Am I saying that Microsoft didn't have a USP?  They did have a USP, it just wasn't very good - as they discovered fairly quickly.  Oh, the whole premise of it was brilliant.  They would take your web analytics data and allow you to segment it against ages, sexes, locations and job titles from their cleverly connected data belonging to those users with a .Net passport.  The only trouble was that most people browsing the web didn't have a .Net passport associated with them and hence the segmentation almost always had large amounts of unspecified data associated with it (rendering it useless).  Or it certainly did with my site.

What is Yahoo!'s USP?  Well it can't be the fact its free, because Google is free too, it was there first and has far more help for users because of that.  Therefore Yahoo! need to think outside the box.  It also can't be that it has built integration with their 'merchant solutions'.  Because that can't possibly have enough users to make it worthwhile for Yahoo! to use it.

Or can it?

Maybe Yahoo! has scoped out the market and realised that although Google has a large percentage market share, its predominantly aimed at bloggers who don't spend that much money on adwords.  Whereas Yahoo! has already got its own store built in that can help them.  Then it can use it to leverage not just built in analytics to sell the 'shop', but also having integrated it into an eCommerce products will allow them to sell more adwords (or whatever they call it).  eCommerce companies have more money to spend, because they make more money back.  Maybe they have gone after the right market.

More over Yahoo! need to make more of their segmentation.  People don't realise that Google's segmentation is ok, but a bit wobbly.   Moreover, most people don't really realise what they are doing with the segmentation or why they are doing it.  This is something that will only be born out if they manage to push the amount of commentary there is out there on web analytics...


2. Yahoo! need a blogging community


This was one of my bug bears with Microsoft Adcenter.  Bar their own blog (which was highly useful), nobody else really blogged about it.  Can't find something you need in adCenter Analytics?  Keep looking, because there aren't any blogs out there to tell you how to do it.

The brilliance of Google Analytics is not the number of professional services that seemed to spring out from nowhere, but the number of blogs helping you out with your analytics.  I was speaking to a friend in the SEO industry once about entering us in for an award.  He flatly declined stating that he'd have to give away the secrets they'd picked up along the way and that would hurt his strategy.  In fact, there are very few industries where people will give away tricks of the trade for free in order to get more people interested.  Why do we do it?  Maybe web analytics really is hard and there will always be a need for a talented and knowledgeable individual.

3. Don't spend forever in Beta


This is something Microsoft are not used to, but Google are.  Google put things in beta and leave them there on the basis that it means they can change them more easily without people moaning.  "What do you mean I can't use this functionality any more?  Oh it was only in beta, that's fine then." 

Yahoo! still haven't given open access yet.  I know it's not in beta (I don't think it is anyway) but whilst it is only free to a limited set of customers, they're not going to win much support (certainly not with the blogging community).  All the time this is going on there are a number of HBX users moving solutions that could have easily been snapped up by Yahoo! given their similar segmentation ability (I know, I know, that was a real generalistic comment).  These customers don't want to go to Google, because they don't see it as giving as in detail analysis as they need and are picking up SiteCatalyst and WebTrends as we speak.


4. Don't give it a name that is spelt differently in different English speaking countries



Just me on this one?  Yup, thought so.

Measuring your SEO efforts

Right, it occurred to me that my last post was particularly dull.  As pointed out by a couple of people on Twitter when I couldn't work out whether I'd posted it correctly or not (thanks again guys).  So this one is going to be more interesting insightful.  Actually this week is going to be a post on how you measure your SEO efforts effectively, which is always a hot topic.  Just out of interest, I'm not going to talk about how to create your SEO strategy or how to create and use your keyword footprint.  These should be things that you are aiming to do through use of competitor intelligence tools (cf Hitwise) as well as your content stance and your keyword analysis from your web analytics tool (external and internal if you have one).

In fact, it surprised me a little bit that I'd never talk about this before.  I suppose there are a few things that so many people have written about that you think you don't need to put them in.  Lets start at the top:

Measure your visits from search engines via organic search

Duh.  Of course you do. Go to your Traffic sources report in Google Analytics and there it is.  It even tells you the percentage from search.  Brilliant.  Make sure you don't include your paid search to start with though.

This is easier in some web analytics solutions than others.  Obviously Google make it easier for you to find in their tool, because, well, it's what they do.

In SiteCatalyst, you still have a similar report in your 'Referrer Type' report.  Again, this is going to be a bit misleading if you don't have the visits metric turned on and are using the 'instance' metric, but hey - lets get over that.  Bear in mind here that you are including your paid search traffic too.  Really you want to be looking in your Search Engines - Natural or Search Keywords - Natural reports.  Remember that if you don't have visits turned on, you should be comparing like with like and comparing your 'Searches' metric with your total instances metric on your referring domains.

