• http://www.example.com/
• http://www.icrossing.co.uk/file1.html
• http://www.icrossing.co.uk/file2.html
• http://example.com/another_file.html
• http://www.acompletelydifferentdomain.com/

You might find yourself in a position where want to take this list and reduce it to just the domain portions like this:

• example.com
• icrossing.co.uk
• icrossing.co.uk
• example.com
• acompletelydifferentdomain.com

It’s possible to use an Excel formula to do this. Here’s one I found by searching:

=IF(LEFT(LEFT(SUBSTITUTE(A1,”http://”,”"),FIND(“/”,SUBSTITUTE(A1,”http://”,”")&”/”)-1),4)=”www.”,MID(LEFT(SUBSTITUTE(A1,”http://”,”"),FIND(“/”,SUBSTITUTE(A1,”http://”,”")&”/”)-1),5,256),LEFT(SUBSTITUTE(A1,”http://”,”"),FIND(“/”,SUBSTITUTE(A1,”http://”,”")&”/”)-1))

This is fine, but as you can see the formula is rather long and combining this with any further functions would be a real headache. To try to solve this, I wrote an add-In for Excel that adds a handy function to perform this step (along with a few others).

Download the URL Tools Excel Add-in Here

To use the Add-in, you’ll need to save it somewhere on your computer (we use ‘C:\Program Files\URL Tools\’ by convention, which helps when sharing workbooks) and install the add-in via the Excel Options menu as follows:

1. Click on the Office Button in Excel

2. Select ‘Excel Options’ from the bottom of the menu

3. Select ‘Add-Ins’ from the left hand menu in the resulting dialog box

4. Click ‘Go’ where it says ‘Manage Excel Add-ins’

5. Click ‘Browse’ and browse to wherever you saved the Add-in file

6 Click ‘OK‘ in the appropriate places to return to Excel

Hopefully, you will now have the Add-in installed. To test this, do the following: Put a URL in cell A1. Any URL, but try something long like http://en.wikipedia.org/wiki/Noel_edmonds. Then, in cell A2 type ‘=wwwsubdomain(A1)’. All being well, this formula should return ‘en.wikipedia.org’. If so, celebrate, otherwise go back and check the installation steps again.

There are a number of other functions included in the Add-in that I will summarise below. There are a couple of options for handling the domain extraction outlined above, but they are subtly different. If in doubt, you probably just want to use ‘=wwwsubdomain’.

• subdomain() – returns the domain part of a URL including any subdomains e.g ‘http://www.example.com/index.htm’ becomes ‘www.example.com’
• nowww() – removes ‘www.’ from the start of a URL (note that www must be at the very beginning of the URL)
• wwwsubdomain() – basically, a combination of the above, equivalent to ‘nowww(subdomain(A1))’. This is what you want to use most of the time!
• urlencode() – Encodes a URL string (try it on a string like ‘Hello World!’)
• urldecode() – Un-encodes a URL string (try it on a string like ‘Hello%20World%21’)
• tld() – Returns the TLD of a URL (not 100% perfect but pretty good)
• geturl() – Extracts the URL from a Hyperlink

I hope you find this useful. It’s very much a work in progress so please let me have your feedback or questions in the comments section below.

Post from: Connect

URL Tools Add-In for Excel

Below are seven ways Google Analytics can be used to help make the right decisions during a redesign.

1. What browsers and windows sizes should we support?

When designing your site you want to ensure that the majority of your customers can view your website and convert without browser versions or sizes preventing this.

I’d personally think that you should design your website in a size that at least 95% of your visitors can see without having to scroll horizontally. An easy way to check what browser sizes your current visitors are using is to go the Screen resolutions report under Visitors > Browser capabilities > Screen resolutions.

By comparing to site average you can also see if any particular screen resolution(s) has a significantly higher bounce rate

In the graph above you can see that the wider screen resolutions are actually performing much worse – this could be addressed by using a liquid layout or JavaScript to position content.

A quick way to check if your site has browser compatibility is to look at the browser report and compare each type to the site average, for the website below it is clear that there is poor support for safari.

The mobile browser report can help to decide whether a mobile version of your site is needed by looking at the bounce and conversion rate of mobile visits.

2. What territory specific pages or languages do we need to support?

The languages report under visitors is a useful way of seeing the proportion of foreign visitors to your site; this is particularly useful if you have an international website as it may indicate territory specific versions required that may help increase conversions.

Additionally, even if your website is UK focused, you may find that foreign speakers use your site.  This is especially relevant if your business is travel related as foreign people living in the UK or visiting short term may be interested in your services.  See this report under Visitors > Languages.

