Someone searched for the following and got to my blog:

change price in google adword text without reducing “”quality score”"

They probably didn’t find a good answer at the time so I’ll have a go now.

A common concern is that if you change anything at all in the ad copy you start afresh with a new ad and lose all benefits of a good quality score.

Quality score is tied to keywords not ads so in fact you don’t lose the quality score history by changing the ad copy. It’s a bit counter-intuitive. People click on ads. Click through rate is one of the biggest factors in determining quality score. However, it’s the click through rate associated with the keyword that counts. The quality score isn’t reset if you change an ad.

Google has just introduced a new facility in Webmaster Central to record the geographic location of a website.

This could be useful for businesses in Australia because the majority of people here use Google Australia (google.com.au) and Australian websites rank highly compared with overseas ones.

The domain name is primarily used to determine the geographic location so a .com.au site is recognised as Australian. Other factors are also considered such as where the website is hosted. Problems arise when the domain is a .com and especially if it is hosted overseas. The new functionality will allow the correct geographic location to be recorded.

All good except I wonder if we’ll see increased competition in Google Australia and “pages from Australia”?

The new feature doesn’t allow multiple countries to be specified at the top level but it does allow a different country for a sub-domain. So, an overseas site can have www.something.com/australia.

I don’t know whether it will make any difference to the rankings but it will be easier for global companies to nominate a portion of content to be specifically targeted in Australia.

Vanessa Fox has all the details.

I’ve been following the commentary on Podcamp Perth, mostly because my friend Ross went and it always looked like it would be a good event. Next year maybe.

Darren Rowse does a good job of maximising the value of an opportunity. He was interviewed in the Sydney Morning Herald and The Age about the page rank drops. He’ll get quite a lot of people clicking through to his site so he addresses them specifically with a welcome post. It’s a good way to capitalise on an expected source of new visitors and do as much as possible to keep their attention and convert them into regular visitors.

Blogs are great for flexible, speedy responses to events.

My Google Analytics stats show that 70% of visitors spend less than 10 seconds on my site and 66% of visitors visit one page only. When I first looked at these metrics I was really concerned that I was doing a terrible job at engaging my visitors. I came up with a list of actions such as linking to related posts, improving the categories, encouraging comments etc . I’m absolutely certain that there’s a lot of room for improvement and that action plan is still valid. However, I do have a much better insight into the metrics after reading Avinash Kaushik’s fantastic book on web analytics.

Time On Site

When someone first visits a website, a session is started and the time is logged. From that point, every time a request is made to the web server, there is a timestamp. The time spent on a page is calculated as the difference between the timestamp when the visitor goes to the page and the timestamp when they visit another page. All well and good except that there is no way of knowing how long the person spends on the last page they visit as there is no time stamp to indicate the end of that event and the beginning of the next. The tools calculate the time spent as zero.

This makes the metric particularly misleading when the website is a blog. Although I might like people to go deeper, read other posts and have a look around, it’s perfectly understandable that someone visits and just reads the latest post. The problem with the stats is that whether they spend 10 minutes reading the post or 5 seconds deciding it’s not interesting, the time on site is still zero which then distorts the overall “average time on site” metric.

Page Views

The number of pages viewed is usually considered to be important when trying to measure the level of engagement. The more pages someone looks at, the more interested they are in the website as a whole. It’s not quite so straightforward with blogs. Subscribers and repeat visitors are a good measure of engagement and yet these visitors will typically come just to read the latest post, or the last few posts which will probably be on one page. And then of course they might not visit the site at all and read the posts somewhere else like a reader.

So what should you measure?

Well I guess the metrics are still useful as long as you understand the inherent flaws. And subscriber stats and number of comments are useful indicators of engagement. Anyone want to offer some feedback on what metrics they find useful and the best way to measure engagement on blogs in particular?

I have a better understanding now of what’s happening when sponsored links appear to be unrelated to the current search query and a hangover from the previous search query.
I blogged about it a couple of weeks ago showing an example where a search for wedding celebrants followed by a second search for string quartets resulted in a mix of ads for both celebrants and string quartets on the same page. Simon from the Found Agency pointed me to their blog post on the same thing where they have screenshots showing combined results for car insurance and child care.

Google’s explanation quoted on Search Engine Land is:

look at the user’s previous query and see how well it intersects with the current query

My initial interpretation was that this is based on some kind of semantic or topic matching. In other words an assessment of how relevant the second query is to the first query. I lost faith in the relevance argument when most of the examples I came across were for two queries that appeared to be completely unrelated - like the car insurance and child care example.

