Digital Analytics Chapter 12 PDF

© Noordhoff Uitgevers bv 525 12 Digital analytics Authors: Tom van den Berg and Marjolein Visser The first eight chapters of this book describe how the marketer can...

© Noordhoff Uitgevers bv 525 12 Digital analytics Authors: Tom van den Berg and Marjolein Visser The first eight chapters of this book describe how the marketer can establish a clear idea of what the market’s wants and desires are (the market sensing process) and how the internet can be used to further product development and improvement as well as for setting up processes surrounding the product or the service (the product realisation process). Once everything is customer-ready, the customer acquisition process commences: defining target segments and recruiting new customers. When the order processing has been executed as smoothly as possible, the relationship can be further expanded and deepened during the customer relationship management process. You have also been able to read what requirements are imposed on effective websites and apps. This chapter addresses the question of what needs to be measured in order to monitor and improve the Digital Marketing process. In addition, it provides you with insight regarding the basic concepts of digital analytics and the most important indicators that are used by analysts. This chapter will cover: digital analytics: definition and starting points performance indicators traffic sources the ABC model recognising the phase that the visitor is in social media metrics digital analytics within the organisation 12 After studying this chapter, you will be able to: describe in your own words what digital analytics entails, what the most important performance indicators are and what the role of digital analytics is within Digital Marketing explain the most important basic values in digital analytics, interpret statistics of these basic values and indicate what additional information you need for a good interpretation indicate how, in digital analytics, the various Digital Marketing communication tools are included in the traffic sources 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 525 26/11/20 5:01 PM  526 © Noordhoff Uitgevers bv identify the relationship between the phase of the Digital Marketing funnel that the visitor is in and the key performance indicators explain how the ABC model works and how you can use it to structure digital analyses identify the most important social media metrics that should be included in a social media monitoring programme, explain why these are important and translate them into recommendations regarding improvements to communications via social media indicate which six factors determine the maturity of an organisation in terms of Digital Marketing analytics and why these factors must be balanced § 12.1 Digital analytics: definition and starting points Web analytics In the past, we referred to web analytics. According to the Web Analytics Association (WAA), the definition of web analytics is ‘the measurement, collection, analysis and reporting of internet data for the purposes of understanding and optimising web usage’. The definition is broad enough to remain applicable. The terms ‘web’ and ‘website’ are somewhat old- fashioned due to the popularity of apps, tablet devices and mobile phones. The focus is shifting towards ‘digital’ or ‘omnichannel analytics’. Omnichannel means ‘all’ channels, both offline (physical stores and print media eg. leaflets) and online (digital media eg. websites and apps). Digital analytics Digital analytics provide market insight, show how target segments respond to campaigns, what visitors are doing on the website and help to evaluate and improve Digital Marketing decisions. The digital world is constantly evolving: website visits from a desktop or laptop have become less important. Brands are also increasingly present on channels that are not self-owned, eg. Facebook, Instagram, Pinterest or Twitter. This means that content is often viewed via apps or mobile browsers on tablet devices or mobile phones. Where the majority of people used to go to a website’s homepage first, they are now often looking for specific pages that have been shared by people from apps or on social channels eg. Facebook, Twitter or Pinterest. In addition, many visitors arrive on a website as a result of a specific search result from Google. You not only want to know what the visitors are doing on their desktop or laptop, but also what interactions they have via apps and mobile pages, and on social channels eg. Facebook or Twitter. Ideally, you would like to link all those separate interactions to one customer or prospect, but this 12 isn’t always easy to do. However, new tools are becoming available to map the complex and personal customer journeys: eg. staying logged in on Gmail, Facebook or on the website/app ensures that you remain recognisable on all your devices. This way, a web analysis programme detects that the evening visit from a tablet device was done by the same account as the short visit earlier that day that came from the mobile newsletter. This is one of the most important developments: analysis is shifting from ‘device’ level to user level, which allows you to observe a real customer journey. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 526 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 527 Tracking code and Google Tag Manager Before you can see how visitors behave on the website, it is essential that you implement the tracking code for the analytics package. This is a piece of JavaScript code that forwards data. You can implement this in two ways: 1 Directly into the code: the tracking code can be added straight into the website’s code. The downside of this is that every change that you’d like to make to a measurement, must be done by someone with sufficient IT knowledge. However, it may be beneficial to add certain tracking codes as high up as possible in the website’s code in order for them to be loaded faster. 2 Tag manager: An increasing amount of companies are using a tag manager. A tag manager is a tool that allows you to manage all your tags (analytics, Digital Marketing, etc.). Previously, all these tags were added straight into the website’s code, one at a time. The problem with this is that IT departments often only deploy a new release once or twice a month. This would mean that if you want to adjust one of your tags, you may have to wait for a month. To tackle this problem, a tag manager was developed. You only have to add a script once: that of the tag manager. After this is done, you can implement and modify as many tags as you’d like using the tag manager. Market insights are not simply gained by solely conducting digital measurements. Avinash Kaushik (2007) depicted this very well in Figure 12.1. He states that by measuring clicks, you only find out what happens online (What) and that by combining the results of digital measurements, you discover the extent of the impact (How much). By experimenting and testing, in combination with listening to the customer, you are able to find out what the reasons are for digital consumer behaviour (Why). In order to listen to the customer, you need more than just measurements, you also need market research (see Chapter 3). In order to make good marketing decisions however, you will also need to know what your customers’ alternatives are (What else). For this, you will have to take stock of the competition and continue to monitor them. Only once you have combined all this information, can you actually gain market insight. Kaushik refers to this as Web Analytics 2.0 (see Figure 12.1). Web Analytics In this chapter you will find more information about the first two steps of this 2.0 model. To get a full understanding of the market, you can combine this chapter with the research and analysis methods from Chapters 2 and 3. FIGURE 12.1 Web Analytics 2.0 Measuring of clicks What 12 Multivariate analysis How much Experimenting & Why testing Listening to customers What else Research competition The gold! Insights Source: Kaushik, 2007 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 527 26/11/20 5:01 PM  528 © Noordhoff Uitgevers bv There is an important precondition for starting an analysis: ask yourself what you hope to gain from the analysis. What will you ultimately be able to do with the results of the analysis? Will these results lead to actions, or is it just nice-to-know? In that case: save your efforts and look for Actionable analyses that will lead to action. You are looking for ‘actionable insights insights’. Online Dialogue has compiled a list of the top 10 analysis questions: 1 Which channels generate the most traffic to the website? What is the aim and the conversion rate for each of the different channels/media? What is the ratio between paid and unpaid traffic? 