Measure your visits from search engines via organic search minus Brand Traffic

This is a bit leftfield.  You should be getting your brand traffic to your site anyway.  You'll rank number one for your brand name, your product names etc (you should do anyway).  What you want to know is how you are doing for the rest of those search terms.  This is where the analysis gets quite tricky, but not impossible.  In Google, the easiest way is probably to create a custom segment:

In SiteCatalyst this should be easier to do due to your 'filter options' in the top right hand corner which allows you to exclude terms very easily.

Measure your conversions from search engines


This sounds fairly obvious, but not always.  You don't want just visits, you want your visits to convert as well.  Remember that segment you just created in Google Analytics?  Well fortunately you can now use that segment on your 'goal conversion' report as well. In SiteCatalyst it is all a bit easier, because you can just change your metrics to your conversion tags (keeping in place your current filter).

 

It's really that easy.  You've got a total at the bottom now.

Also bear in mind that, as I said in one of my first ever posts, you can use other things to measure your success if you don't have any conversions.  If this is the case, let's look at bounce rates, visits lasting more than one minute, etc, etc. 

Measure the number of unique search terms and unique entry pages


Wow - here is a bit of a new metric.  What you are essentially saying here is that you want to increase your keywords to as many search terms as possible (assuming that they all convert) and that you want your entry pages to be equal to the number of pages that you want your users to arrive at.  Theoretically speaking, if you have pages that are indexed in the search engines, but aren't generating traffic, it is because they are lacking the correct keywords.

 

In Google Analytics this is quite easy to do.  You have your total in the bottom right hand corner of your keywords report and if you use the drop down (highlighted with a red circle above) you can look at the landing pages for those terms.  Bear in mind that Google Analytics starts sampling the data above a certain level and hence this may not be the most accurate figure in the world (but is scalable over time).

These metrics are much more difficult to produce for SiteCatalyst.  Why is that?  SiteCatalyst doesn't tell you by default how many unique lines it has in its database.  I reckon that the easiest way of doing this is through the data warehouse.  The Data warehouse is a much under utilised product.  It gives you lots of tit bits of information that you probably wouldn't be able to get normally, but more importantly it gives you the entire dataset on one page.

(Ok, just as an aside, you can probably do this just by downloading a pdf of all of your search keywords from the main keywords report in SiteCatalyst.  But there is a reason for producing this anyway, which I'll come onto in second).

To produce this report what you need to do is create a segment that includes all your search engines (Search engine is greater than or equal to !) and that excludes your paid keywords based on the campaign that defines your paid keywords (there is an exclude tab on there too).  This is key because it just gets your data set to show just your natural search.  This is also the mechanism for creating ASI segments, however your natural search is probably too big for an ASI and would have too much processing, so this is better.

Having created your segment, you are ready to build your report.  You will probably want to do two of these.  The first one will be 'search keywords - all' and visits.  These both appear in your 'standard' Items.  The second report you'll want to run (with just the segment you've created earlier) are the 'entry page' and visits.  This will give you your two metrics as before.

If you are running your search term analysis at the same time as producing this report, it might be wise to add in your conversion (or engagement) metrics into this report to save yourself (and Omniture) a little processing time.  Remember this report we are looking at will view your overall strategy.  If you are trying to establish your keyword strategy and increase conversion for each keyword, you'll need to see the conversions for each keyword and this is an ideal way of doing it.

Measure the technical aspects of your content


This isn't so much a visit based set of metrics, but is vitally important none the less.  You should know how many pages you have available to index (if you're a content site you could try getting this from your database, else you could try the CMS).  You should also know how many pages Google has indexed from your site (here are the ones from this blog).  Ideally these figures should be as close as possible.  Remember this isn't a competition to see how many pages you can get indexed.  Ideally each set of content should only appear once (I know that the search engines have been talking about the canonical link parameter that you can put in your header to avoid duplicate content, but you should still try and have one url per bit of content).

Why stop with Google though - remember you get traffic from Yahoo, MSN (and Live), Ask, etc.  Make sure you are not throwing too many pages at them (or sometimes more importantly, too few).

This also makes it interesting when you compare this figure with the number of unique pages that search users are entering the site at.  It's all well and good having 5 million pages indexed in Google, but if on an average month you only have users arriving at 500,000, then you need to work out how you are going to optimise the other 4.5million.  This can help drive your strategy (should I be creating more content, optimising the current content better or re-evaluating my keyword footprint to get more appropriate terms).


Measure your reach


This is something that can seem not important, but I think is.  When you are a blogger, you spend half your life focusing on what other people think of you.  How can you get people to quote you in their blogs.  You have to write comments on other people's blogs, seed forums with your posts, put up links to it in Twitter, etc, etc (PS, DON'T SPAM).  How do you measure the effectiveness of this?

Well there are two thinks you can do.  You can use the 'link:' parameter in your favourite search engine (mine is Yahoo Site Explorer).  Every month if you see an increase, this is because more people are linking to you.  Also you want to check out your Technorati ranking.  This is a measure of how many other blogs have linked to you and their relative importance.  Mine is quite low, but then my content is quite niche and long tail (plus I don't write often enough).

What do you reckon?  Have I missed anything out?  (please don't say page rank :))