3. Can our customers find what they are looking for?

The internal search reports are really insightful as they show what your users are looking for once they are on your site, giving you an indication about missing or difficult to find content. A word cloud is a quick way to see what search terms regularly appear.

To create a word cloud, download the data, put the search term in column A, the number of times it was searched in column B and paste the formulae =REPT((A1&” “),B1) down column C. Grab all the words from Column C and paste into Wordle.

Word cloud from the internal search of World Vision, one of our charity clients (posted with permission)

4. Which landing/promotion pages need improving?

In a redesign it is important to understand which pages need changing as well knowing which pages are working well and should be left as they are. Google Analytics offers several reports & metrics for interpreting the performance of individual & groups of pages:

Goal funnel visualization – one of the most easy to understand, it graphically shows where visitors are leaking out of your sales funnels. Goal funnels need to be set up manually and when done so can be found under Goals > Goal funnel visualization

Site overlay – Using this report (Content > Site overlay) you can see at a page level what visitors are actually click on helping to show the effectiveness of call to actions and promotions. Not only can you look at the number of clicks on these items but also the ecommerce value i.e. users clicking this button have gone on to generate xx revenue

£ index – this under used metric, shown in the top content report, shows how important visits to a particular page are when visitors convert. You may find that pages like the about us, shipping & returns pages have high £ index values as visitors often view them before making a purchase and help a user decide if they should buy your products. Sounds obvious but it might suggest that including a shipping and returns policy info box on product pages helps improve conversions (particularly likely on high value goods).

Bounce rate/exit rates – bounce rate shows the number of visitors that viewed one page and then left, while exit rate shows visitors that leave the site after viewing a particular page (they may have viewed multiple pages before). These metrics can help you understand which landing pages are under performing or common places that visitors are leaving your site.  

(Page URLs removed from image)

However, these reports, like most analytics, must be understood in context to know if there really is an issue or not. When evaluating a page it is important to consider the objective of the page, how does its performance compare to other pages and are any problems a result of a design, content, product or brand issue (always difficult to know!).

The analysis of the above reports provide a great starting point for user testing as they can help identify issues on your site that are preventing conversions.

After launch

When a new site is built it is obviously important to bench mark performance before and after using KPIs that are relevant to your business such as visits, conversions, average transaction value, revenue etc.

However, it can take a few weeks for these metrics to settle down making it difficult to tell early on if there are any problems with your new website. By setting up custom alerts you can get near real time warnings of any errors or changes that may have occurred. The sensitivity of these alerts depends on your website, suggested values are used below

5. Is tracking set up correctly?

A common error that occurs in Google Analytics is a large amount of self-referrals showing up in the traffic source report i.e. the top referring site is your own! This can happen for a number of reasons such as certain pages not having tracking codes, use of frames, page redirects or sub domain and/or multi domain tracking not set up correctly. These self-referrals overwrite the original referral information preventing you from knowing where these visitors came from.

To be able to fix these tracking errors quickly before they skew your reports, it’s useful to use a custom alert (Intelligence > Custom report) to email you if there are any large daily increases in self-referrals.

Even with this alert I recommend that you add an analytics check into your QA/testing process to ensure no errors or loss of data when an updates are made.

6. Has the redesign affected our search rankings?

With any luck your new site and content will have significant improvement in search engine rankings and subsequent increases in traffic.

But it wouldn’t be a lie to say I’ve seen a big brand push a site live without removing the staging server robots.txt file, which was being used to blocking search engine access, this in turn caused the new site to completely drop out of Google!  Even if large changes in rankings like that don’t happen, its useful to monitor organic traffic after launch to be aware of decreases or increases in visits (as well as changes in rankings using this GA filter).

If you do see drastic falls in organic traffic I’d recommend not reacting unless you’re absolutely sure you know the cause, rankings tend to jostle around a little after a redesign and you could end up changing something for the worse.

7. Tracking 404 pages

When launching a new site it is important to ensure you redirect old pages to new pages, however, errors can easily occur with redirect rules causing users to land on 404 pages.

Tracking visits to 404 pages in Google Analytics is easy, simply add the tracking code to the 404 page template and ensure the page title has 404 in it. Then to see which pages are missing navigate to Content > Content by title > find and click the 404 page, this will show you a list of page URLs that have been requested but not found. Unfortunately a custom alert does not work here as you cannot have the page title as an alert, therefore it is important to check this report regularly after launch. You can also use Google webmaster tools to check 404 pages but I prefer the additional traffic insights Google Analytics can give.

