Web Analytic Matt

What Can I Do With a Web Analytics Tool?2
Matt Hopkins posted in Web Analytics on October 29th, 2007

If you're new here, you may want to subscribe to my RSS feed. Thanks for visiting!

Web analytics tools are big!

Nearly all have the basic reports like keywords used by people coming from search engines, a list of referring sites that drive traffic to your site, the browsers people have used etc. Then there are the tools that allow testing via A/B analysis, bid managment, engagement scoring and multi-dimensional visualization of data.

So web analytics tools provide a mass of data and sometimes nice ways of visualising that data. But what can you do with this data that makes it worth while spending a large amount of money and effort on such a tool? After all specific tasks can be scripted to provide information so you need to ensure that you make the best use as possible from your web analytics tool.

The Basics
The first thing to do with any new web analytics installation is to track your advertising spend and make sure you are spending money in the right place. This is really easy to set up, it can be as simple as adding an extra parameter of data to a landing page, for example.

Original: http://www.abc.co.uk/landingpage
After Tracking: http://www.abc.co.uk/landingpage?keyword={keyword}

The above example will take the dynamic keyword generated by a Google Adwords ad and place it on the landing page allowing you to compare organic keywords with paid keywords to find keywords that convert visitors and make your Google PPC ads more effective.

You can track any paid advert as long as you can give it a unique landing page. If you think this tracked landing page looks ugly then simply create a vanity landing page like http://www.abc.co.uk/promo which will 301 redirect to http://www.abc.co.uk/landingpage?keyword={keyword} or vice versa.

Improving the effectiveness of your ad campaigns may save between 10% and 20% of your ad spend if you did some sort of optimisation, more if this is the first time.

However, unless you’re selling expensive products or services then this saving alone will not help much when convincing your boss to get an analytics package.

Newsletters are the next thing to optimise. If you can capture email addresses and send an opt-in email newsletter then you can do some amazing stuff that will increase your traffic and conversion rates. This is done by using behavioural targetting and the tracking as shown before.

Scenario Time
Imagine the scenario, you go to an online computer retailer and look at some graphics cards, maybe some hard disks and some computer games. You then proceed to sign up for a newsletter with the latest deals from the retailer.

Next week you receive your first email from the retailer and at the top they maybe review or have deals on the latest graohics cards and computer games, maybe a bundle deal that is for a limited period.

The reason they knew what to include in the newsletter is based on the onsite content you ahev consumed on your visit to the website. If the retailer expanded this technique to all their customers then it is likely that as the emails are targetted towards the specific interest of the user, traffic and increased conversion will follow.

Now this scale of behavioural targetting isn’t for everyone as it can take some effort to implement but it will certainly pay for the analytics many times over.

Going along a similar line of though, what if you knew that when someone comes to your site using a specific keyword they normally head for specific areas of your site. Why not make those areas more prominent for these users and make it easier for them to convert in a way that they want to. For example, if you had a stock photography site and someone came from Google using the search term ‘flower’. If you knew that 8 times out of 10, people end up purchasing a picture of a yellow rose when coming via the ‘flower’ keyword, it makes sense to put that picture at the top of the list and so on.

It makes logical sense that doing this will reduce bounce ratre and increase conversion rate. Most importantly the user will have a good online experience and hopefully come back for future stock photography purchases.

In conclusion I guess the real way I see of making the most of your web analytics tool is to use the information it provides to feed your website with the most optimal content in an automated way. Second to this would be to drive the website content creators in which content converts and to make more of it.

If you have any other ideas for using web analytics then please post a comment.

Automating SEO, SEM and Web Analytics2
Matt Hopkins posted in Web Analytics, Search Engine Marketing on October 13th, 2007

There are 2 things that have sparked off this post.

1. The first is that while at work I managed to prove that our SEM strategy is less than optimal but that to fix it on such a large scale would require immense automation.
2. The second was an article I read about a day later reviewing Yield Software which hypes itself as a Google optimizing service

Now, the review is highly critical and pessimistic about the service and I have also read similarly negative reviews of the concept of automating SEO.