2 What keywords lead to people visiting the website? And what is the conversion rate of those keywords? 3 On which days and at what time is the website visited most often? And on which days and at what time are people most likely to convert? 4 What browsers and devices are used most often and is there a difference in the number of conversions between these browsers? 5 What are the top 10 landing pages in terms of visitor counts and conversions? What does the clicking behaviour look like on these pages for the most important segments? 6 What are the most visited pages on the website in general? On which pages do a lot of visitors leave the website? 7 How do people navigate on the website? Use main navigation, filters and so on. 8 What keywords do people use on the website? And how successful are those keywords? 9 What does the funnel look like and where in this funnel do the largest group of exits appear to be. 10 Which channels are responsible for the final conversion? Which channels are responsible for orientation visits (conversion attribution)? § 12.2 The Digital Marketing funnel translated into performance indicators One of Digital Marketing’s greatest charms is that the process is measurable from beginning to end. However, in order to properly monitor and fine-tune the Digital Marketing activities, organisations must determine Key performance in advance what the key performance indicators (KPIs) are. These are the 12 indicators indicators that they use to assess whether their Digital Marketing activities KPI are successful. Eg. a KPI could be the number of visitors per day or the monthly turnover of the ecommerce site. These KPIs vary per industry and per organisation and are highly dependent on the website’s purpose. You can imagine that an online newspaper uses very different KPIs than a website that sells mobile phones. In Chapter 3 you have read about the Digital Marketing funnel (see Figure 12.2). KPIs can be defined for every level of this funnel. For the sake of 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 528 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 529 convenience, from this point onwards we will refer to ‘the website’, but it may also be a mobile site, a landing page or an app. FIGURE 12.2 The Digital Marketing funnel Visit Captivate Decide Order Pay Bind In the following subsections, you will be able to read about the KPIs that are used in each phase. Then, from Section 12.3 onwards, we will talk about how to use digital analytics in practice. 12.2.1 Visit Sometimes, companies will comment on visits in the press, eg.: ‘We have 100,000 visitors per month.’ What does this mean? A visitor is someone Visitor (or actually a new cookie) who goes onto your website. If this visitor returns within thirty days, then they are one visitor who visits twice. A visit means Visit that someone enters a website and performs activities on the website within 1,800 seconds (= half an hour). Suppose someone returns to the website after half an hour has expired, then this is registered as a new visit, but still as one visitor. If someone goes to a website, receives a phone call after five minutes, has a ten-minute conversation and then continues with their website visit, this is still considered one visit. In addition to visitors and visits, there is also a third concept that is often mentioned in digital analytics: page views. During one visit by one visitor, Page views multiple pages are often viewed. Within two days, one visitor may visit three times and view a total of 18 pages. 12 EXAMPLE 12.1 One or two visitors, one or two visits? Suppose you are busy looking for a holiday on a website. You have found a holiday that you want to book. Your partner also agrees. Suddenly the doorbell rings and you leave your computer to open the door. When you return, your partner has already booked the trip. In digital analytics, this is regarded as one visitor and one visit, but in fact this should be two visitors with two visits. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 529 26/11/20 5:01 PM  530 © Noordhoff Uitgevers bv The same problem may arise the other way around. If you visit a website on your mobile phone today and on your laptop tomorrow, digital analytics records this as two visits by two visitors. Whilst in fact, it involves two visits from a single visitor. You are, of course, the same person. The reason this happens is because you are not identified as a person, but on the basis of cookies. You use a different browser on your mobile phone, in which a new, unique cookie is installed. You can however be identified anyway if you are logged into an account or in Gmail/Facebook. Example 12.1 illustrates that there is not always one truth in digital analytics. It is a matter of interpreting the data the right way and not just making assumptions indiscriminately. A hundred thousand visitors per month are basically 100,000 individual people who have visited the website at least once. However, if people use each other’s devices, then those figures are no longer completely accurate. The same applies to visitors that are not logged in and visit the website on multiple devices. Important KPIs for the ‘visits’ phase are the reach, the click rate, the percentage of new visitors and the acquisition costs. Reach The reach of a website is the percentage of the target segment that uses the website. Eg. ‘The reach of Huffingtonpost.com on the internet further decreased in the past year’ means that of all people who read an online newspaper, there is a smaller percentage reading Huffingtonpost.com than the year before. Number of unique users of the website in period T Reach = Total number of unique users in the target segment A visit occurs when users actually end up on the website. The definition of reach however, depends on the circumstances that it is used in. For ads, eg. Google considers the unique reach to be the total number of people to whom an ad is displayed. However, the term reach is also used by Google in a very different context, namely the level that a statistic from Google Analytics relates to: user level, session level, product level or hit level. A hit occurs when the Google tracking code is triggered by a visitor interacting with the website. In short: always properly check the definition used for reach before you draw your conclusions based on these figures. 12 Click rate Another important KPI in this phase is the click rate or click-through rate: the number of clicks on an email, advertisement or other expression in relation to the number of impressions of this expression. ‘The click rate of the banner was 2%’ means that two out of one hundred people that could have seen the banner, clicked on it. This has nothing to do with whether people actually saw the banner: it’s about the banner being displayed on a page that they opened. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 530 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 531 Click rate = Number of clicks Number of impressions The percentage of new visitors to a website is also an important indicator Percentage of for the effectiveness of Digital Marketing communications. A visitor is new visitors ‘new’ if they are visiting the website for the first time in thirty days. Of course, what the ratio should be between the number of new visitors and existing visitors depends on the type of website. Online retailers want to engage their visitors. Whilst orientating or during their decision-making process, people who are looking to buy something will usually view the website multiple times before they make their final purchase. In that case, a high percentage of new visitors can be unfavourable. That’s because this means that many people only visit once. But if the ecommerce retailer has just started a campaign to attract more visitors to their website, the percentage of new visitors will be relatively high and that is a positive thing. In essence, a news website wants visitors to come back daily to read the latest news but will also want an increasing number of new visitors. So, should such a website have a high or low percentage of new visitors? That is, of course, completely dependent on the objectives. The acquisition costs are the cost per click or cost per impression. The Acquisition costs costs of an individual expression, eg. a banner, are expressed in CPM (cost per Mille) or CPC (cost per click). Costs of the expression CPM = CPM (. × the number of impressions) Costs of the expression CPC = CPC Number of clicks In Figure 12.3, the CPM for all expressions together is €15.19 / (0.001 × 12,122) = €1.25. In other words: it costs 1.25 euros to show your banner to 1,000 people. The CPC is €15.19 / 92 = €0.17. In other words: it costs 0.17 euros to receive one click on an advertisement. 12 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 531 26/11/20 5:01 PM  532 © Noordhoff Uitgevers bv FIGURE 12.3 Key Performance Indicators CTR en CPC Click ratio Cost per click Ad group Status Standard Max. CPC for Max. CPC for the Amount Impressions CTR Average. Cost ? max. CPC self-chosen Display Network ? clicks ? CPC placements ? ? Excursions Suitable € 0.20 € 0.20 € 0.20 78 11,861 0.66% € 0.17 € 13.24 for the elderly Excursions Suitable € 0.20 € 0.20 € 0.20 14 261 5.36% € 0.14 € 1.95 for the elderly Total - all, excluding removed ad groups 92 12,122 0.76% € 0.17 € 15.19 Total - search ? 67 4,156 1.61% € 0.16 € 10.83 Total - Display Network ? 25 7,966 0.31% € 0.17 € 4.36 Total - all ad groups 92 12,122 0.76% € 0.17 € 15.19 An overall indicator of the acquisition costs is made according to the costs per contact: the marketing expenses in a period per obtained visitor. expenses in period T Costs per Costs per contact = contact visitors in period T Suppose that, during the week depicted in the previous example, €200 was spent on general advertising costs and this led to 108 extra visitors, then the costs per contact for this period are (€15.19 + €200 / (108 + 92) = €1.08. Connect rate In addition, many digital marketers measure the connect rate: the actual number of people who visited a page, as opposed to the number of people who clicked but did not visit the page. This could be because, eg. it took too long for the page to load or because an error occurred. A visitor on a forum wrote: “I have a rather peculiar problem. I want to visit a certain website (belonging to a travel agency), but I only get to see half a page, showing the message ‘Ready with errors’. The links on that page also not doing what they’re supposed to. When I asked around in my neighbourhood, I found out that all my neighbours are able to access the website without any problems. I, on the other hand, am 12 unable to.” 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 532 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 533 App analytics The number of people that visit your website on their mobile phone continues to increase. Analysing the results of mobile marketing is also known as mobile analytics. People do not only use mobile websites, but also apps. Many apps, eg. Spotify or Facebook, require visitors to be logged in. Within app analytics, the metrics that are reviewed differ slightly from the ones used for web analytics. Important metrics are, eg. the amount of app-installs (the number of times an app is installed), the number of log-ins, the number of pages viewed, the session duration, the number of advertising clicks and how often the app has been used in a week. The Apple and Android store statistics are also important, eg. the number of downloads, installations and searches. Ultimately, your aim is to get a unified picture of the visitor whether it is via the app, mobile website or desktop website. This can be done using universal analytics, this is the version of Google Analytics that revolves around the visitor. 12.2.2 Captivate When someone visits the website, it is important to captivate them so that they can begin to take the first steps towards a purchase. Retention is the Retention extent to which the site holds onto the visitors. Two simple indicators are: Depth of the visit (session): how many pages did the user see? Average duration of the visit: how much time did the user spend on the website? The objectives must be taken into account for both indicators, in order to interpret them properly. Eg. a news website benefits from many pages per visit as this will increase the value of the advertisements. An online retailer will also benefit from many pages per visit, provided this is a visitor who is still in the orientation phase. After all, the purchase process must be smooth, otherwise potential buyers will pull out. Someone who already knows what they want to buy, will want to place their order in as few steps as possible, ie. via as few pages as possible. The value ‘average duration of a visit’, reflects the amount of time a visitor spends on the website on average. If this is extremely short, then the challenge is to make the website more interesting for your visitors. A very long stay however, is not always indicative of anything good. Visitors may 12 have difficulty finding the information or content they want. The relationship between the duration of a visit and the pages per visit is important. By comparing these two values, you are able to deduce a visitor’s average time spent per page. The average time someone stays on a page says something about their behaviour, but precisely what that is depends on what can be found on the page. If a customer stays on the contact page for a short amount of time, that’s a good thing. If they remain on a product page for only a short amount of time, it may mean that they are not interested. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 533 26/11/20 5:01 PM  534 © Noordhoff Uitgevers bv Bounce rate The bounce rate is the percentage of visitors who visit only one page and then leave the website again. If this percentage is large, then the website is not interesting and probably not relevant or not trustworthy enough for the visitor. In practice, this also appears to be more nuanced. Eg. if someone is looking for a phone number and finds it immediately, they will leave quickly, but still be satisfied. A significant reason for a high bounce rate may be the mismatch between a banner and the website. It is relevant eg. that the layout and content in the banner matches the website. Otherwise, visitors may wonder if they are on the right website after they have clicked through. You were able to read more about effective advertising in Chapter 7. Sometimes, a high bounce rate is inherent to the type of website or web page. If there are many outbound links on a page, eg. on a home page like Kadaza.co.uk, then a high bounce rate is a good sign. 12.2.3 Decide It is extremely difficult to see whether a visitor has finally decided that they want to purchase a product or service via the website but have not yet put in an order. An indicator of the number of people who have decided to Abandonment purchase, but for whatever reason fail to follow through, is the abandonment rate rate. This is the percentage of visitors who fail to complete a certain step in the ordering process. An example of such a ratio is the ‘cart abandonment rate’: the percentage of visitors who fill their shopping cart, but then do not complete the order. Some causes of a high abandonment rate could be: an overly lengthy checkout process, a disappointing number of options for paying, sudden display of shipping costs or a long delivery time. For some products, the visitor might return later, after they have consulted someone else about their decision. Number of visitors who add items to their Cart abandonment shopping cart but do not order Cart abandonment rate = rate Total number