Questions for marketers

Key questions for marketing managers related to this chapter are:

●● How do I measure and improve the effectiveness of digital marketing?
●● How much resource do I need to put into managing and improving the site?

Scan code
to find the
latest updates
for topics in
this chapter

Links to other chapters

This chapter should be read in conjunction with these chapters:
●● Chapter 4 describes the development of a digital marketing strategy. The aim
of measurement is to quantify whether the objectives of this strategy have been
achieved
●● Chapter 7 describes issues involved in creating an effective digital experience
including websites and mobile platform, and should be read before this chapter
to introduce the reader to the concepts of website management
●● Chapter 8 describes methods of increasing brand awareness online. It should
be read before this chapter since one aspect of measuring the effectiveness
of digital marketing is aimed at assessing the different online communications
methods.
550 Part 3 Digital marketing: implementation and practice

Introduction

Companies that have a successful approach to online marketing often seem to share
common characteristics. They attach great importance and devote resources to monitor-
ing the success of their online marketing, putting in place the processes to continuously
improve the performance of their digital channels. This approach has been fundamental
to the growth of Amazon, as Case study 10 in this chapter explains about their culture of
metrics.
The importance of defining an appropriate approach to measurement and improvement
Web or digital analytics is such that the term web or digital analytics has developed to describe this key digital mar-
Techniques used to keting activity. A Digital Analytics association (formerly web analytics, www.digitalanalyt
assess and improve the
contribution of digital icsassociation.org) has been developed by vendors, consultants and researchers in this area
marketing to a business, to manage best practice. Their definition of web analytics is:
including reviewing
traffic volume, referrals, Web analytics is the measurement, collection, analysis and reporting of Internet data for
clickstreams, online
reach data, customer the purposes of understanding and optimising web usage.
satisfaction surveys, leads
and sales. You can see this is a ‘catch-all’ definition. How do you think it could be improved? We
think it could reference the commercial aims of optimisation and the measurement of mul-
tichannel usage.
To succeed in a measured approach to improving results from Internet marketing we sug-
gest that there are four main organisational prerequisites, which are broken down as shown
in Figure 10.1 into the quality of the web analytics processes including defining the right
improvement measures and purchasing the right tools and the management processes –
such as putting in place a process where staff review results and then modify their market-
ing activities accordingly.
This chapter is in two parts – the first part is about performance management, where
we review the approach to improving performance through assessing appropriate meas-
ures, tools and the right process to apply them as suggested by Figure 10.1. In the second
part, we review some of the issues involved with maintaining an online presence, looking
at the tools and process for improving different company web presences.

Metrics Tools Analysis Action

Have we selected Can our analytics Are the measures Are the measures
the right diagnostic software collect, reviewed and acted upon?
metrics to improve aggregate and interpreted at the
performance? visualise metrics? right time?

Quality of Quality of
web analytics management
processes processes

Improved channel
contribution

Figure 10.1 Key questions in evaluating process, metrics and tools for improving
the contribution of digital marketing within an organisation
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 551

Digital marketing in practice The Smart Insights interview

Avinash Kaushik, analytics evangelist at Google

Overview and main concepts covered

Avinash Kaushik is Google’s analytics evangelist, well-known for his books Web
­Analytics An Hour a Day, Analytics 2.0 and his Occam’s Razor Blog.

The interview

Q. Some have criticised online customer engagement as being an abstract

­concept that can’t be readily applied in the real world. Can you give some prac-
tical ­examples of how a site owner can apply the engagement concept to get
better results?

Avinash Kaushik: Engagement is a nice goal to have. Create sites that customers
will find engaging and they’ll stick around or come back again or maybe do business
with you.
But that term has been manipulated to a point where it means nothing any more
(or everything to everyone) and is often used as an excuse to not do the hard work of
figuring out what the real outcomes of the site are for the company and the website
visitors.
My encouragement to website owners is to be initially sceptical when someone is
trying to pawn off ‘engagement’ on them and ask the tough question: ‘What do you
­really mean by engagement and how does it specifically apply to my business?’
Secondly, I encourage people to realise that on the most glorious spring day when
the birds are chirping the right song, web analytics tools can measure the degree of
engagement but they fall quite a bit short of measuring the kind of engagement. So
they can report that visitors saw 19 pages on your site (degree) but they can’t tell you
if that was because the visitors were frustrated with your crappy navigation or thrilled
with your content.
People use Google Analytics (or other tools) to measure easily various elements of
the degree of engagement. Perhaps the simplest example is using the bounce rate for
the core landing pages to identify pages that won’t even entice visitors to make one
click! In two clicks you can also get loyalty (recency), frequency, length of visit and
depth of visit to get a solid feel for if visitors are making repeat visits to the site or if
they visit it more frequently and marry that up with content consumption. Doing this
by looking at trends over time is a fantastic way to understand if the site is delivering
value for your customers.
For many ‘social’ websites, website owners also measure the number of people
who sign up and then contribute by writing reviews or comments etc. These are all
­really good examples of (1) measuring the degree of engagement and (2) not confusing
the real metric being measured by calling it engagement.

Q. Which are the best measures or reports you could point to which help market-
ers understand how well an e-commerce site is performing for retention?

Avinash Kaushik: I touched on some of the obvious ones above, the loyalty metrics
(specifically recency and frequency). They immediately tell you if you are acquiring traf-
fic that comes back again and again, and since GA will tell you recency by going as far
back in history as you have data, that is a great way to know when customers come
back (and perhaps also understand why). The other obvious thing to do for shorter
time periods is to look at the trends for percentage of new visits, especially by the
sources of your traffic.
552 Part 3 Digital marketing: implementation and practice

Some retailers want to do retention analysis by looking at repeat purchases. For

this, Google Analytics, like pretty much every tool out there, provides a very strong
complement of e-commerce reports that allow you to segment the data by the types of
purchasers (new or returning) which will help you understand their purchase behaviour
and by applying filters to your data you can dig deeper into sources of traffics, trends
in number of visits, content consumed etc.
This in conjunction with using even simple on-exit website surveys can give you a
great picture of what is happening on your website and where you are missing the boat.

Q. Conversion optimisation. Today there is a lot more talk in large organisations

about using techniques like AB or multivariate testing. How would you advise a
small business owner to set out on this journey?

Avinash Kaushik: This might surprise you, but I am seeing a lot more traction in using
optimisation techniques with smaller companies than with larger companies. There are
a couple of interesting reasons:

1 A/B or MVT is now free with tools like Google Website Optimiser, so you can dis-
pense with RFP’s and all that ‘stuff’ and just go try the tool.
2 Smaller companies are much more willing to try new things and have less politics
and entrenched opinions (and HiPPO’s) that are hard to overcome. This is of course
a bit sad because given the traffic and the sheer opportunities it really is a crime
for larger companies to leave so much more revenue on the table, or the chance to
optimise the customer experience which will improve loyalty and satisfaction.