However it appears that the intersection is based purely on the combination of the actual words in the two queries. Mike Churchill has a detailed explanation but basically if all the words in a keyword phrase are present in the combination of search query one and search query two, then the ad will show on the second search if it didn’t for the first. For example, someone has the keyword phrase “large orange widgets” in their campaign. A search for large widgets will not show the ad but if this is immediately followed by a search for orange cars then the ad for large orange widgets should appear in the second set of search results as the two queries combined contain all the words.

The end result is untargeted and irrelevant ads. Not good! Even worse, as Mike points out, there isn’t a great deal you can do about it without adding a lot of complexity to the campaign management. There’s quite a lot of discussion happening about adwords “features” that are unwanted and detrimental to advertisers. This thread about expanded broad match on high rankings is one example. Maybe we’ll start to see some changes if Google starts to listen to their bigger advertisers or if enough people start recommending wholesale changes like avoiding broad match keywords.

There was an article today in Canada’s national newspaper about the impact of the online search market on ad agencies.

I came across it on Andrew Goodman’s blog and I’m linking to it because I think it’s relevant to the Australian market. The paid search market in Canada is cheaper and less competitive than in the US. There are parallels with the Australian search industry in this respect. We are a couple of years behind the US and this creates opportunities for businesses and agencies.

Google’s president of marketing and commerce makes the point that advertising agencies shouldn’t see the dominance of Google as a threat but as an opportunity. He concludes with the message that (in Canada) “we’re underinvested compared to what the opportunity is.”

Andrew points to Gord Hotchkiss’s post on the same topic. Substitute “Australia” for “Canada”. The post is very relevant to our market as well.

The Australian Blogging Conference was excellent and well worth attending. A few people have already blogged about some of the sessions and no doubt there’ll be more posts over the next few days. Peter Black, did a great job organising the conference and has said that he will do a round up of the coverage on the conference blog so that’s going to be a good resource to get notes from individual sessions.

So, I thought I’d just cover the business blogging session as it was one that I particularly enjoyed and also one of the most relevant to this blog.

Read the rest of this entry »

I’ve started to notice personalisation happening in the delivery of adwords. An example came up today when I searched for “wedding celebrants melbourne”. I noticed an ad for “string quartet melbourne” and then searched separately for “string quartets melbourne”. I was shown a mix of ads for wedding celebrants and string quartets. Any search for string quartets after that had only relevant string quartet ads, no celebrant ads. I was able to replicate these searches with the same results.

Read the rest of this entry »

Is it reasonable to expect to know the rules when paying for advertising? The Google adwords blog stated yesterday that changes to the adwords quality score algorithm will not be pre-announced prior to implementation.

What’s odd about this is that there is quite an extensive help file for adwords. It’s not like the organic search algorithm which has never been outlined officially to the same degree. A search for quality score within the adwords help centre brings up 90 results and they answer questions like “what is a quality score and how is it calculated?”. Ok so at the end of every list breaking down the inputs to quality score there is the catch all “other relevance factors”. So there has always been the assumption that things are taken into consideration other than the main factors like click through rate, ad copy relevance and landing page relevance. However the announcement seems to hint at changes that might be worth announcing but nevertheless won’t be.

This was one part of the announcement. The post was mostly about the types of sites that may be singled out for a poor landing page quality score. These are:

• eBooks that show frequent ads or install malware
• ‘Get rich quick’ sites
• Comparison shopping sites
• Travel aggregators
• Affiliates that don’t comply with our affiliate guidelines

I was surprised to see comparison shopping sites and travel aggregators lumped in with the rest. On reflection it fits in with the philosophy of showing a broad variety of results to searchers as demonstrated by new initiatives such as universal search and the longer standing filtering of duplicate content.

I’m working on a new campaign which fits into the category “comparison shopping sites”. The website fulfils all the guidelines as laid out in the adwords help centre. It has relevant and original content on all landing pages. The campaign is organised into targeted ad groups with relevant ads. It’s only week three so still very early days but the campaign is suffering from a low number of impressions and I’m assuming that the quality scores are predictive and low and it’s a waiting game before realistic quality scores come into play.

It’s a different story if the site will continue to be penalised because it fits into a category that Google treats with special consideration. Comparison sites and travel aggregators are going to have similar content to other sites in their niche. Does it mean that only a subset of sites will be allowed to advertise and the rest will be forced out with high cost clicks due to quality score penalties?