of visitors who add items to their shopping cart in period T 12 Cart abandonment 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 534 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 535 In order to get an impression of the decision-making process, organisations will sometimes measure the velocity: the speed at which the average buyer moves from one step in the Digital Marketing funnel to the next. 12.2.4 Order For websites that sell products or services, the order (and the subsequent payment) is the most important goal. In addition to the usual KPIs eg. turnover and sales per period, the conversion rate and conversion costs are particularly important at this stage. At this stage, the conversion rate expresses what percentage of the visitors places an order. If the website in question is not an ecommerce site, the conversion rate can be, eg. the percentage of visitors who apply for a job, request a leaflet or a quote, or who register for a newsletter. Number of visits resulting in a goal in period T Conversion rate = Conversion rate Total number of visitors in period T Visitors Ecommerce Conversion Rate Target value per visit 72.273 1.59% € 1,91 % of site total: 100,00% Siteav: € 1,91 (0,00%) Conversion rate of 1.59% The previous web statistics show the number of visitors (72,273), the conversion rate (1.59% of the visitors made a purchase) and the target value per visit (€1.91). The target value per visit is the sum that the average visitor brings in. Online retailers also look at the conversion costs in this phase: the cost per sale (CPS) or cost per order (CPO). CPO Costs of the expression CPS = CPS Number of items sold The CPS must of course always be lower than the profit made on the product, otherwise the result is negative. By comparing the CPS of different instruments and websites, the most effective Digital Marketing communication method can be chosen. 12.2.5 Pay 12 If the order is placed but the visitor does not pay, then this is also a form of abandonment. On most websites the order will not be processed in such a case. Discontinuation in the very last phase of the ordering process is also called ‘check-out abandonment’, this ratio can also be determined. Number of visitors who do not fully Check-out complete an order abandonment Check-out abandonment rate = Total number of visitors who have rate ordered a product in period T 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 535 26/11/20 5:01 PM  536 © Noordhoff Uitgevers bv There are often many discontinuations in this part of the process. There are various reasons this could happen. Examples could be: ‘I still have to consult with my partner’, ‘I am now on my mobile, I will finish it at home on my laptop’ or ‘I suddenly see there are delivery costs, I had not seen that before’. Visitors in this phase probably are willing to buy from you because they have found their product or service and have clicked through to pay. Yet, something has prevented them from completing the order. To find out, you could, eg. ask them to complete a survey. An organisation that rents out holiday homes discovered that a significant percentage of the people who indicated that they wanted to rent a house via the reservation page did not complete this reservation with a final confirmation and payment. By contacting these people, they discovered that there were additional questions and wishes that customers felt unable to voice properly online. By solving this, the check-out abandonment rate went down considerably. In this example you can clearly see the combination of number analysis on the one hand, and knowledge of your customer, by contacting the customer personally or using an online survey or feedback tool (feedback button on the site), on the other. 12.2.6 Bind Once a visitor has ordered something once, the digital marketer would like them to come back. In order to establish loyalty, you must look at four indicators: frequency, recency, period of use and monetary value. Frequency Frequency measures how often a user has visited the site. Recency is the Recency time between visits to the website and how long ago the visitor was active Period of use on the site. The period of use is the number of hours a visitor spends on Monetary value the website per visit. The monetary value refers to the amount that a visitor spends within a period. Number of visits in period T Frequency = Number of unique visitors in T Recency = Time that has elapsed since the last visit. Total amount of time spent on the website Period of use = Number of visits in period T Monetary value = (Number of orders × average order value) Per period 12 Universal In Universal Analytics, the digital analytics system introduced by Google in Analytics 2013, it is possible to determine the customer value of visitors from this basic data. An indicator of the extent to which visitors are loyal to a website is the Stickiness stickiness. You can calculate this by looking at how often visitors use the website on average, how much time they spend on the website and how many visitors the website receives in a given period. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 536 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 537 Stickiness = frequency × time period of use × total reach of the site in time period T Attrition occurs if after a fixed period of time a customer is no longer active, Attrition whereas they were before. Churn is an indicator of the number of departed Churn website users in a given period. It is the opposite of retention. Total number of departed website users in period T Churn = Total number of users in period T Data quality Something that is regularly underestimated by companies is the data quality. In this case, we mean the quality of the data within the analytics package (eg. Google Analytics). All kinds of decisions are taken on the basis of this data. If, eg. it turns out that a certain advertisement within Google performs better than another, it may be worth using more of that type of advertisement. You do then have to be sure that the data is correct and not that some of the ads have been clicked on by colleagues, lowering the conversion rate. You should exclude colleagues’ visits as this data is not representative. In addition, a lot of what is called ‘referral spam’ occurs these days. These are bots that send hits to analytics, purely to pollute your data. You should exclude this from your data, eg. by setting up filters on certain IP addresses that these bots use to send their traffic from. § 12.3 Different traffic sources within digital analytics Traffic sources are the sources of origin for visitors to the organisation’s website: places where click-through options are available on the website. Within digital analytics, there are usually three groups of traffic sources: direct traffic, referred traffic and search engines. In addition, there are ‘other’ sources. Below, you will be able to read how to handle the digital analytics for every traffic source and what the most important pitfalls are. 12.3.1 Direct traffic Direct traffic are visitors who enter the web address of a website directly Direct traffic into the address bar of the browser. Another type of direct traffic is favourites. A visitor who is seen as direct traffic is a visitor who is already familiar with a website or at least made a conscious choice to come to the 12 website. Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bounce rate Direct 323,557 5.74 00:03:53 49.04% 29.25% % of total: 35.26% (917,631) Siteav: 5.60 (2.58%) Siteav: 00:03:50 (1.42%) Siteav: 36.87% (33.02%) Siteav: 34.38% (–14.91%) Direct traffic in Google Analytics 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 537 26/11/20 5:01 PM  538 © Noordhoff Uitgevers bv In the example above, you can see data regarding direct traffic. The bounce percentage is rather high, because 29.25% for direct traffic is a little unusual. This requires further investigation: eg. the URL of the website may be very similar to that of another website. An example of this is: www. vouchercodes.co.uk and www.myvouchercodes.co.uk. This can cause user (and even customer) confusion. What also stands out here is the high number of new visitors relative to returning visitors (49.04%). That is special, you would expect a low number of new visitors for direct traffic. The organisation may have implemented an offline media campaign, which includes the web address. Digital analytics is more than viewing statistics. You learn very little if you are not aware of the backgrounds. In general, conversions via direct traffic are always higher than via the other traffic sources, because someone who consciously chooses a site is already further along in their decision-making process and was probably familiar with the brand already. Pitfall Nowadays many visitors type the name of the website in the search bar of a search engine, but that is not counted as direct traffic. They then click on a search result or a paid search engine ad. The actual number of direct visitors can be a lot higher than the statistics show. 12.3.2 Referred traffic Referred traffic Visits via a link on another website fall under referred traffic. It is interesting to find out from which source a visitor came on the website and what this visitor subsequently did on the website. Referred traffic can sometimes be very diverse in terms of bounce rate and the number of pages being visited. The following image shows that the bounce rate is relatively low, and the quality is better than that of direct traffic (more pages per session and longer average session duration). Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bounce rate Referral 2,455 11.20 00:14:30 2.12% 17.84% % of total: 0.27% (917,631) Siteav: 5.60 (100.02%) Siteav: 00:03:50 (277.84%) Siteav: 36.87% (–94.25%) Siteav: 34.38% (–48.10%) Referral traffic in Google Analytics This also illustrates how a digital marketer can control quality if they keep an eye on such traffic. The marketer can use the analytics programme, eg. to see where the visit comes from, on which page the visit lands and what is then done there. A targeted approach can then be used for directing 12 traffic to the website and turning the visits into conversions. Suppose, eg. that many visitors enter a website about indoor plants via women’s magazines’ websites. This could be a reason to try to get more links or to place content in other women’s magazines, or more generally: on websites for women. You can also use referred traffic to find websites that are interesting to advertise on. 12.3.3 Search engines The search engine as a traffic source is divided into sub-groups in almost all web analysis programmes. Eg. you have the traffic from organic results 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 538 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 539 (SEO) and the paid results (PPC). We will now discuss some sub-groups of traffic sources. Traffic from organic results In the results of organic traffic, you often you see that more or less direct traffic is also included. Visitors use the search window on Google and enter part of the domain name there. This is seen as organic traffic, while this is actually direct traffic. Companies that do not do a lot of SEO will score better from this traffic source than companies that do a lot of SEO. That sounds strange, but it is a consequence of ‘disturbed’ direct traffic. If a website does not do a lot of SEO, they will usually be found using the company name or a variant of it. Visitors using this search term genuinely have the intention of visiting the website and will therefore generally stay longer and have a lower bounce rate. If you do a lot of SEO, then you will generate more visitors in what is called the ‘long-tail’. These are visitors who search on specific search terms instead Long-tail of on general search terms. Eg. they have searched for ‘car sleeper train in winter to Innsbruck’ as opposed to ‘train to Austria’. Visitors that perform a very specific search, are often critical of the information and inclined to leave quickly (read: bounce rate). They view the page they need and with that, your website has met their information needs. A low number of pages per visit and a short average visit duration is not unusual for this group. Pitfall Visitors who search in a very targeted manner, as well as visitors who orient themselves extensively on the website, enter via the SEO traffic source. It is important not to see these visitors as a uniform group. Paid search engine traffic The image below shows an example of paid search engine traffic. It is an optimised PPC campaign. You can tell because it has a low bounce rate and an average time on the website that is reasonably proportional to the number of pages per visit (5:09 = 309 seconds / 8.30 = 37.2 seconds). If a paid search campaign is not optimised, the collected data won’t be quite as good. Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bouncerate Paid Search 85,497 8.30 00:05:09 23.79% 17.31% % of total: 2.20% (3,888,763) Siteav: 5.84 (42.07%) Siteav: 00:03:54 (32.03%) Siteav: 37.03% (–35.75%) Siteav: 33.49% (–48.30%) PPC traffic (search network) in Google Analytics Pitfall 12 You often see that digital marketers spend a relatively high amount of money on Ads, unnecessarily so. Low cost per click, or a high CTR (click-through rate = the ratio between impressions and clicks) are then the only indicators for continuing a campaign. The marketer increases the daily budget or the cost per click on the assumption that more turnover will be generated on the site. Spending more money on advertising is often easier than analysing the behaviour on the site or implementing a content strategy on social media. That ‘more Ads’ will attract more visitors is certainly the case, but are these extra visitors of good quality? Could not you have attracted people to your site at lower cost? 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 539 26/11/20 5:01 PM  540 © Noordhoff Uitgevers bv Advertising for your customers? One of the special things about Google Ads is that these results appear in Google ‘above’ organic (unpaid) results. Many people who have been your customers for a long time already, use your expensive Paid Ads as a sort of bookmark; a quick way to get to your site without typing the URL. The search engine benefits from this more than if a visitor types the URL in the address bar. That is why the cursor is already ‘ready’ in the search bar. You then spend a lot on advertisements that are clicked on by visitors who are already your customer or are a regular visitor. By looking closely at the behaviour of people who arrived on your website via paid ads, you will see that they do not convert, but eg. log into the ‘My environment’ or have come to your site many times before. A good technical analyst can fit the advertisements with a piece of code where people who have ever logged in before or have visited the website more than 10 times are not targeted. For some companies, the percentage of customers who click on ads is 85%. So they only use 15% of their advertising budget for recruiting new customers or visitors. Display network Display network The display network is a paid form of advertising that includes everything outside of the search engines (see also Section 7.3). Eg. ads on ITV.com or YouTube are display ads. In the following picture you can see an example of a traffic source that consists of so-called display visits. This kind of visit is often characterised by a high bounce rate, a short time spent on the website and few pages per visit. This is logical because visitors in the display network were actually busy doing something else, before they were tempted by the advertisement to visit the website. There more care that is taken when selecting the websites that the ads are placed on, the more favourable the results will be. It is important to mark or tag the ads in the Analytics programme UTM code separately with what is called a UTM code. This allows you to measure which variant of the banner or which location works best on any given site. Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bouncerate Display 45,896 2.64 00:01:23 40.34% 74.10% % of total: 5.00% (917,631) Siteav: 5.60 (–52.87%) Siteav: 00:03:50 (–64.11%) Siteav: 36.87% (9.43%) Siteav: 34.38% (115.55%) CPC or CPM traffic (display network) in Google Analytics Pitfall 12 If a digital marketer is running a campaign with a high bounce rate, there could be several causes. It does not necessarily mean that the wrong websites have been chosen for the campaign. Eg. the banner can be misleading or even ‘too nice’. If a banner is very exciting or appeals to the imagination, visitors will also come to your website without any intention of purchasing anything. It is also possible that a landing page does not match an advertisement. Also, very few pages per visit combined with a high average time on the site, can be a sign that something is wrong. Eg. it may be that the page is not clear or that the visitor is stuck. So your visitor does want to make a conversion, but simply cannot figure it out. Analysis, visitor research or A/B testing of banners and landing pages could provide a solution in this situation. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 540 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 541 A/B tests, multivariate tests and experiments. What are they? Which images, texts and buttons influence the conversion on websites? Using A/B testing and multivariate testing web pages can be optimised for the best possible result. When conducting an experiment, different elements of a web page are compared. Think of headlines and headers, images, text, call-to-actions and page layout (Jansen, Klaassen & Visser, 2015). It is important to have a clear idea beforehand about the factors that influence the conversion based on the objective of that specific web page and the pre-analysis that has been done. Interesting insights into the data, eg. a high drop-out rate, a lot of clicks between product and category pages or a short time spent on the page, can be input for a hypothesis that is then validated using an A/B test. In principle, only one element is tested at a time in A/B testing. More than two variants of an element can also be tested, such a test is also called an A/B/n test. In practice, digital marketers often change multiple elements within an A/B test. This has the advantage that there are bigger differences between A and B. The possible impact might also then be bigger. The disadvantage is that if a positive or negative result comes out of the A/B test, you do not know exactly which element caused it. In multivariate tests, multiple variants are tested simultaneously, eg. A/B/ C/D. In such a case, you test which combination of elements yields the highest conversion. In multivariate tests, eg. the header and a picture can be tested simultaneously. It is also possible to determine which element has the greatest influence on the conversion. If multiple variants are tested (eg. four), this means that the number of visitors per variant will be lower than with an A/B test (two variants). This means that this a type of test needs a longer running time or a higher uplift, a greater positive effect, in order to be able to recognise a significant difference. These tests usually use specific software eg. Visual Website Optimizer, Optimizely, AB Tasty or Google Optimize. Google Optimize is available in a free version (in addition to the paid version). The software sends visitors to different versions of the page and calculates which elements yield the highest conversion. Seasonal or temporal influences are excluded because different versions of a web page are tested simultaneously. Here is an example in the case of a luxury hotel website: On the page containing the form, it was found that approximately 50% of the visitors dropped out (ie. exited the site). To understand why visitors drop out here, a survey was conducted among 12 visitors who appeared to be about to leave the page (mouse outside the active browser screen). This is also called an exit survey. The question was: ‘Would you like to help us by indicating why you are about to leave the website?’ Perhaps surprisingly, 100 visitors answered this and about 30% of them complained about the fact that check-in and check-out times were not mentioned anywhere. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 541 26/11/20 5:01 PM  542 © Noordhoff Uitgevers bv These times were then simply added to the page quoting the price for the room. This optimisation resulted in 5.5% conversion increase. This means that on the B variant, 5.5% more visitors eventually booked a room compared to the A variant. 5.5% more conversion means 5.5% more turnover: a considerable sum! Before you start optimising and A/B testing, it is important to do good research. The combination of quantitative and qualitative research is ideal in this situation. Data shows you visitors’ behaviour (in the example, the 50% drop-out percentage), but this does not tell you why that is the case. By carrying out qualitative research you can get to the bottom of the visitors’ motives (in the example, an exit survey). 12.3.4 Other traffic sources The other traffic sources are usually categorised in campaigns and are also viewed using the so-called UTM codes. If you do not use UTM codes, this traffic will be seen as referred traffic. We will use newsletters and affiliate marketing as an example. Then we will discuss conversion attribution. Newsletters The traffic source ‘newsletters’ is a tricky one. You have read more about email marketing in Section 6.5. Several factors depend on the success of the newsletter and influence the statistics. Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bouncerate Email 10,823 1.75 00:00:46 46.49% 52.58% % of total: 10.89% (99,363) Siteav: 3.60 (–51.32%) Siteav: 00:03:46 (–79.78%) Siteav: 41.80% (11.24%) Siteav: 30.36% (73.19%) Newsletter as a source in Google Analytics In a qualitatively good newsletter system an analysis feature will always be included. The digital marketer can check whether the mail has been opened, which article of the newsletter is clicked on most often and sometimes you can even see who has clicked on a particular article. You can also see what email the client uses. Data eg. this will help you to use the knowledge of such a newsletter system and optimise it for frequently used devices. If your email is opened on an iPhone in 60% of cases, you can advise the marketing department to optimise the length of the header. The site can then also be 12 optimised for navigation on a mobile with a mobile template or a site can be built that is ‘responsive’ and adjusts the content to the screen size. How well a newsletter performs, depends entirely on the content of the newsletter and what the organisation’s connection to the recipients is. If the newsletter is strongly focused on the needs of the recipient and the target audience finds the newsletter interesting, then the quality will be significantly higher than if the digital marketer approaches large numbers of people in an unfocused way. The example above shows that on average this newsletter performs much worse than the rest of the traffic sources. Compared to the costs 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 542 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 543 of advertising, the low costs of creating and sending a newsletter mean that this is a frequently used method. As a result, people are receiving an increasingly greater amount of emails and they don’t pay as much attention to them. Pitfall You can conclude that the organisation mentioned should not send newsletters or at least change its approach. But, the recipient’s behaviour can be explained logically. In principle, the marketer determines the moment the newsletter is sent. Often a subject line or the attractive layout acts as a trigger to click through but having to read the landing page comes at an inconvenient moment, or the recipient actually has no intention of making a purchase at that time. Often, those who click through are curious visitors. However, these visitors can contribute to a longer-term conversion. This is called conversion attribution (see later in this section). Affiliate networks Affiliates are websites that receive a commission for, eg. a purchase or a registration. You have read more about this in Section 7.5. An affiliate visit can best be compared to referred traffic, with one significant difference. Referred traffic often has