My recommendations for any company are perhaps similar:

●● Start with A/B testing. In my experience starting simple will ensure that you will get
out of the gates fast and be able to start the critical process of cultural shift with
easily understandable experiments. Then you can move to the 1.8 billion combina-
tion page test.
●● For the highest impact try dramatic differences in your test versions. Trying shades
of blue might sound interesting but the test might take a very, very long time to pro-
vide you with statistically significant differences. But trying a page with only text and
one with text and images might get you on the path to understanding your custom-
ers faster.
●● Run a report for your top 25 landing pages (entry pages) on your site, then look at
the bounce rates for each of them. Pick three with the highest bounce rates, these
are the pages letting you down the most. You’ll win big by testing these first.
●● Have an active ‘customer listening channel’. Remote usability testing, market
­research, customer call centres or surveys (even a free excellent solution like 4Q,
which I helped create with iPerceptions). The best focus points about what is not
working on your site come from your customers (sadly not you) and likewise the
greatest ideas on how to improve your site (and hence test) also come from your
customers. Listen and you will prosper.

Q. What excites you most about potential developments in web analytics in the
future?

Avinash Kaushik: The thing that excites me most is that no one has a clue where this
is all headed. We have no idea what ‘web analytics’ will look like in five years. That is
exciting because there is a tonne of change and growth to come and being a part of
that change is simply fantastic.
There are new data collection methods to come, there are new ways of doing
­superior analysis of data, there is so much more we could do with Artificial Intelligence
 Chapter 10 Evaluation and improvement of digital channel performance 553

in optimising customer experiences, there are opportunities to bridge the various

islands of data (on the web or outside) to create something amazing, there are … it
goes on and on.
They are going to get a lot better about what you should look at. Visualisation is
great and tables are good but what is killing analysts right now is their ability to figure
out, from megabytes and megabytes of data, what is actually worth looking at. Most
tools still simply spew data out relying on the analyst (or the data consumer) to figure
things out. That is a bad strategy, yet most tools follow it.
Visualisation has got a lot better – but I’m not seeing any form of intelligent rec-
ommendations. It’s very tricky and the web analytics companies are too busy copy-
ing other functionality! It’s good you mention that ClicksTracks features though.
Using ClickTracks as an example, I demonstrated how to look at only the data that
has shifted in importance by a statistically significant amount. Your top 20 of anything
never changes, but using this type of report, What’s Changed, you can look at just the
data that really matters. Now it is easier to take action.
I expect all tools to get much, much better at applying advanced mathematics and
statistics to help their users identify where to focus their attention. The other thing
I would highlight as an evolution for web analytics tools is that they are going to do a
lot more than page view reporting on your site. I don’t mean doing clever things like
event logging to measure Web 2.0 experiences – that is cool of course. I am refer-
ring to their ability to measure content no matter how it is distributed (widgets, rss,
etc.) and where it is consumed (websites, feed readers, mobile phones, your home
refrigerator or washing machine!).
Opportunity is, I suppose, what I find most exciting about the future of web analytics.

Performance management for digital channels

To improve results for any aspect of any business, performance management is vital. As
Bob Napier, Chief Information Office, Hewlett-Packard was reported to have said back
in the 1960s, ‘You can’t manage what you can’t measure’. The processes and systems
Performance
intended to monitor and improve the performance of an organisation are known by busi-
management system ness operations researchers as performance management systems and are based on the
A process used to study of performance measurement systems.
evaluate and improve
the efficiency and Today, nearly all organisations have different forms of online presence, but the ques-
effectiveness of an tions highlighted in Figure 10.1 aren’t answered adequately. So, a good starting point is to
organisation and its
processes.
understand the current improvement process and the organisational barriers which prevent
a suitable improvement process.
Performance
measurement system In this section, we will review approaches to performance management by examining
The process by which three key elements of an Internet marketing measurement system. These are, first, the pro-
metrics are defined,
collected, disseminated cess for improvement and, secondly, the measurement framework which specifies groups of
and actioned. relevant digital marketing metrics and, finally, an assessment of the suitability of tools and
Digital marketing techniques for collecting, analysing, disseminating and actioning results. We will review
metrics three stages of creating and implementing a performance management system.
Measures that indicate
the effectiveness of
digital marketing activities
integrated across different Stage 1: Creating a performance management system
channels and platforms
in meeting customer,
business and marketing The essence of performance management is suggested by the definition for performance
objectives. measurement used by Andy Neely and co-workers of Cranfield School of Management’s
554 Part 3 Digital marketing: implementation and practice

Centre for Business Performance. They defined performance measurement as (Neely

et al., 2002):
the process of quantifying the efficiency and effectiveness of past actions through acqui-
sition, collation, sorting, analysis, interpretation and dissemination of appropriate data.

Performance management extends this definition to the process of analysis and actioning
change in order to drive business performance and returns. Online marketers can apply
many of the approaches of business performance management to digital marketing. As
you can see from the definition, performance is measured primarily through information
Effectiveness on process effectiveness and efficiency, as introduced in Chapter 4 in the section on objec-
Meeting process tive setting, where we noted that it is important to include both effectiveness and efficiency
objectives, delivering the
required outputs and measures.
outcomes, ‘doing the right The need for a structured performance management process is clear when we examine
thing’.
the repercussions if an organisation does not have one. These include: poor linkage of
Efficiency
Minimising resources or
measures with strategic objectives or even absence of objectives; key data not collected;
time needed to complete data inaccuracies; data not disseminated or analysed; or no corrective action. Many of
a process ‘doing the thing the barriers to improvement of measurement systems reported by respondents in Adams
right’.
et al. (2000) also indicate the lack of an effective process. The barriers can be grouped as
follows:
●● senior management myopia – performance measurement not seen as a priority, not
understood or targeted at the wrong targets – reducing costs rather than improving
performance;
●● unclear responsibilities for delivering and improving the measurement system;
●● resourcing issues – lack of time (perhaps suggesting lack of staff motivation), the neces-
sary technology and integrated systems;
●● data problems – data overload or of poor quality, limited data for benchmarking.
The Web Analytics Association (2011) Outlook survey of companies using web ­analytics
gives insights on the specific challenges of performance management for digital marketing.
The top five challenges were:
●● actionability of the data (36 per cent);
●● business decisions driven by analytics (35.3 per cent);
●● social media (34.9 per cent);
●● executive management awareness and support for web analytics (34.9 per cent);
●● failure to take action on the data (31.0 per cent).
The top two issues reported suggest the problems of performance management, taking
­action based on the data.
To avoid these pitfalls, a coordinated, structured measurement process such as that
shown in Figure 10.2 is required. Figure 10.2 indicates four key stages in a measurement
process. These were defined as key aspects of annual marketing plan control by ­Kotler
(1997). Stage 1 is a goal-setting stage where the aims of the measurement system are
defined – this will usually take the strategic digital marketing objectives as an input to the
measurement system. The aim of the measurement system will be to assess whether these
goals are achieved and specify corrective marketing actions to reduce variance between
target and actual key performance indicators. Stage 2, performance measurement, involves
collecting data to determine the different metrics that are part of a measurement frame-
work, as discussed in the next section. Stage 3, performance diagnosis, is the analysis of
results to understand the reasons for variance from objectives and selection of marketing
solutions to reduce variance.
In a digital marketing context, corrective action is the implementation of these solu-
tions as updates to content, design, ongoing marketing communications and CRO. At this
stage the continuous cycle repeats, possibly with modified goals.
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 555

CONTROL
What do we want to achieve?
Who?

