Back

Measurements

WHAT PEOPLE DO

The metrics generated by web analytic tools tell us how many pages were viewed, how many unique visitors there were, the bounce rate on pages, the number of pages viewed, the average time spent on a web site and more such measurements.

It is important to understand that these metrics only confirm the actions that people have taken on a web site.  They do not tell us WHY people have taken these actions.

A simple analogy would be to sit outside two main street stores and count how many people enter each store. This might tell us that 30% of all visitors have entered Store A and 70% have entered Store B.  But these measurements do not tell us why they entered these stores in these proportions.  We could go further by entering Store B and counting how many of each products are purchased.  Once again, while this might tell us that Product X was purchased three times more often than Product Y, the measurement does not tell us why people purchased those two products in those proportions.

It may not be necessary to understand why people took the actions they took.  Measuring the rates of activity may be sufficient information to use for a strategy to improve conversions.   However, if the product lines change or some external factor changes, then past measurements can no longer be used as a guide to customer behavior.  We would have to continuously measure activity and act on the most recent set of results.  This can lead to rapid fluctuations in the data which lead us to constant changes in the marketing strategy.  This may be a suitable strategy to follow when deciding which product lines to promote but may not be suitable when the product line or service remains stable.

The other issue with merely using the numbers is that interpretation is a guessing game.  Did people buy more of Product Y because it was discounted, because it was promoted more heavily, because it is seasonal or because it has fallen into fashion?  The reason for the levels of activity is unclear.  Attempting to use variations of control tests to discover the true reason would be a complex and lengthy process.

And finally any results from the data should be compared with the results that could be obtained from a sample created by random probability.   The two can look very much the same.  A chart of results that at first indicates a "cause and effect" pattern can often be duplicated from a random sample of results.  This is known as a normal distribution.  A set of results deviating from a normal distribution would be evidence that there is a genuine cause and effect taking place.

If we want to know why people have taken the actions they do we need to develop models of social and economic behavior.   The metrics can then be used to confirm whether these models are correct or not.

INCOMING TRAFFIC

The source, volumes and frequency of the web traffic reaching your web site can be easily measured using a web analytics tool.  These results represent the entire external audience that is reaching the web site.  The value of this traffic is determined by conversion and return on investment measurements.

The nature of the incoming traffic is directly comparable to the number and type of people you could measure entering a real estate location.  And yet these online measurements can also reveal the intent behind the visit and the context in which they occur. 

For example, a measure of the search phrase "cheap jewelry" compared with the search phrase "quality diamond jewelry" may allow us to segregate web site visitors into two income brackets.  This is an opportunity to provide different web content to the user based upon their search phrase and implied income bracket.

The volume of traffic to each web page can indicate how popular the content on that web page is relative to all other content on the site.  This may allow us to develop services and products that target this popularity.

The frequency measurements, such as time between each visit and time spent on site, can indicate how well we a web site is engaging with its audience.  This is commonly referred to as "stickiness".  "Engagement" may be a better term since it reflects a lifestyle union between the visitor and the web site. 

MEASURING TOOLS

For web sites a range of tools are available to measure web site activity and perform analysis.  These include web traffic logs, the free Google Analytics, WebTrends, Omniture and services such as Google Adwords to measure conversion rates and return on investments for pay per click campaigns.   The commonly preferred method is to use a page tagging method, such as that provided by Google Analytics, because it allows the activity on cached web pages to be captured.

It is also feasible to build your own measuring tools by creating a bespoke web traffic database and updating it with every page event.

The selection of the right tool to use could be influenced by price or scalability.

Other metrics can be provided from sales databases, payment records, ERM and CRM software products.

DEPARTING TRAFFIC 

Page exit counts on web pages show the location where visitors have decided that they have had enough and leave the web site.  These could be called Departure Pages.  Just as Landing Pages are doorways into the web site, Departure Pages are the doorways out of the web site.

The question most web sites owners want to answer is "Why are the visitors leaving my site?"  The Page Exit metrics in your web traffic reports will highlight those pages where most of your visitors left.  It is these pages that you will need to examine to determine how to keep your visitors.