link building as a primary goal, so that a better score is achieved in the search results. Affiliate marketing is primarily concerned with conversion, eg. a notification or a purchase. Affiliate sites often present themselves as a ‘friend’ of the visitor, as a critical comparison site, but actually earn their money from a commission from the sites to which they refer visitors, eg. moneysavingexpert.com and moneysupermarket.com. Default Channel Grouping Sessions Pages/sessions Av. session time New sessions Bounce rate Affliates 86,342 5.31 00:04:04 11.40% 31.65% % of total: 9.41% (917,631) Siteav: 5.60 (–5.16%) Siteav: 00:03:50 (5.91%) Siteav: 36.87% (–69.08%) Siteav: 34.38% (–7.92%) Affiliate networks as a source in Google Analytics In principle, the affiliates should score well, both in the number of new visitors they deliver and on the depth of the visit. Using analytics, digital marketers remove badly scoring affiliates from their network and determine the price per lead. Online advertising, what should you pay attention to? As a digital marketer, you will often be asked for your opinion on online advertising. You will also be approached by sellers who promise you great results if you advertise on their website. What should you pay attention to in 12 these situations? Advertising on other websites has been made very transparent by digital analytics. As a rule, you can find this traffic at the ‘referred traffic’ source, but you can also attach tracking codes, eg. UTM codes. You can then view what kind of traffic an advertiser forwards, what exactly this has yielded and what exactly it cost. You can also compare various banners and banner positions to achieve maximum return on investment. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 543 26/11/20 5:01 PM  544 © Noordhoff Uitgevers bv An advertising salesperson wants the most revenue with as little effort as possible. This means that an advertising salesman prefers to sell a banner at a fixed price. As digital marketer, you want to pay as little as possible for a guaranteed success. That means that you actually prefer to pay on a CPS (cost per sale) basis, in that case you only have to pay if you have actually sold something. In between are all kinds of variants eg. paying per click (CPC = cost per click), paying per 1,000 impressions of a banner (CPM = cost per mille), or paying per registration, application or registration (CPL = cost per lead). It’s common knowledge that advertising salespeople tend to make things appear more favourable than they actually are. As a digital marketer, you obviously mustn’t fall for this. What to look out for: The advertising salesperson: ‘We have 100,000 visitors on our website!’ Visits and visitors are great, but is it the right target audience in terms of age, willingness to buy, phase of orientation process? The advertising salesperson: ‘We have a lot of unique visitors!’ You often hear this, but you already know that many unique visitors are not necessarily a good thing. The advertising salesperson, after an initial analysis of the results: ‘We deliver 1,000 people a month to you!’ Are they the ‘right’ visitors? Advertisers tend to focus on delivering people to your doorstep, whereas you are focused on those people who help you to meet your KPIs. As an analyst or marketer, you will therefore have to ‘follow’ visitors on your site to determine whether an advertiser provides you with the right ‘type’ of visitors. Prices of ads do not say very much, other than that you have to be able to pay them of course. More important is the cost per conversion, this must be relatively low compared to the expected revenue and compared to other channels, marketing or sites where you could just as easily place your banners. Finally, always request a ‘test’ period. If the advertising salesperson is so convinced of their own product, then this advertising salesperson will have no objection to testing an ad for a month. You can then look at your web analysis to examine the ad’s effect and provide management with valuable advice (based on Kremer, 2012). Conversion attribution Conversion Conversion attribution simply means that you do not assign a conversion to attribution one traffic source, but to multiple. Suppose someone makes a purchase in response to an ad via Google Ads, you would normally be able to attribute this conversion to this ad. This would also make you more inclined to spend extra money on Google Ads and less on other ads or marketing. 12 But a visitor comes to a website in different phases and often needs several visits to eventually come to a conversion. The visitor may have entered via a newsletter during the first visit, then via a referred link Referral (referral) and finally via the Google Ads campaign. This conversion can therefore be attributed to three traffic sources, namely newsletter, referred Assisted traffic and Google Ads. Google Analytics also shows so-called assisted conversions conversions: key figures that show which online channels have played a role in the preliminary stage of the conversion. As a result, a digital marketer is better able to estimate how conversions come about and might find that they do not have to spend all their budget on Google Ads, because newsletters are a very important source of conversions. 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 544 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 545 The image below shows an example of a conversion attribution summary from Google Analytics. Conversion attribution in Google Analytics In Google’s Universal Analytics and also in the App + Web version of Google Universal Analytics, a user ID is used to measure the behaviour of users across various Analytics browsers, devices and channels. In the older versions of Google Analytics, a visitor’s visits were measured separately, and you were not able to fully establish the complete ‘customer journey’ that precedes the conversion. There are a number of different conversion attribution models: 1 Last click model: this is the traditional payment model. 100% of the Last click model conversion value is assigned to the last click. This model is easy to calculate but is less than optimal for basing your budget distribution on. All other clicks are given no value at all, while they have contributed to the conversion. If consumers type your company name in Google and click on your ad as a result of a radio commercial, the source of traffic in the ‘last click’ model would be a paid search machine. A specific form of the last click model is the last-non-direct-click model. Sometimes a Last-non-direct- visitor has been in contact with the website or another digital click model communication expression several times but goes directly to the website at the time of purchase. Using the last-non-direct click model, direct traffic is disregarded, and the conversion is fully allocated to the last channel the customer clicked through from, prior to the conversion. Google also uses the last Ads click model, where 100% of the conversion Last Ads click value is attributed to the most recent Ads ad that the customer clicked on model prior to the conversion. 12 2 First click model: 100% of the conversion value is assigned to the first First click model click. This is the opposite of the last click model. In this case, the model is also fairly simple, but again sub-optimal. A lot of touchpoints are ignored in the allocation of conversion value. Someone clicks on a banner and comes to your site. By placing a cookie, you will recognise them the next time they come via a comparison site. But the ‘click’ then goes to the banner. 