MEASURE
Who?
What is happening?
When?
How?
REVIEW
Why is it happening? Who?
When?

What should we do about it?

Figure 10.2 A summary of the performance measurement process

Stage 2: Defining the performance metrics framework

Measurement for assessing the effectiveness of digital marketing should assess the
­contribution of digital marketing at different levels:
1 Are corporate objectives defined in the digital marketing strategy being met?
2 Are marketing objectives plans achieved?
3 Are marketing communications objectives achieved?

These measures can also be related to the different levels of marketing control specified by
Kotler (1997). These include strategic control (question 1), profitability control (question 1),
annual-plan control (question 2) and efficiency control (question 3).
Efficiency measures are more concerned with minimising the costs of online marketing
while maximising the returns for different areas of focus such as acquiring visitors to a
website, converting visitors to outcome or achieving repeat business.
Chaffey (2000) suggested that organisations define a measurement framework or create
a management dashboard which defines groupings of specific metrics used to assess digital
marketing performance. He suggested that suitable measurement frameworks will fulfil
these criteria:
●● Include macro-level effectiveness metrics which assess whether strategic goals are
achieved and indicate to what extent e-marketing contributes to the business (revenue
contribution and return on investment). This criterion covers the different levels of mar-
keting control specified by Kotler (1997), including strategic control, profitability con-
trol and annual-plan control.
●● Include micro-level metrics which assess the efficiency of digital marketing tactics
and implementation. Wisner and Fawcett (1991) note that organisations typically use
ah ­ ierarchy of measures and they should check that the lower-level measures support
the macro-level strategic objectives. Such measures are often referred to as performance
drivers, since achieving targets for these measures will assist in achieving strategic
­objectives. Digital marketing performance drivers help optimise online marketing by
­attracting more site visitors and increasing conversion to desired marketing outcomes.
These achieve the marketing efficiency control specified by Kotler (1997). The research
by Agrawal et al. (2001), who assessed companies on metrics defined in three categories
556 Part 3 Digital marketing: implementation and practice

of attraction, conversion and retention as part of an e-performance scorecard, uses a

combination of macro- and micro-level metrics.
●● Assess the impact of digital marketing on the satisfaction, loyalty and contribution of
key stakeholders (customers, investors, employees and partners) as suggested by Adams
et al. (2000).
●● Enable comparison of performance of different digital channels with other channels as
suggested by Friedman and Furey (1999).
●● The framework can be used to assess e-marketing performance against competitors’ or
out-of-sector best practice.
Since 2000, many marketing and digital marketing dashboard services have been
­introduced, but it is difficult to understand which are the best types of solution and
whether they are worthwhile beyond the main dashboard system such as Google Analytics
or Adobe Analytics. Complete Activity 10.1 to explore the types of dashboards available
and their benefits.

Activity 10.1 Selecting the right type of digital marketing dashboard

Purpose
To assess the benefits and disadvantages of dashboards and review some of the
­options available.

Activity
1 Review some of the most commonly used dashboards by digital marketers to
understand their features and how they differentiate.
●● Google Analytics (which offers standard and customised dashboards)

●● Geckoboard

●● Kissmetrics

●● Klipfolio

●● Sproutsocial

●● Tableau

2 Identify different types of dashboard solutions.

3 List essential requirements of dashboard solutions which differentiate them from
tools such as Google Analytics.
4 What are the potential disadvantages of dashboard systems which require
management?

When identifying metrics it is common practice to apply the widely used SMART
­ nemonic and it is also useful to consider three levels – business measures, marketing meas-
m
ures and specific digital marketing measures (see the objective setting section in Chapter 4).
Figure 10.3 shows a framework of measures, which can be applied to a range of differ-
ent companies. The groupings of measures remain relevant, although they are centred on
sites or online presence, measures for engagement with social media should also be consid-
ered. In Chapter 4, we also reviewed two alternative frameworks (see Tables 4.6 and 4.9)
that can also be used for creating a performance dashboard.
Channel promotion
Measures that assess
why customers visit a
Channel promotion
site – which adverts they
have seen, which sites
These measures evaluate the volume, quality and value of where the website, social pres-
have they been referred ence or mobile site visitors originate – online or offline – and what are the sites or offline
from. media that prompted their visit. Web analytics can be used to assess which intermediary
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 557

The WebInsights™ diagnostics framework

1. Business
includes these key metrics:
contribution
1. Business contribution:
Online revenue contribution (direct and indirect),
category penetration, costs and profitability.
2. Marketing
2. Marketing outcomes: outcomes

Organisation’s targets

Organisation’s tactics
Leads, sales, service contacts, conversion and
retention efficiencies.
3. Customer satisfaction: 3. Customer
Site usability, performance/availability, contact satisfaction
strategies. Opinions, attitudes and brand impact.
4. Customer behaviour (web analytics):
Profiles, customer orientation (segmentation), 4. Customer
usability, clickstreams and site actions. behaviour

5. Site promotion:
Attraction efficiency. Referrer efficiency, cost of
acquisition and reach. Search engine visibility 5. Site
and link building. E-mail marketing. Integration. promotion

Figure 10.3 The five diagnostic categories for digital marketing measurement

Referrer sites customers are referred from (the referrer) and which keywords they typed into search
The site that a visitor engines when trying to locate product information. Similar information on referrer is not
previously visited before
following a link. typically available for visits to social media sites. Promotion is successful if traffic meets
objectives of volume, quality, value and cost (as explained in Chapter 8). Quality will be
determined by whether visitors are in the target market and have a propensity for the ser-
vice offered (through reviewing conversion (online or offline), bounce rates and cost of
acquisition for different referrers).

Key measure
Referral mix. For each referral source such as paid search or display ads it should be pos-
sible to calculate:
●● percentage of all referrals or sales (and influence in achieving sale last click or assist);
●● cost-per-acquisition (CPA) or cost-per-sale (CPS);
●● contribution to sales or other outcomes.

Channel buyer Channel buyer behaviour

behaviour
Describes which content Once customers have been attracted to the site we can monitor content accessed, when
is visited and the time and
duration.
they visit and how long they stay, and whether this interaction with content leads to satis-
factory marketing outcomes such as new leads or sales. If visitors are incentivised to regis-
ter on-site it is possible to build up profiles of behaviour for different segments. Segments
can also be created according to visitor source and content accessed. It is also important
to recognise return visitors for whom cookies or login are used. In Chapter 6 we saw how
hurdle rates can be used to assess activity levels for return visits, e.g. 30 per cent of custom-
ers return to use the online service within 90 days.

Key ratios
●● Bounce rates for different pages, i.e. proportion of single page visits.
●● Home page views/all page views, e.g. 20 per cent = (2000/10,000).
558 Part 3 Digital marketing: implementation and practice

Stickiness ●● Stickiness: page views/visitor sessions, e.g. 2 = 10,000/5000.