However I think that everything to do with optimizing your rankings in Google is possible to automate given a very skilled programmer and a lot of time.

A few years ago when I first came into contact with the theory of online marketing, as a programmer I thought, “why is all this SEO stuff not automated?”. I then later found that much of it is using software like WebCEO and IBP, together with bid automation tools for SEM, would it be much of a stretch to combine these tools, along with behavioural targeting and page optimization tools like Google Optimizer to make a people and search engine optimized site? I don’t think so.

But what about when Google changes it’s algorithm?
As that question states, we know that Google mainly works off one or more algorithms, so surely you can write an algorithm to optimize for Google’s?

SEOmoz wrote a great example of how to test a search engine algorithm on a simple scale. Basically it uses a random domain name with random text and random links, you then modify the elements that you wish to test and as a result you see how well each of the pages perform in the SERPs (search engine result pages). So, if you had a number of these tests up running there is no reason why you could not use the resulting information to feed your ‘worker’ algorithms that optimize your sites.

The only thing that cannot be automatically optimized is written content. Now, you can certainly have huge database driven websites and optimize them quite automatically but if you have a marketing site then chances are you will have to write some content that will appeal to your visitors. You could get an algorithm to write your optimized content first using a complex Markov chain which would be ok for search engines but you would need to ‘fix’ it to make it human readable.

Some more examples of similar things being done by less than white hat SEOs are control pannels where you enter your domain and some scripts create content, link to it from other sites like social bookmarking, blogs and other domains owned by the user. Personally I think that without looking at Yield Software it is difficult to tell exactly how their system works but I imagine its probably a paid version of one of these control pannels fed by testing domains.

So is it black hat?
I don’t know if their methods are black hat, but i’m sure that if it works then the search engines will make using the software non-compliant with their T&Cs, thus making it black hat.

In conclusion
It’s an ambicious idea and I like it, if they have pulled it off then fair play and anyone that doesn’t use them will dissapear from the SERPs, and it will make SEOs very unhappy and very poor.

Managing Social Media using Web Analytics0
Matt Hopkins posted in Web Analytics, Social Media on September 27th, 2007

It’s quite amazing what you do with web analytics because it is basically your portal into understanding your online presence. This means that you can look at the effect of social media, your revenue metrics and even your more subtle engagement metrics.

So what is social media?
Social media is news or content that is created by every-day people, this could be a bookmark of some useful content, a blog like this made of one person’s thoughts or even a video. Really it can be any piece of content that is generated by a user of the internet rather than a regular publisher.

And this social media is good because…?
It comes down to the search engines, they want to serve their customers by providing the most relevant content at the relevant time. To do this they constantly look for new ways in which the internet is changing and new ways of identifying relevant content. And what could be more relevant than what individual people are saying.

For example if the Sun newspaper (a tabloid newspaper in England) wrote about a new form of cheap and renewable energy you might not believe it entirely as it is only one publisher and may possibly not be as trusted as other publishers. But then you have 50 individuals that are blogging and making videos about this new form of cheap renewable energy. Who would you believe?

Personally I would go for the 50 individuals as the news is likely to be up to date, a blog can be updated any time whereas a newspaper only comes out once a day. Also there is safety in numbers and 50 publishers saying the same thing is probably more accurate than 1 publisher.

We know that more and more people are reading blogs for niche information and so the big search engines must do the same. This means that people interested in their online presence need to be looking at using social media to their best advantage as every-day people trust it as do the search engines.

But how do you measure social media?
There are a couple of trains of thought when it comes to measuring social media, some people use things called buzz monitors to track particular keywords or keyphrases that are appearing in certain mainstream blogs or aggregation channels. Some just use their keyword and referrer reports of their web analytics applications to measure the increase or decrease of interest. Obviously you would need to add relevant metrics like engagement or interactivity or time on site etc. to find out if your social buzz translates into the desired on site behaviour.