There is little point of improving the exit rates of rarely viewed pages before improving the exit rates of the most heavily viewed pages.

ON SITE NAVIGATION

Analysis of navigation paths to and from a web page can indicate the entrance and exit points from the page.   Any web traffic tool will provide navigation path analysis and may allow you to view them as heatmaps with click rates over each available url link.  Each entrance point and each exit point will be displayed as a specific web page on your site.

If your navigation analysis is showing low traffic flow through the url links then you may want to consider a redesign of those links or even permanently removing those that are not receiving any interest.

Path navigation analysis can be interpreted in two ways.

a) You can view the entire navigation path itself from one page to another as a goal in itself.

or

b) You can treat the entrance and exit measurements on each page in isolation with no connection to other pages.

The second method is easier to understand because you are just analyzing the behavior on one page.  The first method requires the asusmption that visitors are following a defined path.  This may be true if the pages involved in the navigation have offer few links, such as just Next Page, First Page, Last Page links.

KEY PERFORMANCE INDICATORS

A Key Performance Indicator, or KPI for short, is the name given to any particular online measurement you wish to monitor.  It should relate to an online activity that can be actioned upon  It can be defined as either an incremented value such as 3,927 or as a ratio such as 3.87.  A KPI should be created for each specific online event you want to measure.  The KPI should also be relevant to the person using it and be related to a specific goal that person wants to achieve.

An example of a calculated KPI may be as follows:

Number Of Visitors For Each Organic Search Term
-----------------------------------------------
Number Of Sale Conversions

LANDING PAGES

Any page that is the first page on a web site accessed by a visitor is a landing page i.e. they are the entrance point to the ENTIRE web site.  It could be the home page, an inner page within the site or a page specifically designed as a page accessed in a banner or pay per click campaign.

Examining the web traffic metrics for your primary landing pages is an ideal way to begin to understand how these measurements can influence web page content.  This is because the measurements are easily accessible and you can influence the results very quickly because these pages are already receiving substantial traffic.

For example, you may find that your most frequent landing page incurs a bounce rate of 60%.  So 60% of visitors are not willing to step further into your web site from this landing page.  That is a hefty percentage.  What could be done to erode that bounce rate down to 30% and encourage more visitors to continue on to other pages on your site?

Landing pages are most successful where they are both a) relevant to the link or search phrase that generated the traffic and b) offer further paths that add value to that visit. 

SEARCH PHRASES

The analysis of search phrases is the entry point to the relationship between the visitor and the web site.  The majority of visitors visiting a web site from the use of a search engine are "fire and forget" visitors.  They search, they visit and then the majority of them will return to the search engine.  Their visit is a fleeting opportunity to form a relationship.

The web traffic metrics relating to search phrases show us how the search engines view our web site.  Is that view correct?  Do we want to alter the way they see our site?  Are we trying to capture search engine traffic that is out of our reach?  Are we ignoring potentially valuable search engine traffic by not catering for particular search phrases?  Are we concentrating too much on low value, high volume search phrases, and not paying enough attention to high value, low volume phrases?

RETURN ON INVESTMENT

Calculating an accurate ROI is a fairly complex calculation in most cases.  It involves correctly calculating the total cost of acquisition of visitors, the revenue generated from online sales, the profit retained from those sales, reductions from product returns and costs associated with processing online payments.  A web traffic tool and pay per click price reports alone cannot cover all of those factors.  It would require further data from ERM retail systems and payment systems to correctly calculate the financial gain from online activity.

It is however possible to formulate a revenue estimate based upon the number of sales generated from the cost of acquisition through organic, pay per click, banner and email campaigns.  This would indicate a ratio between the online revenue generated and the total online promotional expenditure.  An estimate could then be made as to the percentage of profit on these sales. 

There are also other benefits which are difficult to measure financially.  These would include increasing brand recognition and building up a database of future sales leads.  These have real value in the longer term but are intangible assets in the short term.

ROI can always be measured in terms of customer loyalty.  Since it is much more expensive to gain new customers than it is to retain existing customers, monitoring and improving customer loyalty is a genuine financial benefit.