An indication of how long ●● Repeats: visitor sessions/visitors, e.g. 20 per cent = 1000/5000.
a visitor stays on-site.

Channel satisfaction Channel satisfaction

Evaluation of the
customer’s opinion of the Customer satisfaction with the online experience is vital in achieving the desired channel
service quality on the site
and supporting services
outcomes. Online methods such as online questionnaires, focus groups and interviews can
such as email. be used to assess customers’ opinions of the website content and customer service and how
it has affected overall perception of brand. Benchmarking services such as Foresee (www.
foreseeresults.com) based on the American Customer Satisfaction Index methodology are
published for some industries. These assess scores based on the gap between expectations
and actual service.

Key measure
Customer satisfaction indices. These are discussed in Chapter 7 and include ease of use,
site availability and performance, and email response. To compare customer satisfaction
with other sites, benchmarking services can be used.

Channel outcomes Channel outcomes

Record of customer
actions taken as a Traditional marketing objectives such as number of sales, number of leads, conversion
consequence of a visit to
a site.
rates and targets for customer acquisition and retention should be set and then compared
Conversion rate
to other channels. Dell Computer (www.dell.com) records on-site sales and also orders gen-
Percentage of site visitors erated as a result of site visits, but placed by phone. This is achieved by monitoring calls to
who perform a particular a specific phone number unique to the site.
action such as making a
purchase. Key marketing outcomes include:
●● registration to site or subscriptions to an email newsletter;
●● requests for further information such as a brochure or a request for a call-back from a
customer service representative;
●● responding to a promotion such as an online competition;
●● an offline (phone or store) lead or sale influenced by a visit to the site;
●● a sale on-site.

Key measure
●● Channel contribution (direct and indirect).

A widely used method of assessing channel outcomes is to review the conversion rate,
which gives an indication of the percentage of site visitors who take a particular outcome.
For example:
●● Conversion rate, visitors to purchase = 2 per cent (10,000 visitors, of which 200 make
purchases).
●● Conversion rate, visitors to registration = 5 per cent (10,000 visitors, of which 500
register).
Attrition rate A related concept is the attrition rate which describes how many visitors are lost at each
Percentage of site visitors
lost at each stage in
step of a conversion funnel from landing page to checkout. Figure 10.4 shows that for a set
making a purchase. time period, only a proportion of site visitors will make their way to product information,
a small proportion will add an item to a basket and a smaller proportion still will actually
make the purchase. A key feature of e-commerce sites is that there is a high attrition rate
between a customer adding an item to a basket and subsequently making a purchase. It is
surmised that this is due to fears about credit card security, and that customers are merely
experimenting.
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 559

Wrong
audience
100
Slow page
load

Percentage of site visits

Unclear
marketing Clumsy site
message navigation

Unengaging Awkward
look and feel selection
Price
uncompetitive
Card validation
No real-time error
stock information No e-mail
notification
High shipping
costs
Failed
delivery
0
Acquisition First impressions Product selection Payment and fulfilment
Depth of relationship

Figure 10.4 Potential reasons for causing attrition on an e-commerce site

Channel profitability Channel profitability

The profitability of the
website, taking into A contribution to business profitability is always the ultimate aim of e-commerce. To assess
account revenue and cost
and discounted cash flow
this, leading companies set an Internet contribution target of achieving a certain proportion
of sales via the channel. Assessing contribution is more difficult for a company that cannot
sell products online, but the role of the Internet in influencing leads and purchase should be
assessed. Discounted cash flow techniques are used to assess the rate of return over time.

Multichannel evaluation
The frameworks we have presented in this chapter are explained in the context of an indi-
vidual channel, but with the contribution of the channel highlighted as percentage sales or
profitability. But as Wilson (2008) has pointed out, there is a need to evaluate how different
channels support each other. Wilson says:
Traditional metrics have been aligned to channels, measuring resource input or leads in at
one end and the value of sales generated by the channel at the other end. For companies that
have been operating in a single channel environment, this might have been relatively e
­ ffi­cient –
but it no longer works when the organisation diversifies to a multichannel approach.

He suggests the most important aspect of multichannel measurement is to measure ‘chan-

nel cross-over effects’. This involves asking, for example: ‘How can the impact of a paid
search campaign be measured if it is as likely to generate traffic to a store, salesforce or call
centre as to a website?’ and ‘How can the impact of a direct mail campaign be tracked if it
generates website traffic as well as direct responses?’
An example of a balanced scorecard style dashboard developed to assess and compare
channel performance for a retailer is presented in Figure 10.5.
560 Part 3 Digital marketing: implementation and practice

Results (6) Customers & stakeholders (5)

• Revenue • Overall customer satisfaction
• Multichannel contribution • Customer propensity to defect
• Degree multichannel sells up • Customer propensity to purchase
• Costs per channel • Customer perception of added value
• Degree of sweating assets • Integration of customer experience
• Multichannel infrastructure costs

Core processes (3) People and knowledge (4)

• Productive multichannel usage • Staff satisfaction
• Price (relative to competitors/other • Appropriate behaviours ‘Living the brand’
channels) • Willingness to diversify/extend the brand
• Quality of integrated customer view • Knowledge of target customer

Figure 10.5 Multichannel performance scorecard example for a retailer

Source: Wilson (2008)

Stage 3: Tools and techniques for collecting metrics

and summarising results

Organisations need to select the most appropriate tools for collecting and reporting met-
rics which meet requirements such as reporting of marketing performance, accuracy, analy-
sis and visualisation tools, integration with other marketing information systems (import,
export and integration using XML standards), ease of use, configuration (e.g. creation of
custom dashboards and email alerts), support quality, cost of purchase, configuration and
ongoing support.
Techniques to collect metrics include the collection of site-visitor activity data such as
that stored in web analytics systems and in site log files; the collection of metrics about
Site-visitor activity data outcomes such as online sales or email enquiries and traditional marketing research tech-
Information on content niques such as questionnaires and focus groups which collect information on the custom-
and services accessed by
e-commerce site visitors. er’s experience on the website. We start by describing methods for collecting site-visitor
Hit
activity data and then review more traditional techniques of market research.
Recorded for each
graphic or text file Collecting site-visitor activity data
requested from a web
server. It is not a reliable
Site-visitor activity data captured in web analytics systems records the number of visitors
measure for the number
of people viewing a page. on the site and the paths or clickstreams they take through the site as they visit different
Log file analyser content. The terms used to measure visitor activity are summarised in Table 10.1.
A separate program such In the early days of Internet marketing, in the mid-1990s, this information was typically
as WebTrends that is
used to summarise the collected using log files. The server-based log file is added to every time a user downloads
information on customer a piece of information (a hit) and is analysed using a log file analyser (as illustrated by
activity in a log file.
­Figure 3.8). Examples of transactions within a log file are:
Page impression
A more reliable measure www.davechaffey.com – [05/Oct/2006:00:00:49 -000] ‘GET/index.html HTTP/1.0’ 200 33362
than a hit, denoting one
person viewing one page. www.davechaffey.com – [05/Oct/2006:00:00:49 -000] ‘GET/logo.gif HTTP/1.0’ 200 54342
Also known as page view.
Unique visitors Despite their wide use in the media, hits are not a useful measure of website effective-
Individual visitors to a ness since if a page consists of ten graphics, plus text, this is recorded as 11 hits. Page
site measured through
cookies or IP addresses impressions or page views and unique visitors are better measures of site activity. Auditing
on an individual computer. companies such as ABC electronic (www.abce.org.uk), which audit sites for the purpose
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 561

