This was the ‘Search Advertising Advanced Exam’. I got 88%. This makes me a Google Advertising Professional (since I think you have to pass 2 of the 4 exams to qualify). Last month I passed the Google Advertising Fundamentals Exam.

You can see the contents of the various different exams at the AdWords Certification Programme Learning Centre, which also links to all the training materials that are available.

So, Google Analytics reports the search keyword that was used to find the website – you can also see the number of sales or other conversions that resulted from that query. However for Adwords traffic the keyword that is reported is the “bid term”, the keyword that is targeted in the Adwords account. This is different from the real search term that people used in their search, because Google matched the real term to your bid term using “broad match”.

It is possible to get the real search term to be shown in Analytics, and therefore to look at the performance of the real keyword using the metrics available within Analytics, (e.g. Bounce Rate, Time on Site, Goals and Ecommerce). This is done using a custom filter for Analytics. This will rewrite the campaign term to show both the bid term and the real search term in brackets:

As you can see, often the bid term is the same as the real search term, but sometimes it is not. In the example above, the 4 entries for ‘Clinicians’ would normally be combined into a single entry for that keyword.

You can use this information to adjust your Adwords campaign, for example you can add additional keywords into the Adwords account (e.g. “new zealand clinicians” shown above) and depending on the ROI of the keywords you could adjust the bids or even prevent the ads being shown for a particular variant if that didn’t convert well.

Instructions

First of all, remember, as with all filters for Analytics, make sure you test the filter out on a test profile, and always have a Raw profile to which no filters are applied.

This requires 2 custom filters …

1. This filter restricts the data to CPC / PPC traffic only, using a Custom -> Include filter. If this filter were not in place then natural search traffic would also have a second keyword added, and this would always be exactly the same as the first one. The details are:

Filter Type: Custom Filter -> Include
Filter Field: Campaign Medium
Filter Pattern: (cpc|ppc)

2. This filter extracts the real search term out of the referrer string and also extracts the campaign term (the bid keyword) and then it writes both the bid keyword and the real search term back into the campaign term. The details are:

Filter Type: Custom Filter -> Advanced
Field A -> Extract A: Referral: (\?|&)(q|p|query|search)=([^&]*)
Field B -> Extract B: Campaign Term: (.*)
Output To -> Constructor: Campaign Term: $B1 ($A3)

Contact First Rate to maximise the return your organisation gets from Google Analytics.

A couple of weeks ago we blogged on New Zealand Facebook statistics, in particular we reported:

• There are more Kiwi women on Facebook than men (57.5% female vs 42.5% male) and
• Most of the New Zealand users of Facebook are in the 18-24 or 25-34 age range.

This led me to wonder, what is the percentage of each gender in each age range? Are there any interesting trends? Are there any age ranges where men are more common?

And also, now that Facebook is the most popular site in NZ (aside from Google), what percentage of the NZ population is on Facebook now, and how does that change for each age range?

By the way, in case you hadn’t heard, Facebook overtook TradeMe in July as the most popular destination site for New Zealanders – here’s a graph from Google Trends for websites. (P.S. If you want to know what the other most popular sites in NZ are, here’s the Top 100 according to Alexa and the top 20 according to Hitwise).

Graph of visits to Trademe.co.nz and Facebook.co.nz (New Zealand visitors only)

So, let’s use the demographic information from the Facebook advert tool to understand New Zealand’s Facebook demographics a bit better:

The gender gap is strongest for older users

The younger age ranges are much closer to gender parity than the older age ranges. The 20-24 age range has the highest proportion of men, with 47% male and 53% female. As the age increases though, the percentage of men decreases. The 50-54 age range has the lowest proportion of men, 35% male and 65% female.

% of each gender on Facebook, by age range (% of those who stated a gender)

Nearly all young kiwis are on Facebook

A famous quote about Tolkien’s The Lord of Rings is that the world is divided into two types of people, those who have read it, and those who are going to read it. It could be that the same is true of Facebook in New Zealand – kiwis are divided into those who are on Facebook already, and those who are going to be on Facebook soon!