Please share in the comments any other reports, filters and advanced segments that can help during a redesign.

Post from: Connect

7 ways Google Analytics can help when redesigning your website

(click for a larger image)

Above we looked at how the top six cinema chains rank for search in each location. As you can see from the graphic above, there is no one outright winner with different brands performing well in each city. In London, Odeon took just over 60% the natural search cinema market, while in Edinburgh My Vue took 70%.

(click for a larger image)

Next we can see how many people are searching each month for cinemas at each of the 10 cities we looked at. Perhaps unsurprisingly the more people living in that city the more people are searching for cinemas at that location. For smaller cities such as Nottingham people are quite straightforward in the ways they search, searching using phrases such as ‘cinemas in nottingham’, ‘nottingham cinema times’ or ‘cinema listings Nottingham’. While in larger cities such as London they are more specific in their search ‘cinemas central london‘, ‘north london cinemas’ or ‘kings road cinema london’

Certain sectors are highly location specific (e.g. the hospitality sector) and as such their online strategies will need to be different to areas were location is not as crucial (e.g. Insurance). Search is becoming increasingly personalised and in sectors like cinema and movie search mobile is growing in importance. As a brand it is important to know how people find your product, who your competitors are and their strengths and weaknesses. Online all this information is easily quantifiable, much of it from free data sources therefore there really is no excuse not to be well informed when planning your online marketing campaigns.

Post from: Connect

Its summer, time to go to the movies…

Methodology

In order to perform this analysis a two week benchmark period and a test period were set up. The periods were selected to be as similar as possible, both Monday – Sunday periods, however some seasonal variation was inevitable, therefore we analysed the traffic for non brand terms for the two periods to see if there was any change. The results showed a 5.1% uplift in the second period over the first, therefore a 5.1% uplift was added to the first period to normalise the results.

Results

The results showed that the majority of visits and revenue from the paid campaign moved across to natural search. When we look at the top 20 brand terms (by visits) we saw that 98.5% of the traffic received previously now went to natural search (100% would represent no loss what so ever).

Conclusions

There are many factors that need to be considered when deciding whether to bid on your own brand terms, such as the relative strength of your brand (e.g. what is the propensity of consumers to click on your brand), is your brand also a generic term (e.g. cheapflights) and how well are mis-spells covered in natural search. Google has become better at correcting spelling mistakes in search queries but many do still fall through the net.

In the test we ran we found that the poorest performing brand terms, the ones where the lowest percentage of clicks were taken up by natural search were terms that had been misspelled and brand URLs. In these cases we found competitors were aggressively bidding on popular misspellings, therefore it made sense to maintain the brand campaign where competitor activity is high.

Additionally when the brand campaign was turned off click through rates fell substantially, this could potentially have an affect on the quality score of the account increasing the cost of the generic PS campaigns. Therefore the overall cost of turning off the brand campaign is not just the potential lost visits/conversions but also the potential increase in cost of the generic PS campaigns.

In the past other pieces of research have suggested an overall increase when combining both paid and natural search however in this particular instance we found that natural search took the majority of the clicks lost in paid. This test was only for one client and for one vertical and results will vary between sectors. It was interesting to see the role competition plays on the brand terms and click throughs. In this instance for a strong brand where there was no competition, turning off the brand campaign and letting natural search pick up the slack made sense, but if there is fierce competition around your brand you would be wise to keep bidding on those terms.

Post from: Connect

What happens when you stop bidding on Brand?

Social media is growing quickly in popularity and shows no sign of slowing down. Facebook the biggest social property in the UK currently has 400 million users worldwide and is growing quickly.

So Social Media is big, but how do we make money from it? Search engines found a way to monetise the traffic they receive by providing targeted ads based on users queries. When users search they reveal their intent and this allows search engines to match adverts to those queries which allows them to charge for that space. The trouble with social media is that people don’t type in queries and therefore reveal intent, they just browse and interact and comment and chat. Social has yet to find an effective way of monetising all the traffic they’re getting, but could all this be about to change?

Facebook recently released its Open Graph platform which allows content publishers to easily integrate Facebooks social elements such as the Like button onto their site. There has been a lot of discussion regarding this new feature over the last few weeks. People have suggested that this could change the web, potentially challenging links as the primary way to index and rank websites the discussion of which is outside the scope of this post, but what I will talk about is all the data they are collecting.