Table 10.1 Terminology for key website volume measures

Measure Measure Definition

1 How many? Unique users A unique and valid identifier [for a site visitor].
‘audience reach’ Sites may use (i) IP + User – Agent, (ii) cookie
and/or (iii) registration ID
2 How often? Visit A series of one or more page impressions,
‘frequency metric’ served to one user, which ends when there is a
gap of 30 minutes or more between successive
page impressions for that user
3 How busy? Page impression A file, or combination of files, sent to a valid user
‘volume metric’ as a result of that user’s request being received
by the server
4 What see? Ad impressions A file or a combination of files sent to a valid user
as an individual advertisement as a result of that
user’s request being received by the server
5 What do? Ad clicks An ad impression clicked on by a valid user
Source: ABCe (www.abce.org.uk)

of proving the number of visitors to a site to advertisers, use unique visitors and page
­impression as the main measures.
An example of visitor volume to a website using different measures based on real,
representative data for one month is presented in Figure 10.6. You can see how hits are
much higher than page views and unique visitors and are quite misleading in terms of the
‘opportunities to see’ a message. We can also learn from the ratio between some of these
measures – the figure indicates:
●● Pages per visit (PPV) – the average number of pages viewed per visitor to a site (this is
indicative of engagement with a site since the longer a visitor stays on a ‘sticky site’, the
higher this value will be). PPV is a more accurate indication of stickiness than duration
on a site in minutes since this figure is skewed upwards by visitors who arrive on a site
and are inactive before their session times out at 30 minutes.
●● Visits per (unique) visitor (VPV) – this suggests the frequency of site visits. Readers will
realise that this value is dependent on the period that data are collected over. These data

Hits = All files downloaded

e.g. = 4,000,000

Page views = ‘Impressions’ viewed

e.g. = 1,200,000
PPV = 10
Visitor sessions = Visits
e.g. = 120,000
VPV = 2
Visitors = Unique visitors
e.g. = 60,000

Figure 10.6 Examples of different measures of visitor volume to a website

562 Part 3 Digital marketing: implementation and practice

are reported for a month during which time one would not expect many returning visi-
tors. So it is often more relevant to present these data across a quarter or a year.
Other information giving detailed knowledge of customer behaviour that can be reported
by any web analytics package includes:
●● top pages;
●● entry and exit pages;
●● path or clickstream analysis showing the sequence of pages viewed;
●● country of visitor origin (actually dependent on the location of their ISP);
●● browser and operating system used;
●● referring URL and domain (where the visitor came from).

Digital marketing insight 10.1 Focus on measuring social media marketing

Social media marketing has its own range of specialist measures that can appear confusing, but are best
­understood in the context of a combination of website and PR measures. These show the volume, quality, senti-
ment and value of interactions. Analyst Altimeter (2010) has created a useful framework (shown in ­Figure 10.7)
that helps map out different social media measures in the context of level of business management.
You can see that there are three levels of KPIs:
●● Business-level KPIs to measure contribution from social media. These KPIs include contribution to
revenue through direct sales attributed to social media. Softer measures include reputation and customer
satisfaction (CSAT).
●● Reach and influence KPIs to review reach, share-of-voice and sentiment. These show the relative
comparison of a brand’s reach.
●● Engagement KPIs to manage social media. These are the easiest measures to collect, but the least valu­
able since they don’t directly show contribution to business value. Although easy to collect, data on

Role Metrics Specific data (examples)

Business Business Revenue,

executives metrics reputation, CSAT

Business Social media Share of voice, resonation, WOM,

stakeholders analytics support response, insights intake

Community managers Engagement Clicks, fans, followers,

and agencies data RTs, views, check-ins

Figure 10.7 A framework for different measures used to evaluate and manage social media
marketing.
Source: Altimeter (2010) with permission (Creative commons)
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 563

interaction with social sites is often supplied separately by the owners of the different social presence and
tools for managing social interaction. A new class of social analytics tools have been created to bring this
data together. Figure 10.8 shows an example from the social media management tool Hootsuite where
sharing of shortened URLs linking to different social media sites have driven traffic back to a main website.
Direct traffic is where visitors click direct through from a social media messaging application like Hootsuite
or Tweetdeck to the site.

A common question within social media is how to assess the value of a consumer connecting with a brand,
by liking on Facebook, following on Twitter or placing in a brand in a circle on Google+. Since the tracking
of social media can’t show what an individual does on the network, specific value is difficult to ­establish.
Instead what we can assess is the relative purchase rates of visitors from social media sites to websites
­compared to other channels using measures like conversion rate and revenue per visitor.

Figure 10.8 Example of measure from Hootsuite application for measuring social media
marketing

Design for analysis

Measurement is often highlighted as an issue once the first version of a site has been ‘up
and running’ for a few months, and employees start to ask questions such as ‘How many
customers are visiting our site, how many sales are we achieving as a result of our site and
how can we improve the site to achieve a return on investment?’ The consequence of this is
Design for analysis that performance measurement is something that is often built into an online presence ret-
(DFA) rospectively. Preferable is if a technique known as design for analysis (DFA) is designed into
The required measures
from a site are considered the site so companies can better understand the types of audience and their decision points.
during design to better For example, for Dell (www.dell.com), the primary navigation on the home page is by cus-
understand the audience
of a site and their decision tomer type. This is a simple example of DFA since it enables Dell to estimate the proportion
points. of different audiences to its site and, at the same time, connect them with relevant content.
564 Part 3 Digital marketing: implementation and practice

Other examples of DFA include:

●● Breaking up a long page or form into different parts, so you can see which parts people
are interested in.
●● A URL policy (see Chapter 8) used to recommend entry pages for printed material.
●● Group content by audience type or buying decision and setting up content groups of
related content within web analytics systems.
●● Measure attrition at different points in a customer journey, e.g. exit points on a five-
page buying cycle.
●● A single exit page to linked sites.