Dividing the number of Facebook users at each age range (from Facebook’s advert tool) by the total number of New Zealanders in each age range (from the NZ census website) shows that the younger age ranges are almost totally assimilated already. Nearly 90% of the 20-24 age range is on Facebook already. The percentage of Kiwis on Facebook falls quickly as age increases – only 30% of 40-44 year olds are on Facebook. And the percentage continues to fall slowly as age increases – only 17% of the over 60’s are on Facebook.

Estimated percentage of the NZ population on Facebook, at each age range, for each gender and total.

Of course this is just an estimate, and in particular there is no guarantee that the number of users reported by Facebook is indeed correct. But given the popularity of Facebook in New Zealand, it can’t be too far wrong.

So if you want to try an advertising medium that will reach a very high proportion of New Zealanders, then Facebook should be your second choice after Google Adwords. First Rate incorporates Facebook ads in the overall online marketing mix (which also includes SEO, Adwords PPC and performance advertising, all tracked by Google Analytics). Please do contact us to understand how Facebook advertising can benefit your business.

The fine print

Facebook allows advertising on a cost-per-click basis as we illustrated last time. And these adverts can be targeted by age, city, gender, education, relationship status and keyword, allowing very strong demographic targeting. The advert tool has these options for the advert targeting:

Demographic targeting options for adverts on Facebook

As you can see, the estimated number of people is shown, and this is where I got the estimated number of NZ Facebook users at each age range from, including gender:

Estimated total number of people in NZ on Facebook

Estimated number of women over 30 in NZ on Facebook

Estimated number of married women over 30 in NZ on Facebook

P.S. Plenty of other blogs have used the same trick to get useful insights about the demographics on Facebook. Here are a selection of some of the best:

• Facebook Market Research Secrets and 2/3 of Facebook users aren’t (necessarily) women by Alec Saunders
• Facebook Secretly Reveals Demographic Data on the All Facebook blog
• Where Does Facebook Grow From Here and Facebook Growth By Country and the Slowdown in App Usage on the O’Reilly Radar blog
• Facebook Gender Demographics – A Comparison by Country on the Ad Chap blog

Have you ever typed in the address of a Google service and been surprised when that website didn’t exist?

Google uses a sub-domain for some of its products, (e.g. adwords.google.com, earth.google.com, mail.google.com) but uses a path name for others, (e.g. www.google.com/analytics, www.google.com/adsense).

This inconsistency causes confusion amongst Google users who may type in a URL containing the product name, but use the wrong format. The classic example of this is Adwords: Type in www.google.com/adwords/ (with the trailing slash) and you may be astonished to discover:

So, this got us thinking here at First Rate, how many of Google’s products have both the path name and the sub-domain version set up, and how many of them have only one of them set up. And is there any rhyme or reason behind Google’s choice of a sub-domain or a pathname?

First Rate tested the pathname and sub-domain version for 50 of Google’s different services and products. The names tested were:

First Rate tested the following six URL formats:

• http://name.google.com/
• http://www.name.google.com/
• http://google.com/name/
• http://www.google.com/name/
• http://google.com/name
• http://www.google.com/name

The results confirmed First Rate’s suspicions: Many of Google’s products do not have the alternative URL set up:

• 64% of the products had both the sub-domain and path name versions set up (e.g. earth.google.com and www.google.com/earth or wave.google.com and www.google.com/wave).
• 14% had the path name set up but not the sub-domain (e.g. www.google.com/postini/ but not postini.google.com).
• 12% had the sub-domain set up but not the path name – some of the services had the path name set up for the version without a trailing slash, but not the version with the trailing slash.
• 10% had neither the path name nor the sub-domain version set up (e.g. android.google.com and www.google.com/android or insights.google.com and www.google.com/insights).
  Sometimes, this was because the product exists on its own domain, (e.g. www.android.com or www.orkut.com) and sometimes it was because the product in question has a longer URL than that tested, (e.g. insights is actually found at www.google.com/insights/search).

