Improving web marketing effectiveness

Introduction

Do you know what proportion of your target audience actually visits your web site? Or how many visitors go beyond the home page to take the actions you desire and how many convert from visitors to customers?

This article outlines approaches answering these questions. It shows how metrics can be used to identify changes to your online marketing which will increase visitors and convert them to the required marketing outcomes.

We also review the types of tools available for analysing web site visitor data since these have changed significantly over the last year or so. Finally we outline approaches to collecting primary data through online surveys and focus groups.

Key metrics for assessing Internet marketing

Before examining the tools and technique for collecting metrics, we must define the data that needs to be collected to improve the performance of online marketing. As explained in the previous article, organisations should ensure that the data collected covers all aspects of Internet marketing performance. It can be suggested that a comprehensive framework will cover these 5 diagnostic categories and answer these questions:

1. Business contribution
How does Internet marketing contribute to the bottom line? What is the online revenue contribution (direct and indirect), costs and profitability?

2. Marketing outcomes
How many marketing outcomes are achieved online? For example, what proportion of leads, sales, service contacts occur online? How effective is online marketing at acquiring, converting and retaining customers?

3. Customer satisfaction
What are the customers’ opinions of the online experience and how does this affect their loyalty?

4. Customer behaviour (Web analytics)
This assesses how different customer segments interact with web site content and assesses how the actions they take are influenced by usability, design, content, promotions and services.

5. Site promotion
How effective are the different promotional tools such as search engines, e-mail, direct marketing and advertising at driving quality traffic to the web site? Measures include attraction efficiency, referrer efficiency, cost of acquisition, reach and the integration between tools.

We will refer back to these 5 categories of metrics when considering the different tools later in this article.

Primary measures of visitor activity

Web site visitor activity data is used to assess customer behaviour (category 4 in the diagnostics framework). It is recorded using three key primary measures that are widely used for reporting across a time period such as month, week or day:

• Page impressions. One page impression is one individual viewing one page and is sometimes referred to as page views. Together page impressions form a clickstream which can be used to understand online buyer behaviour.
• Visitor sessions. Usually defined as a series of page requests by a visitor without 30 consecutive minutes of inactivity.
• Unique visitors. The number of individual visitors to the site within the reporting time period (it is also highly desirable to measure repeat visitors).

Note that the term ‘hit’ is no longer used by web professionals since it is not a reliable measure for the number of people viewing a page (it is recorded for each graphic or text file requested from a web server, so if a page contains 10 graphics, plus HTML file, this is recorded as 11 hits).

Key ratios of online marketing effectiveness

Although the primary measures are now widely used for reporting on web site visitor activity, to enhance a web site further, it is valuable to calculate additional ratios. Try these key ratios which each give an indication of the value of the online content to the visitor. Online marketers should aim to increase these ratios through time.

1. Average page views per visitor. Calculated by page impressions/visitor sessions – this gives an indication of duration of visit – the more pages viewed, the longer the duration. It is a better indication of content viewing than average visit time since this is skewed by visitors who keep a page loaded after they have finished reading it.

2. Average visit frequency. Calculated by visitor sessions/unique visits – indicates average number of visits within a time period.

3. Repeat visit percentage. Calculated as the percentage of visitors who are repeat visitors within a time period.

As suggested by the questions at the start of this article, further ratios should be used to assess visitors’ engagement with the site and conversion to different actions.

Let’s assume that in a given month we attract 10,000 visitors to the home page of a site. However, only 4,000 venture beyond the home page.

The proportion of visitors venturing beyond the home page (40%) and other pages (the bounce rate) should be recorded and design and content modified to improve this proportion – it is often surprisingly low in first generation web sites.

Similarly, conversion to different types of action should also be recorded and reviewed for methods of increasing conversion to the required outcome.

For example, in this example conversion rates might be 200 or 2% completing a registration form; 350 or 3.5% entering a competition and 75 or 0.75% completing a sale.

Adjusting incentives, copy and promotions can all be used to increase conversion rate – these are cost-effective methods of increasing the marketing performance of a site. However, if there are more fundamental problems with the site design or proposition, then these will, of course, be more costly to rectify.

Many of the tools for collecting the primary data described below do not calculate these key ratios, neither do they show time-series.

Yet, plotting the data through time is crucial to understanding which site promotion and on-site marketing techniques work and which don’t.

Download Metrics spreadsheets from http://www.davechaffey.com/Spreadsheets/.

Metrics tools and techniques

There are several methods of collecting primary measures data. These primary measures are used to calculate metrics in category 4 (customer behaviour). Although managing these tools is often the realm of the web master, it is useful for marketers to be aware of the strengths and weaknesses of the different approaches. The main collection techniques are:

1. Server-based log file analysis.

A log file is a feature of web servers that records all requests for web pages including the page requested, and the time and source of the enquiry.