Another way is to actually use a tool provided by the search engine Google. It is their trends application and this week they made what I think is a major change by having daily updates as opposed to monthly or weekly updates so you can measure the buzz of your project using the search engines. You can see below that social media is a particularly hot topic at the moment, one of the reasons I’m writing about it is it’s relevance at the moment.

So what am I using social media for at the moment?
At the moment I’m creating a bit of buzz around a website I’ve created by using social bookmarking sites, writing relevant quality content blogs around issues my new product can address, creating social groups on relevant topics in Facebook and Yahoo groups, answering relevant questions on Yahoo answers and some other good stuff.

And to measure this buzz i’m using my install of Buzz monitor as well as Google trends and of course my onsite analytics. This way I can see at what stage my buzz has reached and where I need to create buzz in order to generate more quality traffic and in the end more sales.

But remember that social buzz is not the only thing on the internet and your website does not exist in a social vacuum, it relies on other interaction from the usual places.

If you’re doing any fun internet marketing with social media and don’t mind sharing your experiences then please add a comment.

Using packet Sniffing for Web Analytics3
Matt Hopkins posted in Web Log Files, Web Analytics on September 4th, 2007

Firstly a packet sniffer is a really simple application that passively listens to any network traffic that runs through or past a network card. When it ’sniffs’ the network it picks up all the packets for every protocol such as tcp/ip and ARP, it also picks up encrypted SSL packets.

This all sounds very technical and worlds away from anything related to marketing or web analytics so how does it fit in?

Well, using a packet sniffer you can pick up all the packets contained within a HTTP or HTTPS request. If it is HTTPS traffic then you can provide the SSL certificate to the packet sniffer and access the requests in their unencrypted form.

Once the packet sniffer has recreated the HTTP and HTTPS traffic it can then create a log file, similar to one created by a web server. From this you can use your favourite web log analyzer to process the log files and provide you with website visitor data.

So where does packet sniffing fit into the data collection methodologies?

You might already know that the main difference between page tags and log files is that page tag data is collection on the client side whereas log files are generated on the web server. Packet sniffing also resides on the web server or at least the Local Area Network (LAN). This means it has the same problems as log files with proxy caching and so is likely to be less accurate than page tags.

But there are advantages, packet sniffers pick up every piece of tcp traffic including form data that has been sent using the POST method and all packet sniffer applications will output that data. For technically minded web analysts there are loads of performance statistics about the network that are also output to the log files.

Another extremely useful aspect of packet sniffers is te ability to amalgamate data from multiple web servers into one log file. For example, lets say that a large content provider has 20 servers that are load balanced and in front of them there are 10 proxy servers. If we use standard log files then we need to either use the proxy logs assuming the proxy servers are all on the same platforms and can be configured correctly to output the required information, or cluster the 20 server log files during analysis. Using a packet sniffer in front of the proxies we can pick up all of the data from one point and because it uses passive sniffing it will not slow down the network traffic.

In any other situation I would suggest page tags or log files depending upon your preference. If you are currently using a packet sniffer(like Clipen) in your analytics environment I would be interested to hear of your experiences which you can detail in a comment below.

First Party Cookie Confusion5
Matt Hopkins posted in Web Analytics, Visitor Segmentation on August 1st, 2007

Cookies in their own right are really simple little things, the idea being that every time you view a web page on a website, you will be given a code to store on your local PC. The next time you come back to that site it can read your stored cookie and so count you as the same visitor who has returned.

So where is the confusion?
Well, the confusion comes when you start talking about first party and third party cookies and how they are treated differently by web browsers.

A first party cookie is a cookie that is given to the website visitor by the same domain (www.domain.com) that the web page resides on. Whereas, a third party cookie is one that is issued to the website visitor by a web server that is not on the same domain as the website. I’ve made a diagram below which shows how the first party cookie and third party cookie differ.

So when might a third party cookie be used?
Generally third party cookies are issued when are third party is interested in tracking your website visitor traffic, this could either be by a banner advertiser who places a number of banners on your site and wants to know how many times it has been requested, or it could be a third party hosted analytics vendor that issues a page tag for each of your pages that forces a cookie on your site.