AB and multivariate testing

Often site owners and marketers reviewing the effectiveness of a site will disagree and the
only method to be certain of the best-performing design or creative alternatives is through
designing and running experiments to evaluate the best to use. Matt Round, then director
of personalisation at Amazon, speaking at the e-metrics summit in 2004, said the Amazon
philosophy, described further in Case study 10 is:
Data trumps intuition.
AB testing
A/B or AB testing refers
to testing two different
AB testing and multivariate testing are two measurement techniques that can be used to
versions of a page or review design effectiveness to improve results.
a page element such
as a heading, image or
button for effectiveness. AB testing
The alternatives are In its simplest form, A/B or AB testing refers to testing two different versions of a page or a
served alternately with
the visitors to the page
page element such as a heading, image or button. Some members of the site are served alter-
randomly split between nately, with the visitors to the page randomly split between the two pages. Hence it is some-
the two pages. Changes times called ‘live split testing’. The goal is to increase page or site effectiveness against key
in visitor behaviour can
then be compared using performance indicators including click-through rate, conversion rates and revenue per visit.
different metrics such as When completing AB testing it is important to identify a realistic baseline or control
click-through rate on page
elements like buttons page (or audience sample) to compare against. This will typically be an existing landing
or images, or macro- page. Two new alternatives can be compared to previous control, which is known as an
conversion rates, such
as conversion to sale or
ABC test. Different variables are then applied as in Table 10.2.
sign-up. An example of the power of AB testing is an experiment Skype performed on its main
Control page topbar navigation, where it found that changing the main menu options ‘Call Phones’ to
The page against which ‘Skype Credit’ and ‘Shop’ to ‘Accessories’ gave an increase of 18.75 per cent revenue per
subsequent optimisation
will be assessed. Typically
visit (Skype was speaking at the 2007 e-metrics summit). That’s significant when you have
a current landing page. hundreds of millions of visitors! It also shows the importance of being direct with naviga-
When a new page tion and simply describing the offer available rather than the activity.
performs better than the
existing control page,
it becomes the control Multivariate testing
page in subsequent
testing. Also known as Multivariate testing is a more sophisticated form of AB testing which enables simultane-
‘champion-challenger’ ous testing of pages for different combinations of page elements that are being tested.

Table 10.2 AB test example

Test A (Control) B (Test page)

Test 1 Original page New headline, existing button, existing body copy
Test 2 Original page Existing headline, new button, existing body copy
Test 3 Original page Existing headline, existing button, new body copy
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 565

This enables selection of the most effective combination of design elements to achieve the
desired goal.
An example of a multivariate test is shown in Mini case study 10.1.

Mini case study 10.1 Multivariate testing at National Express Group increases
­conversion rates

The National Express Group is the leading provider of travel solutions in the UK. Around 1 billion journeys
a year are made worldwide on National Express Group’s bus, train, light rail and express coach and airport
­operations. A significant proportion of ticket bookings are made online through the company’s website at
www.nationalexpress.com/home.aspx.
The company uses multivariate testing provider Maxymiser to run an experiment to improve conversion
rate of a fare-selection page which was the penultimate step in booking (Figure 10.9). The analysis team iden-
tified a number of subtle alterations to content (labelled A to E) and calls to action on the page with the aim of
stimulating visitor engagement and driving a higher percentage of visitors through to successful conversion
without changing the structure of the page or National Express brand identity. In order to aid more effective
up-sell to insurance add-ons, changes to this call to action were also proposed.
It was decided that a multivariate test would be the most effective approach to determine the best per-
forming combination of content. The variants jointly developed by Maxymiser and the client were tested with
all live site visitors and the conversion rate of each combination monitored; 3500 possible page combinations
were tried and during the live test the underperforming combinations were taken out to maximise conversion
rates at every stage.
At the end of the testing period, after reaching statistical validity, results gave the best combination of ele-
ments, showing a 14.11 per cent increase in conversion rates for the page – i.e. 14.11 per cent more visitors

Figure 10.9 National Express page assessed through multivariate testing

566 Part 3 Digital marketing: implementation and practice

were sent through to the fourth and final step in the registration process, immediately hitting bottom line
­revenue for National Express (Figure 10.10).

Content Lift on
Maxybox A Maxybox B Maxybox C Maxybox D Maxybox E
combination control
1 Variant 3 Variant 2 Variant 4 Variant 3 Variant 1 14.11%

2 Variant 3 Variant 3 Variant 4 Default Default 14.09%

3 Variant 6 Variant 3 Variant 4 Default Default 11.15%

4 Variant 3 Variant 3 Variant 2 Default Variant 3 10.57%

Default content Variant 3 Variant 2 Default Default Default 0.00%

Conversion rate uplift by page combination:

1 14.11%

2 14.09%

Page
3 11.15%
combination

4 10.57%

Default 0%

0% 2% 4% 6% 8% 10% 12% 14% 16%

Figure 10.10 Results of multivariate testing for National Express

Clickstream analysis and visitor segmentation

Clickstream analysis refers to detailed analysis of visitor behaviour in order to identify
improvements to the site. Each web analytics tool differs slightly in its reports and termi-
nology, but all provide similar reports to help companies assess visitor behaviour and diag-
nose problems and opportunities. Table 10.3 gives an indication of the type of practical
questions asked by web analyst and consultant Dave Chaffey (www.davechaffey.com) when
reviewing clients’ sites.

Path analysis
Aggregate clickstreams are usually known within web analytics software as forward or
reverse paths. This is a fairly advanced form of analysis, but the principle is straightfor-
ward – you seek to learn from the most popular paths.
Viewed at an aggregate level across the site through ‘top paths’ type reports, this form
of clickstream analysis often doesn’t appear that useful. It highlights typically paths which
are expected and can’t really be influenced. The top paths are often:
●● Home page: Exit
●● Home page: Contact Us: Exit
●● News page: Exit
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 567

Table 10.3 A summary of how an analyst will interpret web analytics data. GA is terminology for Google
Analytics (www.google.com/analytics), one of the most widely used tools

Analyst question Typical web analytics Diagnosis analyst used to improve performance
report terminology
How successful is Conversion goals (GA) • Is engagement and conversion consistent with other sites
the site at achieving Bounce rates (GA) in sector?
engagement and Pages/visit (GA) • What are maximum engagement and conversion rates
outcomes? from different referrers?
Where are visitors Top entry pages • How important is home page compared to other page
entering the site? Top landing pages (GA) categories and landing pages?
• Does page popularity reflect product popularity?
• Review messaging/conversion paths are effective on
these pages
• Assess source of traffic, in particular keywords from
search engines and apply elsewhere
What are the sources Referrers • Are the full range of digital media channels relevant for a
of visitors (referrers)? Traffic sources company represented?
Filters set up to segment • Is the level of search engine traffic consistent with the
visitors brand reputation?
• What are the main link partners driving free traffic (poten-
tial for more?)
What is the most Top content (GA) • Is page popularity as expected?
popular content? • Are there problems with findability caused by navigation
labelling?
• Which content is most likely to influence visitors to
outcome?
• Which content is most popular with returning visitors
segment?
Which are the most Site search (GA) • How popular are different forms of navigation, e.g. top
popular findability menu, sidebar menus, etc?
methods? • What are the most popular searches?
• Where do searches tend to start?
• Are they successfully finding content or converting to sale?
Where do visitors Top exit pages (GA) • Are these as expected (home page, About Us page,
leave the site? transaction completion)?
• Are there error pages (e.g. 404 not found) which cause
visitors to leave?
Which clickstreams Path analysis • How can attrition in conversion funnels be improved?
are taken? Top paths (GA) • What does forward path analysis show are the most
effective calls-to-action?
• What does reverse path analysis indicate about the pages
which influence sales?