Search managed to monetise itself by delivering highly targeted ads that people clicked on, now the data that Facebook is collecting can help them do something similar. As people move around the internet and click on all those ‘like’ buttons, facebook is building up a database of user interests, the sites they visit, where they spend their time, how often they come back and what they like. Using this information Facebook could know for example that I visit a lot of technology sites, that I have read a lot of articles on the new iphone 4, that I have posted links on my friends Facebook wall about the new iphone, visited carrier websites to check out pricing and even mobile phone recycling website to see how much I can trade in my old phone for. Using this information they can serve up ads offering me the new iphone at a discounted price, offer to buy my old phone or even tell me which of my friends are also thinking of buying the iphone 4 (adding an extra dimension to my purchase decision).

“(your friend) John just bought a new iphone 4 and traded in his old phone for cash, click here to read John’s review of the phone and get 10% a new iphone today”

Social has the potential to engage at a more emotional level than search has ever been able to. It can leverage personal relationships, psychology and add interaction to encourage clicks, something search cannot do. One of the strengths of traditional media, and something that search is not great at is branding campaigns. In search Chanel can bid on the term ‘perfume’ using sponsored ads but that ad won’t have the same emotional and brand building impact as a full page glossy ad of an attractive young couple in a sensual embrace with the perfume in the foreground.

With Social Media you can create immersive and engaging brand experiences. Imagine if a company such as Coke was to create an online experience as social and engaging as Farmville, the hugely successful Facebook game that has more players then there are twitter accounts. In such an engaging environment users would spend hours and hours, playing the game, interacting with their friends all within a world Coke created. Players could collect in game credits slowly by performing tasks, or quickly by buying bottles of coke in the real world. Collect enough points and you can win prizes. Products could be advertised and purchased right within the game. All the while users are living in coke’s world, subject to their brand message, they are building brand awareness, positive sentiment and brand loyalty, this is something than search engines cannot offer. Search is not emotional or engaging, social is and that’s what builds brands.

Social has the potential to be a much bigger market than search. It has emotion, engagement, detailed user data and friend networks. You can leverage social status among friends, encourage competition and impulse purchases. This is the space brands need to be in, it is no longer a question of if, but how.

So is Search dead? No. Search currently outperforms Social with respects to driving sales and revenue but the tide is changing and I think the next 12 months will be an interesting time in this space.

Post from: Connect

Social Media bigger than Search in the UK – Should Google be Worried?

An interesting analysis is to see a breakout percentage by number of terms used across different countries:

Source: iCrossing -  Trellian data / Forrester

Nordic countries for example have a large percentage of one word search over 50%, compared to American geographies where the percentage is for less than 40% using one word. One can think that less number of keywords means or has something to do with the language itself used in the country. Conversely,  one can immediately see that this hypothesis is wrong. If you compare UK and US or CA split by number of words, this differs completely in percentages despite of these countries tending to use the same language structure and share a lot of vocabulary terms. Another good way to check this is checking Germany as an example as one can expect -one word %- to be in large proportion as Germans use very long word construction that can include adjectives and nouns.

A second interesting analysis is to check if there is any potential correlation between internet penetration and number of words that people use. The hypothesis behind is that one can expects that more mature markets with internet usage have evolved in search and therefore  tend to use more long-tail terms. Again, having a look to the graph above, there is no substantial theory behind this as Nordic countries with heavy internet penetration have low rates for 3-4 word searches.

Trying to have a more detailed analysis of how the number of terms used impact organic and paid search results, one arrived to different interesting conclusions.

If we have a first look to organic, the graph below shows the percentage of searches that generated a click on one of the search results. Clearly 3 and  4 words show the largest percentage of searches that generate a click on the search results.

On the contrary, if we have a look to paid search, the graphs below shows the importance of 8 , 9 and 10+ words in generating a click. This shows that paid search is a more targeted technique and therefore provides higher relevancy.

In conclusion, seems that the number of words that people use when they search is still important and of course matters particularly across different geographies.

Post from: Connect

The number of keywords used in a search still matters

Facebook has published a Gross National Happiness Index where you can track levels of happiness across different countries. Users rate their happiness as positive or negative. I found this at least interesting but of doubtful methodology although 400 million of Facebook users is a fair sample.

Gross National Happiness is the difference between the positivity and negativity scores.The model is taking into consideration the words used in users’ status updates breaking them out by positive or negative words, for later assess that the day as a whole is counted as happier than usual.

Additionally, the model has been calibrated differently to ensure consistency for different countries which eliminates effects due to differences in the countries’ population and language use.Computers do the calculations.

The graph below shows a positive and negative word use in status updates which covaries with self-reported satisfaction with life. Also note that the graph shows peaks and valleys on days that are culturally and emotionally significant.