In the last situation, where an analytics vendor issues a cookie through a page tag the cookie is seen as a third party cookie because it is being generated by the analytics server which is having the tracking 1×1 invisible gif image requested from it by the page tag. It is however possible to have an analytics cookie issued by the third party vendor but still look like a first party cookie.

There are 2 ways of achieving this:-

1. Create a DNS alias for third party analytics server so that it looks like it is actually part of your domain and so anything issued by this server because 1st party (including cookies)
2. Have the Javascript page tag create a cookie at run-time and then pass the cookie value back to the analytics server so the cookie is created within the page and so becomes a 1st party cookie.

The obvious advantage of the DNS alias option is that you can have a smaller page tag which is quicker to load, however the cookie making page tag has an advantage over the DNS alias because no structural changes need to be made to the site’s infrastructure and the implementation of the tag should be more straight forward.

So, in the end you want to aim for a first party cookie as these are typically blocked by fewer browsers than third party cookies. To give you an example, I did a test a few weeks ago using on a site using a third party cookie and measured that over 70% of the cookies were being blocked. After a similar test using first party cookies only 30% were being blocked. This shows that although cookies aren’t as accurate as we might all like, all cookies are not created equal.

My First Engagement0
Matt Hopkins posted in Web Analytics, Visitor Segmentation on July 24th, 2007

No, i’m not getting married just yet.

But I have just released my first white paper which focusses on website visitor engagement and it follows on nicely from my orginial post on engagement scoring.

Let me know what you think.

Web Analytics Jobs0
Matt Hopkins posted in Web Analytics, Web Analytics Jobs on July 9th, 2007

Curious about the current state of the web analytics market to see if we in the UK are getting close to the US I did a quick search on Google to find this great site on web analytics jobs.

Another thing to mention is that there is currently a web analytics job survey happening which is organised by Anil Batra. I think this is a great idea so we can all learn about how this market is maturing and allows career planning for the future in terms of the skills required.

What is Visitor Segmentation and How Will it Help Me?0
Matt Hopkins posted in Web Analytics, Visitor Segmentation on July 8th, 2007

When looking at the topic of web analytics you will no doubt run into something called visitor segmentation. This is really just looking at different behaviours of the visitors that come to your website.

I’ll explain it a bit further by using the analogy of a TV programme we have here in the UK called Big Brother (BB). Personally I’m not a great fan of the show but hopefully it’s something that people can relate to.

In BB there are a number of people that live in a house for a period of 8 weeks or so, this house is covered with cameras so that everything that happens in the house can be monitored. In the same way a website can record everything that happens when someone is on the site using page tags or web server log files.

Within the BB house people move from room to room, they chat to each other and group together, they might swim in the pool or eat some food. All of this behaviour is tracked by the house cameras and then analysed later by psychologists and the public. On your website you can identify which pages people go to, whether they fill out a form or even if they play a Flash game and get the highest score. This site activity is then later analysed to identify behaviour patterns, the most popular types of content among other things.

So what’s the point of this analysis?
The result of BB is that Endemol, the programme creators earn money from a variety of sources including public phone voting, advertising and television appearances. Also the main purpose of commercial websites is to generate revenue by interacting with website visitors and getting them to part with their cash in a variety of ways including purchasing products, services or clicking on advertising.

And where does visitor segmentation come into it?
Well, for BB they need to identify people within the house that do similar activities be that talking on the sofa or eating all the time. Then after grouping these people they can identify which behaviours are the most popular with the public and cause Endemol to generate the most amount of money. Once this has been achieved they can plan what type of people they need to recruit for next year’s show. We are now on BB 8 and we can see that the housemates have become more and more irritating with each iteration of the show, but maybe that’s what generates interest in the show and so creates more revenue.