Clickstream analysis becomes more actionable when the analyst reviews clickstreams in
Forward path analysis the context of a single page – this is forward path analysis or reverse path analysis.
Reviews the combinations
of clicks that occur from a
page. This form of analysis On-site search effectiveness
is most beneficial from On-site search is another crucial part of clickstream analysis on many sites since it is a key
important pages such as
the home page, product way of finding content, so a detailed search analysis will pay dividends. Key search metrics
and directory pages. Use to consider are:
this technique to identify:
messaging/navigation ●● number of searches;
combinations which work
best to yield the most
●● average number of searches per visitor or searcher;
clicks from a page. ●● percentage of searches returning zero results;
568 Part 3 Digital marketing: implementation and practice

Reverse path analysis ●● percentage of site exits from search results;

Reverse path analysis
indicates the most
●● percentage of returned searches clicked;
popular combination of ●● percentage of returned searches resulting in conversion to sale or other outcome;
pages and/or calls-to- ●● most popular search terms – individual keyword and keyphrases.
action which lead to a
page. This is particularly
useful for transactional
pages such as the first Visitor segmentation
checkout page on a Segmentation is a fundamental marketing approach, but is often difficult within web ana-
consumer site; a lead
generation or contact us lytics to relate customer segments to web behaviour because the web analytics data isn’t
page on a business-to- integrated with customer or purchase data.
business site; an email
subscription page or a
However, all analytics systems have a capability for a different, but valuable form of
call-me-back option. segmentation where it is possible to create specific filters or profiles to help understand one
type of site visitor behaviour. Examples of segments include:
●● First- time visitors or returning visitors.
–– Visitors from different referrer types including:
–– Strategic search keyphrases, brand keyphrases, etc.;
–– Display advertising.
●● Converters against non-converters.
●● Geographic segmentation by country or region (based on IP addresses).
●● Type of content accessed, e.g. are some segments more likely to convert? For example,
speaking at Ad Tech London ’06, MyTravel reported that it segments visitors into:
–– site flirt (two pages or fewer);
–– site browse (two pages or more);
–– saw search results;
–– saw quote;
–– saw payment details;
–– saw booking confirmation details.

Selecting a web analytics tool

There has been consolidation of web analytics tools, such that there is now a basic choice
of a free service such as Google Analytics or a paid service from the main providers such
as Omniture (owned by Adobe Systems), Coremetrics (owned by IBM) and WebTrends,
which may cost hundreds of thousands of dollars a year for a popular site. All will report
similar measures for digital marketing activity to those explored earlier in the chapter, so
often the selection of the best system will depend on factors such as:
●● Integration with other data sources (for example, social media marketing, customer
data and financial reporting). Figure 10.11 gives an indication of the types of data that
need to be integrated; these include operational data, tactical and strategic data.
●● Accuracy. Potential sources of inaccuracy are reviewed in Table 10.4 comparing tradi-
tional log file analysis to the more common browser-based or tag-based measurement
system that records access to web pages every time a page is loaded into a user’s web
browser through running a short script, program or tag inserted into the web page.
The key benefit of the browser-based approach is that it is potentially more accurate
than server-based approaches. Figure 10.12 indicates how the browser-based approach
works. The free version of Google Analytics uses sampling on large sites which can
­decrease accuracy.
●● Media attribution. We saw in Chapter 8 that the ‘last-click-wins’ model of attributing a
referral source to sale is inaccurate and weighted models based on the whole customer
journey are more accurate. The capability of analytics system to display this is impor-
tant for companies investing a lot in online media.
●● Visualisation. How data are displayed through reports and alerts. Vendors continually
introduce new features in this area.
●● Customisation facilities. For creating and distributing new reports and alerts
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 569

Strategic
Performance ‘Management
management scorecards
systems and dashboards’

Audience data
Tactical

Lifetime
(reach, characteristics,
value models
opinions)
‘web analytics’
Operational

Referrer or Site or Customer Sales

campaign clickstream response and transactions
data data profile data (legacy)

Figure 10.11 Different types of data within a performance management system

for Internet marketing

Table 10.4 Inaccuracies caused by server-based log file analysis

Sources of undercounting Sources of overcounting

Caching in user’s web browsers (when a user Frames (a user viewing a framed page with
accesses a previously accessed file, it is three frames will be recorded as three page
loaded from the user’s cache on their PC) impressions on a server-based system)
Caching on proxy servers (proxy servers are Spiders and robots (traversing of a site
used within organisations or ISPs to reduce by spiders from different search engines
Internet traffic by storing copies of frequently is recorded as page impressions. These
used pages) spiders can be excluded, but this is
time-consuming)
Firewalls (these do not usually exclude page Executable files (these can also be
impressions, but they do assign a single IP recorded as hits or page impressions
address for the user of the page, rather than unless excluded)
referring to an individual’s PC)
Dynamically generated pages, generated
‘on the fly’, are difficult to assess with server-
based log files

●● Support services. For configuration of data feeds and reports and consulting to assist in
auctioning the results. In 2011 the free service Google Analytics introduced a premium
version for large corporate customers which included account management.
●● Privacy considerations. Web analytics systems store personal data. As we saw in
­Chapter 4, it is important that data collection and disclosure about the method of
­collection by the system follow the latest laws about use of cookies.
570 Part 3 Digital marketing: implementation and practice

Strategic data
Performance management systems for senior managers will give the big picture presented
as scorecards or dashboards showing trends in contribution of digital channels to the
organisation in terms of sales, revenue and profitability for different products.
An example of the output reporting from a web analytics service is shown in
Figure 10.12.

Marketing research using the Internet

Internet-based market Internet-based marketing research can help determine the influence of a website and
research
The use of online
related communications on customer perception of the company and its products and
questionnaires and services. But it also has wider applications of gaining feedback from customers about a
focus groups to assess brand and how it could develop in future. Smart Insights (2010) identifies these five differ-
customer perceptions
of a website or broader ent classes of online feedback tools:
marketing issues.
1 Website feedback tools. Provide a permanent facility for customers to give feedback
by prompts on every page. They are run continuously to enable continuous feedback
including ratings on page content, and also products and services.
2 Site user intent-satisfaction surveys. These tools measure the gap between what the user
had hoped to do on the site and what they actually achieved. We gave the example of
4Q in Chapter 7 that covers four questions to assess site effectiveness.
3 Crowdsourcing product opinion software. These are broader than web feedback, ena-
bling customers to comment about potential new services. This is the approach used by
Dell in the IdeaStorm (www.ideastorm.com).
4 Simple page or concept feedback tools. Again a form of crowdsourcing, these tools give
feedback from an online panel about page layout, messaging or services.

Figure 10.12 Differences between browser-based and server-based measure-

ment systems
 Chapter 10 Evaluation and improvement of d
­ igital channel performance 571

5 General online survey tools. Tools like Zoomerang ( www.zoomerang.com ) and

­SurveyMonkey (www.surveymonkey.com) enable companies to survey their audience
at a low cost.
The full options for conducting survey research include interviews, questionnaires and fo-
cus groups are summarised in Table 10.5. Each of these techniques can be conducted of-
fline or online.
We will now briefly review the strengths and weaknesses of the different research
­techniques and some approaches to best practice.