Some interesting insights, Argentina has a peak for the International Woman’s day and Puerto Rico is on the contrary a country with large negative fluctuations. For the UK, anything surprising as Christmas is still the happiest day.

It would have been great to have had Greece in the list of countries.

Now we only need to understand if there’s any correlation between this graph and shopping patterns. A good source of insights during the next Football World Cup or another important sports events.

Maybe Facebook will bring in the immediate future models allowing partners to check automated brand engagement (done by computers) and how happy consumers are with a particular brand or service based on people’s status updates.

Post from: Connect

An unobtrusive model of “gross national happiness”

“As people spend increasing amounts of time conversing on social networks, monitoring what they’re saying about your brand is crucial. So what tools are there to help you listen in?”

Here’s what I said:

‘Another popular paid-for product is Brandwatch, favoured by Mark Higginson, Head of Social Media at search agency iCrossing, for being one of the most cost-effective solutions available. He says iCrossing tweaks the network of sites crawled and reported on to match its sector knowledge, combining this with analysis. ”We call our quantitative and qualitative research a ’stories and numbers’ approach. It’s through this narrative we ascertain what content to create for which audience and who it’s best to approach in those networks in order to gain the greatest share of available attention.”‘

It’s interesting to read what other people had to say, particularly those that favoured Radian6. They’ve revamped their dashboard since I last tried it out; all well and good but I remember being pretty unimpressed with the quality of the actual data collected, although given how quick they are at responding to blog posts about them there must be something to be said for it. Andrew Girdwood of Bigmouthmedia is quoted as saying:

‘… that the data supplied by Radian6 is “pretty rudimentary” and comparable to that from some of the free tools, but its front end is “sexy”. “If you’re preparing social media reports to show someone else, there’s a lot to be said for a package that will wrap it up and can be presented to the board. Sometimes you’re paying for that.’

I would recommend in the strongest possible terms that you spend your budget on human analysis and insight over an expensive tool with weak data but pretty charts. At iCrossing our reports are generally bespoke, the size of client we deal with means we need to be flexible enough to fit in with existing reporting and a self-serve dashboard just doesn’t serve that need. Also, as I say above, this analysis does not exist in a vacuum; it needs to be acted on and that requires it to be substantiated.

There is also an emphasis from other commentators on the importance of real-time reporting. This is actually a distraction from the core features you need. There are exceptions but this level of response is really only applicable to activity around your own profiles, which is simple enough to do. Until online social behaviours become more embedded most people I speak to say that they would find a brand responding to them directly around something they’ve said online that did not specifically occur on a brand-managed property as more akin to stalking than useful customer service. Analytics tools should assist you in gathering useful data in aggregate, from which you can derive intelligent and considered insights. They should not encourage you to make unnecessarily rapid responses.

If you are interested in real-time social results Google launched this feature about the same time this New Media Age article was coming out. It’s available from the regular search results. Just click ’show options’ above the first item on the search results page, then click ‘updates’ in the section entitled ‘all results’. This will give you a timeline you can scrub through as well as mentions as-they-happen. It’s a really nice implementation.

Essentially all analytics tools are search engines with a few useful filters on the data. They have an index of sites to crawl, process the data collected and allow the user to filter the results. It’s a hard job and there’s only one business out there doing it outstandingly well and that’s Google. Despite millions invested, the most intelligent engineers and an ever-expanding cloud of hardware even their results have significant flaws. It is a tough task to condense meaning from the vapour of nuance where language is concerned.

Post from: Connect

Social Media Analytics: one size fits no-one

I’ve now updated it so it works on all Google regional engines, as well as Yahoo and Bing including all their regional engines.

It also now has another feature which I thought would be quite helpful; one-click switching of the number of displayed results.

Links are placed at the top of Google and Yahoo, which link through to SERPS of different result sizes for the same search term. This utilises parameters in the search engines URL, and will not change your default setting.

You can install the extension by clicking here in Chrome, and extensions homepage can be found here.

Post from: Connect

Our Google Chrome extension v2 – Numbered search results

They say the simplest ideas are the best. If that’s true, then this Chrome extension is the best one ever built.

Inspired by Ben’s recent post on the top Chrome SEO extensions, I created a simple Chrome extension that will number the organic search results in Google.

This can be useful for quickly checking a ranking without having to count your way down to the result. For the next version, I’ll add Yahoo and Bing, plus some other search result enhancing functionality.

You can download the extension here. For updates, please check back on the extensions homepage.

Any feedback is welcome.

Post from: Connect

Our new Chrome extension – numbered SEO results