It’s not been too mad in recent months to be honest, since the arrival of the fabled Bounce Rate measurement in Google Analytics (and it’s common inclusion into almost any Web analytics analysis done today), there was definetly a small panic movement as to “omg omg omg! My bounce rates are higher than 20% and my site is soo b0rked and my sales are flying downhill!”

Well, that really isn’t the case.  Before you go and get too worried about your bounce rates, don’t worry too much.  In fact, as has been said before on the web and as I’m mirroring here – High Bounce Rates Are Not Always Bad. In fact, on some sites…they’re a good thing.

If your site is a blog or information providing site such as Wikipedia, with lovely SEO and lots of nice direct SERP links to topical pages, then you will have high bounce rates and that will be good!  It will mean people are landing exactly where they want to be, fulfilling their need and then leaving.  People searching for information about a topic, store opening times, where to find you, prices, or any other action that can be fulfilled all from one page will be happily getting this and then leaving. Clocking up a bounce for you.  But it’s ok!

If you’re a retailer and trying to lure then into the site with promises of discounts and gold however, then it’s probably not so good.  Stronger call to actions needed and yada yada yada…But just remember, please -

High Bounce Rates Are Not Always Bad – Sometimes They Are Good!

Just a quick note to say a big ‘Thanks!’ to the guys over at Receptional Internet Marketing. They’ve given me free access to a Yahoo! Web Analytics Account, and I’ve added codes for this site to run tracking on.   It’s one of the few (getting smaller by the month!) web analytics tools which I’ve not spent time working in over the past 4 years, so it’s a great opportunity to increase my scope of analytics tools and have a play around with a new toy! Part of me hopes it will be as awesome as Vistorville

I’ve added some initial data, once I’ve got a few weeks to play around with (and another couple of sites) I’ll publish a review of the system with all my thoughts and findings

Implementation

Implementation of analytics can be achieved through two main methods, each with their own advantages and disadvantages – via a page ‘tag’ solution (such as Google Analytics, HBX analytics), or through server log file analysis (such as ClickTracks or Affinium NetInsight).

Page tagging provides a relatively quick implementation, dropping a provider’s specified tag onto pages within the website. At its most basic level, this tag will then record and report back customer data to an online system, which you can access. Google Analytics provides a very good solution for this – more so because it’s free – the only cost to you being implementation. More advanced tags can be added where online purchases take place, and the system can be setup to provide you with conversion rates, bestsellers, and revenue generation across pages – all in all quite a thorough tracking solution.

It does however have its limits in terms of reanalysing data and ‘profiling’ visitors – if a tag isn’t implemented and specific information is required then you can’t go back into the data and reanalyse to dig deeper into the information. Generally this tracking would only be recorded going forward, so it’s important to perform a thorough audit of all the tracking and tagging required initially before implementing page tags to ensure no information is missed.

Server logfile analysis can provide more flexibility and depth of reporting, but can take more expertise and time. Database driven log file tools physically analyse the logs generated by the web server each day, compiling the information into a database which can then be manipulated. These logs are stored physically, so data can be queried and filtered, segregated and compared multiple times – once it is in this database format it can be manipulated as required and across multiple tools – something which is difficult for tag solutions to do. This is particularly useful when doing a detailed site analysis as filters can be constantly changed and re-generated as the analysis evolves and more issues are discovered.

Performance Must be Measured

Any company that trades online should have some tracking implementation at a minimum – in the same way shops and superstores change their layout to increase sales and promote product lines. With good analytics a site can be constantly optimised to drive conversions – without analytics, you will lose market share to the competition.

With so many good solutions now on the market such as Google Analytics, there’s really no excuse. Simple on site optimisations through results analysis can boost online sales and provide some informative insight into how your site is being used. Even if making a few simple changes increases conversions by 0.1% – on 100,000 visitors a month that’s an extra 100 orders @£65 AOV = £6,500) – a fraction of setup AND analysis costs.