Table 10.5 A comparison of different online metrics collection and research methods

Technique Strengths Weaknesses

Server-based log file • Directly records customer behaviour • Not based around marketing out-
analysis of site activity on site plus where they were referred comes such as leads or sales
from • Size – even summaries may be over 50
• Low cost pages long
• Gives insight on robot crawling • Does not directly record channel
for SEO satisfaction
• Undercounting/overcounting
• Misleading unless interpreted carefully
• Most site analytics tools are now
browser based
Browser-based site activity • Greater accuracy than server-based • Relatively expensive method
data analysis • Similar weaknesses to server-based
• Counts all users, cf. panel approach technique apart from accuracy
• Limited demographic information
AB and multivariate testing • Structured experiments to review • Often requires cost of a separate tool
influence of on page variables (e.g. or module from standard web analytics
messaging and buttons) to improve package
conversion from a website • Content management systems or page
templates may not support AB/multi-
variate testing
Panel activity and • Provides competitor comparisons • Depends on extrapolation from
demographic data • Gives demographic profiling data-limited sample that may not be
representative representative
• Avoids undercounting and
overcounting
Outcome data, e.g. • Records marketing outcomes • Difficulty of integrating data with other
enquiries, customer emails methods of data collection when
service collected manually or in other
information systems
Online questionnaires • Can record customer satisfaction and • Difficulty of recruiting respondents
Customers are prompted profiles who complete accurately
randomly – every nth • Relatively cheap to create and analyse • Sample bias – tend to be advocates or
customer or after customer disgruntled customers who complete
activity or by email
Online focus groups • Relatively cheap to create • Difficult to moderate and coordinate
Synchronous recording • No visual cues, as from offline focus
groups
Mystery shoppers • Structured tests give detailed • Relatively expensive
Example is customers feedback • Sample must be representative
are recruited to evaluate • Also tests integration with other
the site, e.g. www. channels such as email and phone
emysteryshopper.com
572 Part 3 Digital marketing: implementation and practice

Questionnaires and surveys

Malhotra (1999) suggested that Internet surveys using questionnaires will increase in
­popularity since the cost is generally lower, they can be less intrusive and they have the
ability to target specific populations. Register et al. (2014) confirmed that Internet surveys
are fast becoming the preferred mode for survey delivery as they afford researchers con-
venient use of design options such as ‘forced answering’ (FA) that can virtually eliminate
item non-response error.
However, a recent review of the technique by Stern et al. (2014) shows that the Internet
has failed to deliver in terms of response rates and it can be more challenging to get a repre-
sentative sample. Questionnaires often take the form of pop-up surveys. The key issues are:
A Encouraging participation. Techniques that can be used are:
interruption on entry – a common approach where every 100th customer is
●●

prompted;
●● continuous, for example click on a button to complete survey;

●● on registration on-site the customer can be profiled;

●● after an activity such as sale or customer support, the customer can be prompted for

their opinion about the service;

●● incentives and promotions (this can also be executed on independent sites);

●● by email (an email prompt to visit a website to fill in a survey or a simple email

survey).
B Stages in execution. It is suggested that there are five stages to a successful question-
naire survey:
1 attract (button, pop-up, email as above);
2 incentivise (prize or offer consistent with required sample and audience);
3 reassure (why the company is doing it – to learn, not too long and that confidential-
ity is protected);
4 design and execute (brevity, relevance, position);
5 follow-up (feedback).
C Design. Grossnickle and Raskin (2001) suggest the following approach to structuring
questionnaires:
●● easy, interesting questions first;

●● cluster questions on same topic;

●● flow topic from general to specific;

●● flow topic from easier behavioural to more difficult attitudinal questions;

●● easy questions last, e.g. demographics or off-putting questions.

Typical questions that can be asked for determining the effectiveness of Internet marketing are:
●● Who is visiting the site? For example, role in buying decision? Online experience? A
­ ccess
location and speed? Demographics segment?
●● Why are they visiting? How often do they visit? Which information or service? Did they
find it? Actions taken? (Can be determined through web analytics.)
●● What do they think? Overall opinion? Key areas of satisfaction? Specific likes or
­dislikes? What was missing that was expected?

Focus groups
Malhotra (1999) noted that the advantage of online focus groups is that they can be used
to reach segments that are difficult to access, such as doctors, lawyers and professional
people. This author also suggests that costs are lower, they can be arranged more rapidly
and can bridge the distance gap when recruiting respondents. Traditional focus groups
can be conducted, where customers are brought together in a room and assess a website;
this will typically occur pre-launch as part of the prototyping activity. Testing can take the
form of random use of the site or, more usefully, the users will be given different scenarios
 Chapter 10 Evaluation and improvement of digital channel performance 573

to follow. Focus groups tend to be relatively expensive and time consuming, since rather
than simply viewing an advertisement, the customers need to actually interact with the
website. Conducting real-world focus groups has the benefit that the reactions of site users
can be monitored; the scratch or slap of the head cannot be monitored in the virtual world!

Mystery shoppers
Real-world measurement is also important since the Internet channel does not exist in
isolation. It must work in unison with real-world customer service and fulfilment. Chris
Russell of eDigitalResearch (www.edigitalresearch.com), a company that has completed
online customer service surveys for major UK retailers and travel companies, says:
we also needed to make sure that the bricks-and-mortar customer service support was
actually supporting what the clicks-and-mortar side was promising. There is no doubt
that an e-commerce site has to be a complete customer service fulfilment picture, it can’t
just be one bit working online that is not supported offline.

An eMysteryShopper survey involves shoppers not only commenting on site usability, but
also on the service quality of email and phone responses together with product fulfilment.
Mystery shoppers test these areas:
●● site usability;
●● e-commerce fulfilment;
●● email and phone response (time, accuracy);
●● impact on brand.

Customer experience and content management process

As part of the process of continuous improvement in online marketing, it is important

to have a clearly defined process for making changes to the online presence of a com-
pany. If pages remain static, as is the case with some brochureware sites we still see, then
the opportunity to engage customers and prospects with a brand is missed. With search
engines and social media sites featuring real-time data posted in blogs, companies that
have a static site are missing an opportunity to get better visibility. A static site also has
a missed opportunity to make the site more effective at generating value for a business by
increasing conversion rates using the AB and multivariate testing approaches we talked
about in the last section.
The key to keeping a website dynamic is to have a clear content and e-communications
strategy based on a content or social hub, as we have seen in previous chapters. The site
and content update process should be understood by all staff contributing content to the
site, with their responsibilities clearly identified in their job descriptions. To understand
the process, consider the main stages involved in publishing a page. A simple model of the
work involved in maintenance is shown in Figure 10.13. It is assumed that the needs of the
users and design features of the site have already been defined when the site was originally
created (as described in Chapter 7). The model only applies to minor updates to copy, or
perhaps updating product or company information. The different tasks involved in the
maintenance process are as follows:
1 Write. This stage involves writing the marketing copy and, if necessary, designing the
layout of copy and associated images.
2 Review. An independent review of the copy is necessary to check for errors before a
document is published. Depending on the size of organisation, review may be necessary
by one person or several people covering different aspects of content quality such as