And on your website you can group visitors by their online behaviour like which content types they consume, whether they comment on articles or come back to the site on a regular basis. Once you have identified these groups of people you can then see which generate the most revenue for you and attract more visitors like this via your marketing campaigns, or maybe create more content that satisfies your revenue generating groups.

So in essence visitor segmentation is about group people who act in similar ways and then if you can make one person in each group happier then you can hopefully please most of the people in each group and ultimately generate more revenue/interest/brand awareness. Of course it is never that simple as people are notoriously unpredictable but habits do form over time and so it is possible to capitalise on these.

Happy segmenting!

What Makes a Good Web Analytics Consultant?1
Matt Hopkins posted in Web Analytics, Consultancy on June 22nd, 2007

I have been a web analytics consultant for some time now and during my consultancy engagements I have learned a lot which I thought would be quite useful to someone wanting to enter into the market as a web analytics consultant. Recently I have seen the web analytics market explode with independant consultants, mainly US based but I know that there is certainly a growing demand for independant consultancy.

What Makes a Good Web Analytics Consultant?
In my opinion a good consultant needs the following skills:

• Be able to listen - this is a key skill that will save you many hours or back tracking, just listen to what the customer wants, not what you think they want. Assumptions are the root of all evil.
• Take into account other people’s suggestions - being thrust into a company that you don’t know means that you will need to take other suggestions on board. you may know best practises, but company employees know their business and are more likely to understand the ramifications of major process changes.
• Be able to take control - as a consultant people look to you to lead and so you must be able to manage projects and small teams of people when the time calls for it.
• Communication to all levels of people - being able to speak to someone on their wavelength is a great ability, whether they be top level management or at the day to day operations level.
• Present - web analytics consultants must always be able to present themselves, their company and most importantly their solution in a manner that is well understood by the entire audience. Rarely simple task.

These are my thoughts on what it takes to be a web analytics consultant, but for general consultancy take a look at another great article on how to be a consultant.

Web Analytics as a Marketing Tool: Scorecards0
Matt Hopkins posted in Web Analytics, Visitor Segmentation on June 16th, 2007

We should all know that to do web analytics for web analytics’ sake is not a good thing, all those numbers, a few words, but without meaning they are pretty useless.

As a web analytics consultant I like to find ways in which web analytics data can be used in new and interesting ways to assist in making better business decisions.

This is where I introduce the concept of scorecards, banks and lending institutions use a method called credit scoring to identify people who they consider to be eligible for receiving credit. The way this scoring works is to take a number of factors about each person, such as their gender, age, employment status and where they live. They then apply a value to each factor, so if the person is below 25 years old then they get a score of 1 for age. If they are employed and have a salary of over 30K pounds then they may get a score of 15 for employment status. Add all these scores together and you get a scorecard per person.

Each person’s scorecard then has an overall score which determines how eligible they are for credit and how much credit. Now, currently in web analytics we have the concept of visitor segmentation which takes different visitor behaviours as differentiating factors but I’m not aware of a product that allows scoring of these factors in a similar way to scorecards.

So, what is the benefit of scoring website visitors and what should we score them on?
In my opinion the main thing website visitors do is view pages, or on web 2.0 sites they may trigger events. So let’s say that we apply a score to certain pages or events. If we use the example of an e-commerce site where books are sold, a search result page could have a score of 1, a book description page may have a score of 2 and the sign up page for the latest books email newsletter can have a score of 5.

Now when we take a site visit that includes 2 searches, a view of a book description page and a sign up for the newsletter the total score for this visit is 9. If a visit consists of only 4 searches then the total score is 4.

In this example the same number of page views were present but because we score each of the page views, we know that the initial visit was worth more than the latter visit. This also means that we can group visits together by their score and find out commonalities between high scoring visits and also low scoring visits.

Also there is no reason why we can’t use this technique to filter out website spiders when using log file analysis, because spiders will want to visit every page so they will have outrageously high scores.

Whatever you decide to use this technique for, you will need to decide which pages to score, what the scores will be for each page and the thresholds for visit/visitor scores you are interested in.