Since log files quickly grow, separate log file analysis software has to be purchased to summarise the data. Webtrends is that with the largest market share most commonly used, but there are many similar tools.

Although these tools present a vast amount of data, finding the nuggets to improve your web marketing, or even to report the primary measures is not straightforward – some monthly reports amount to 100 sides of A4! Server-based log file data is sometimes audited, usually for advertising purposes by ABC Electronic (www.abce.org.uk).

When interpreting log file data care should be taken since there may be major sources of undercounting or overcounting as outlined in Table 1.

Table 1 Inaccuracies caused by server-based log-file analysis.

Dynamically generated pages, generated ‘on the fly’ are difficult to assess with server-based log files.

2. Browser-based analysis.

This approach is a response to the problems of undercounting and overcounting shown in Table 1. This technique assesses the primary measures by recording access to web pages every time a page is loaded into a user’s web browser through running a short script (usually a Javascript program) inserted into the web page. Services that use this approach include Redsherriff (www.redsherrif.com), Nedstats (www.nedstats.com) and Hitbox (www.hitbox.com). These services are usually paid for monthly according to traffic volume. This approach removes many of the sources of inaccuracy and such services are becoming more important.

3. Panel data.

Internet panel data is collected in a similar way to home-based TV panels. Panel members agree to have software installed on their PC that sends data that is collected by the monitoring organisation.

The main providers of this data are Nielsen Netratings (www.netratings.com) and Comscore (www.comscore.com).

Critics note that there may be a problem with panel accuracy since the type of person who agrees to have the software installed may not be representative of the wider population although efforts are made to ensure they are demographically representative.

4. ISP data.

This is an approach provided by Hitwise (www.hitwise.com), a company originating from Australia. Each day Hitwise analyses the networks of a range of ISPs, counting the visits that any website receives from their visitors. The figures from each ISP are then aggregated and analysed to calculate a site's ranking relative to other sites. The key difference with this collection technique is that it is possible to gain data on competitors’ online activity. Although detailed numbers of page impressions or visitors are not provided, the ranking and relative audience share for different competitors within a sector are available, so a bank, for example, can compare its online share compared to its offline market share. This data can be used to answer the first question posed at the start of this article.

Note that great care needs to be exercised when interpreting site activity data as a basis for revenue estimates for forecasting.

As well as measurement of primary data, more sophisticated analysis is increasingly being used to illuminate the behaviour of customers online and to identify sales potential.

These are some of the novel techniques which we will return to in a later article:

• Clickstream analysis and reverse path analysis are used to increase the number of visitors who follow the optimal site path to a desired outcome.
• Visitor segmentation is used to categorise groups of customers according to their demographics and their buying behaviour (time of visit, types of content visited, length of time on site and actions taken).
• Web mining can identify hidden patterns within the web logs which can be used to devise new offerings and revise the site design.
• Real-time optimisation is an exciting approach where the marketer can immediately identify particular types of customer or behaviour and can personalise content accordingly.

Assessing customer satisfaction

Different types of market research can be used to assess customer satisfaction with an online presence (category 3 diagnostics). Table 2 compares different alternatives for collecting market research data online. Some of these market research techniques are best suited to a particular stage of web site development. Their offline equivalents also need to be considered. For example, to explore a new site concept, an offline focus group may be best; to assess site visitor opinions, a pop-up survey may be best; for registered customers, e-mailing details of an online survey is best while accompanied surfing is best for detailed usability studies.

Online surveys are one of the most cost-effective methods of determining visitor satisfaction. Great care needs to be taken to ensure respondents are representative of visitors as a whole. Pete Comley of Virtual Surveys (www.virtualsurveys.com) suggests that three key areas of questions that should be gathered via online surveys are:

1. Who is visiting site? Typical questions concern, Internet experience, access location /speed, demographics/segment and role in buying decision (B2B).

2. Why they are visiting? How often do you visit? Which information/service? Did they find it? Actions taken?

3. What do they think? Overall opinion? Key areas of satisfaction? Specific likes/dislikes? What was missing?

Links

Agrawal, V., Arjona, V. and Lemmens, R. (2001) E-performance: the path to rational exuberance. Mckinsey Quarterly, No 1. 31-43. Online at www.mckinseyquarterly.com.

Defines key measures for business contribution and marketing outcomes with an evaluation of US e-tailers.

Note: This article is one of a series originally written between 2000 and 2003 for The Chartered Institute of Marketing What's New in Marketing Newsletter. Many of the concepts remain valid, but some more recent concepts such as Web 2.0 and social networks aren't referenced.

For the latest on these concepts search my DaveChaffey.com blog site for web analytics articles.