Further details

First of all, let’s ignore some of the tests – almost 100% of the tested URLs with no www. redirected to the www. version.

This is true both for URLs that exist at the redirected location (e.g. google.com/analytics redirects to www.google.com/analytics) and for the URLs that do not exist,(e.g. google.com/android redirects to www.google.com/android). Therefore, we haven’t considered the non www. versions of the URLs any further.

62% of the products did not have the www.subdomain.google.com version set up.

Perhaps we can let Google off the hook, as many websites with sub-domains setup fail to allow the www.subdomain version.

However, it is inconsistent and perhaps careless nonetheless, that half of the of the Google products that do have a subdomain have a www.subdomain version – and half don’t!

Finally, here are the full details for all the tests. (Note that the link checking software that we used failed to report on all the 302 redirects, it reported all of these as 200 ok – so the status code for the test may be incorrect.)

Product
admanager
adsense
adwords
affiliatenetwork
alerts
analytics
android
androidmarket
base
blogger
blogsearch
books
checkout
chrome
code
dirhp
docs
earth
feedburner
finance
friendconnect
grandcentral
groups
health
images
insights
labs
local
mail
maps
mobile
news
orkut
pack
picasa
postini
reader
scholar
sell
sites
sitesearch
sketchup
talk
translate
trends
video
voice
wave
websiteoptimizer
zeitgeist
Icons ref

Some other notable broken URLs include:

Blogsearch
www.google.com/blogsearch/ (with the trailing slash) – weirdly this one redirects to http://blogsearch.google.com/blogsearch/ which then doesn’t exist.

Feedburner
www.google.com/feedburner – you’d have thought that feedburner was a big enough product that there must be plenty of people who type in google.com/feedburner

Website Optimiser
websiteoptimizer.google.com is another important Google product with a path name but no sub-domain. First Rate loves this product, it is excellent for improving a client’s website conversion rate.

The Conclusion

Well, really the conclusion is obvious: Google as a “multi-armed hydra” has a variety of different business units and divisions, and perhaps this is the reason that there are no consistent website standards enforced?

And on top of that, Google don’t really need to care – because as a search company, Google knows that if someone fails to find a website they can always do a Google search to find it (and that’s why SEO is so important..!)

Sometimes, the search results for Google’s own products are less than good, e.g. they may lead to an older version of the page which has no obvious link to the newer version, or the phrase used may not match the product name and so the official page is pushed down the search results. Or sometimes even the link may be broken! – But that is a story for another blog post.

UPDATE: 10th September 2009

Matt Cutts has just pointed out to us that some of the inconsistencies are in fact intentional, on the basis of performance. Thanks Matt!

Don’t believe that there are too many websites with speech bubbles in their logos? Look at the graphic that everyday have assembled – it has 60 different Web 2.0 logos. You can’t argue! It’s been done to death, it’s passe now.

Our client Mesamis has launched their summer range. I have just completed designing various newspaper and magazine ads. Photographs are supplied by each clothes Label and I photo retouch and colour manage each image to the publishers specifications for high quality reproduction.

First I had to assemble my sample. I used two Google searches, for Accessibility Policy and for Accesskey (both were obviously restricted to .nz websites only). Then I took the first 30 or 40 results from each, removed duplicates and pages that were not the list of accesskeys for that site, and that gave me a list of 50 websites’ accesskeys. I have published the full list of all 50 sets of accesskeys in XML format. It can be seen that the list is dominated by government websites – this is not a surprise because the NZ Government web guidelines place a high level of importance on the accessibility of government websites, so many of them have implemented accessibility policies and accesskeys.

So what lessons can be learnt? Um, I’m working on that. Stay tuned!