In our experience, online activity is generally not being tracked as well as it could be across the majority of online ecommerce sites – and those that do have tracking, have it implemented incorrectly or at a very basic level. Usually in these cases, a small amount of work can provide so much more information and insight into how a website is being used – improving performance and sales as a whole.

Since writing this article, Google Analytics has seen some great improvements.  The addition of segementation for one, has been highly valuable and has brought it up to speed with many other analytics packages, allowing detailed analysis of specific visitor types to be analysed.  It’s such a useful tool I can’t recommend it enough – if you don’t have it on your site, implement it now!  For more info on the page tag vs log files debate, see my previous article.  Also, since doing this, HBX to my knowledge has ceased to exist, and has since been bought up by Omniture.  Expect to see some big developments there; it wasn’t a groundbreaking tool but was very straightforward and simple to use.

The morale of the story? If you’re in the business of having a website and using it for any customer/visitor activity – get some tracking on it! Would you open a shop and not be aware of how busy the shop is? How many sales you made? How popular it was? Doubtful!

Previous Blog Post – What to measure Online & How – Part 1

When looking at web analytics, there is always the face-off, the showdown – which tracking method of the two to use should I use? Logfiles or Pagetags?

In recent years, logfiles seems to be getting less and less attention, less dedicated software solutions & less talk.  Most of big players now are all using Javatag solutions, with little to no mention of any logfiles.  Several years ago, this was a different story. With Javatags in their infancy, logfiles were a readily available method of tracking a site that anyone could work with, and there were some excellent software packages out there to analyse them.  The most widely recognised one left to my knowledge would be ClickTracks.

Outlined below is an overall comparison between the two methods of tracking.  There is no right or wrong one to choose, but their maybe a preferable method of tracking for you, based on your experience, setup, & resources.

Log Files

When you load a web page in your web browser (Internet Explorer etc.), a request for the page is made to a web server; the server then returns the page. Whenever the server does this and any other task, such as loading images or Flash content, it records the actions it takes in a log file. The log file stores data on who requested the page, what site referred them to your site, what time they made the request, what page/image/object was requested etc.

Logs can then be thoroughly processed through various pieces of software, which allow you to extract relevant information about your site, and to manipulate and filter this information.

Page Tags

Using a piece of JavaScript code in each page of your website allows tracking through software. Every time the page is requested the JavaScript automatically runs in the web browser and sends information to a log file or database stored on a remote server which can then be processed to reveal relevant information.

Page tagging also allows access to capture user data such as screen resolution, screen colour depth and the java version they are running (allowing you to see how user friendly your website is), etc.

Comparison

There is no right or wrong method, just simply two methods of measuring – different methods may suit different requirements of the user. Below is a simple comparison.




As you can see each has advantages and disadvantages, combining methods can be used, but results from these are not really comparable because of the differences in collection.

More and more vendors are providing flexible solutions to be able to filter and manipulate pagetag information in the same way that logfile analysis can be. It’s been a while coming and pagetag solutions have been playing catch-up, but they are getting closer. Coremetrics Explore, for example, allows detailed segmenting & filtering across endless data fields and reports, albeit in a slightly long winded way. Unfortunately, you are still tied into a vendor contract with no access to the physical data itself, but you do get a detailed pagetag solution with lots of extras.

When conducting any online analytics the most important aspect is to ensure all data being analysed is derived via the same method and is therefore robust and comparable.

Last year I had an article published in Catalogue e-business magazine (July 2008) under the ‘ask the experts’ feature; the topic being:

“What ecommerce marketers should be measuring & how to do so”

Over the course of this week, I’ll publish the content of that article here in two parts, with some extra comments and observation added.  It at a quite basic level, but provides a good introduction to measurement & analytics for marketers new to the world of online.

Part 1 – Why Measure?

As is being realised more and more in today’s technology driven, fast paced business world – the web can be a highly effective marketing mechanism. It allows seamless communication and interaction with customers on a personal level, allowing companies to market to their customers in new ways. Not only is it effective in its execution of campaigns, but an element often overlooked by many companies new to this form of marketing (or even in some cases to those who have been doing it a while), is that with the right implementation, practically every aspect of its performance can be accurately measured. This in turn, gives marketers a robust feedback loop to their campaigns – the likes of which has never been seen before.

Online marketing is one of the fastest growing marketing sectors with more and more companies leveraging to this form of marketing. In the UK alone we spent £4.2 billion to July 2007 compared with £2.34 billion in the previous year . It is also becoming an increasingly competitive space, and as the playing field is level, even smaller companies with the right strategy and implementation can outperform some of the bigger players online. The number of households online is growing (61% of UK households in 2007 – 7% increase on 2006 ) as well as the audience widening (ABCDE ) – providing essentially a consumer ‘playground’ for marketers to test new ideas and campaigns. Unfortunately with this great opportunity also comes many potential areas to get it wrong – all of which may have a negative impact on a business.

Increasing competition means that pay per click advertising becomes more costly, natural search and affiliate marketing become more important – so we must work smarter not harder. Analytics is fundamental to any campaign; ensuring that whatever media investment we are making, it is the right one.

Establishing Measurements

In terms of measuring, with all the options online gives, marketers must first decide their key ‘primary’ measurements – those that are most important to them, their relative market, business and growth strategies.  These measurements then give marketers and the team as a whole clear direction.  Are they trying to migrate more offline customers to the web (potentially via web incentives and promotions)? Increase average order frequency and encourage repeat visits through loyalty programmes? Campaigns can be performance measured and controlled throughout the process, ensuring the optimum response is achieved and adjusted to compensate for.  These are direct marketing principles effectively applied to a retail environment 24×7.

At its most basic level, measurement must include key top line metrics.  Overall visitor numbers, conversion rates, frequency of purchase, time online, and page information (interest, entry & exit points) can provide enough information to understand how a site is performing, and also establish some KPI’s.  These can then act as a baseline going forward, and monitored regularly based on changes to a site or marketing activity.  Entry and exit points can give a good insight into where visitors are landing on your site through search, and also highlight area that maybe potentially losing too many visitors.

One thing to remember – measurements need to be decided before undertaking improvements, campaign activity etc.  An initial baseline needs to be set and a control measure taken before anything is undertaken in order to measure improvement!

Metrics you should be measuring (at a minimum):

• Unique Visitors
• Conversions (could be orders, sign ups, enquiries, brochure requests etc)
• Online Conversion Rate
• Sources of Traffic (Referrers)
• Referring Keywords (What people are entering on Search Engines to find you)
• Key points visitors Enter the site
• Key points visitors Exit the site

At the advanced level, key metrics can be filtered and manipulated as required – how long visitors spend online before a purchase, optimum number of product views before a purchase, the common customer journey through the site (and performance analysis of each stage), performance of onsite keyword search and other functionality, detailed analysis of the purchase process  – even going so far as to measure which precise form field was last filled in before a visitor left the checkout process.  All of these metrics can be used to keep tight control over things like bounce rate and cart abandonment, and ultimately ensure purchase conversion is at maximum.

In any recruitment campaign our objective is to get that ‘first order’.  Consequently our strategies include landing customers directly onto product pages in order to minimise the number of clicks to purchase and checkout.  These customers no longer enter at the homepage but throughout the site.  A comprehensive analytics setup is the only way of determining (in the event that the campaigns don’t work) as to why they don’t work.

Once these metrics have been established, more detailed marketing measurements can be taken into account and applied – segregating visitors by channels (direct, advertising, natural search, email, etc) and measuring performance of these groups individually.  By looking at these different channels, a site can be analysed specifically for a certain type of visitor – you may find for example that ‘direct’ customers find your homepage great and view three or more products, but those who come in via your non-brand natural search content suffer high rates of drop-off after one or two pages as their expectations are not being met.

Part 2 coming soon…