3 Linear model: in this model the total conversion value is evenly Linear model distributed over all clicks in a conversion path. In this model, you could wonder whether all clicks deserve the same value. Isn’t a direct visit 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 545 26/11/20 5:01 PM  546 © Noordhoff Uitgevers bv worth more than a visit from a banner? Eg. someone clicks on a banner and comes in again later via a comparison site and finally via a search engine. The attribution is then 33.3%, 33.3%, 33.3%. Time decay 4 Time decay model: this factor takes into account the factor of time. model Clicks that took place on the same day of the conversion will receive more value than the clicks that took place more than 7 days ago. If the banner click was 10 days ago and the comparison site, search and purchase were all on one day, then the attribution is 10% for the banner, 40% for the comparison site and 40% for the search engine. Position based 5 Position based model: the first and last click of a conversion path get more model value than the intervening clicks. The attribution would then be 40% for the banner, 20% for the comparison site and 40% for the search engine. 6 Own attribution model: in many analytical packages it is also possible to define your own conversion attribution model, so that it suits your company. Attribution Example 12.2 describes how Google uses the various attribution models. models EXAMPLE 12.2 Google’s attribution models A customer finds your site by clicking on one of your Google Ads ads. She comes back to your site a week later via a social network. That same day she comes back a third time via one of your email campaigns. A few hours later she comes back again, this time directly, and she makes a purchase. According to the attribution model ‘Last interaction’, the last point of contact, in this case the ‘Direct’ channel, would get 100% of the value points for the sale. Using the attribution model ‘Last non-direct click’, the direct traffic is completely disregarded and 100% of the value points for the sale go to the last channel the customer clicked through from prior to the conversion, in this case the channel ‘Email’. According to the attribution model ‘Last Ads click’, the last Ads click (in this case the first and only click to the ‘Paid search results’ channel) would get 100% of the value points for the sale. In the attribution model ‘First interaction’, the first point of contact (in this case the channel ‘Paid search results’) would receive 100% of the value points for the sale. 12 According to the attribution model ‘Linear’, every point of contact in the conversion path (in this case the channels ‘Paid Search Results’, ‘Social Network’, ‘Email’ and ‘Direct’) would receive the same number of value points (25% each) for the sale. According to the attribution model ‘Time Decay’, the contact points that are closest to the time of the sale or conversion, receives the most value points. In this specific sale, the channels ‘Direct’ and ‘Email’ would receive the most value points, because the customer interacted with these two channels a few hours before the conversion. The ‘Social Network’ channel 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 546 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 547 would receive fewer value points than the ‘Direct’ or ‘Email’ channel. Since the interaction with ‘Paid search results’ took place a week earlier, this channel receives considerably fewer value points. According to the attribution model ‘Position based’, 40% of the value points are attributed to both the first and the last interaction, and the remaining 20% are evenly distributed over the intermediate interactions. In this example, the ‘Paid search results’ and ‘Direct’ channels would each receive 40% of the value points, while the ‘Social Network’ and ‘Email’ channels would each receive 10% of the value points. Source: https://support.google.com/analytics/answer/1662518 A nuanced view is required It is important to know that there is no uniform truth! You will have to experiment to find out which model best suits your customers’ customer journeys. If the customer requires only an average of two contact moments, then a linear model is probably sufficient. However, if a customer takes an average of 60 days to make a purchase and there are many more touchpoints, then a time decay model is probably better. People also tend to use different devices, meaning they often do not fit into the model. There are also consumers who discard their cookies in the meantime. Many corporate networks even do this daily. Segmentation and personalisation When a visitor comes to the website from one of the previously described traffic sources, it is important that the website delivers against the visitor’s requirements. A common technique that is used for this is personalisation or segmentation. As you were able to read in the previous chapters, personalisation is a trend that will continue to develop in the coming years and continue to increase in popularity. Nowadays it is at least technically possible to provide each of your website visitors with a personalised experience. This allows the marketer to communicate the right message, to the right person, at the right time. This, of course, is the objective of all advertising, offline or online and, provided it is done well, it can be highly beneficial to the visitor/customer as well as to the brand owner. Most organisations begin using segmentation before actually deciding to personalise and show each individual a unique message. Common segments whose behaviour can be analysed using Digital Analytics are: 12 Customers/non-customers Male/Female Age Behavioural segmentation: – Which products have they previously viewed/bought? – How often do they visit the website? – What behaviour do they exhibit on the website? 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 547 26/11/20 5:01 PM  548 © Noordhoff Uitgevers bv If a marketer chooses to use personalisation, they may ask: is the website visitor a customer? To find out, it is important that this fact is recorded as data within Digital Analytics. Is this customer male or female? And what is his or her age? This is all interesting and potentially relevant data to use for personalisation. In addition to the availability of this data with respect to these segments, it is essential that they are linked to an action. By doing so, Digital Analytics brings our website traffic numbers to life in the shape of real users doing specific things, ie. an audience to whom we can make personalised offers. Eg. ‘If visitor X is a customer and female, show them this product or this offer.’ It is important to take privacy law and ethics into account and only record the data that the user has agreed to share. What do I know about the customer based on previous orders? Suppose I ordered a new phone with a subscription from a telecoms provider last week. If I return to the website a week later, what are the chances that I’ll want to order a new phone again? Or am I more likely to have a service- related question? Or, perhaps an extra product that I can add to my subscription? What do I know about the customer or visitor based on their behaviour during the visit? Which products or pages have they viewed in the past few minutes and what message is now the most suitable for them? Remember that personalisation costs. It is important that companies continue to measure what the added value of personalisation is. An A/B test (see box in 12.3.3) is one method used to validate this. § 12.4 The ABC Model: the Digital Marketing funnel In the previous sections, we introduced the concept of digital analytics and discussed the most common metrics in relation to the Digital Marketing funnel. In this section you will learn about a different model that is often ABC model used in digital analytics: the ABC model (see Figure 12.4). In this section, we will now take an in depth look at the use and application of the ABC model. 12.4.1 Use of the ABC model Using the ABC model, look at the marketing funnel from the bottom to the top. Start at the bottom of the funnel, at the conversion (C) and then move up to the behaviour (B) and the channels and segments (A) (Jansen, 12 Klaassen & Visser, 2015). Analyses for a website only become meaningful when you make a distinction between segments. So, ask yourself which important visitor segments can be distinguished: prospects versus customers desktop versus mobile visitors one-time buyers versus loyal customers browsing versus focused visitors visitors from a comparison site versus from a newsletter visitors coming in via different landing pages 12_275468_DIGITAL_MARKETING_FUNDAMENTALS_CH12.indd 548 26/11/20 5:01 PM  © Noordhoff Uitgevers bv DIGITAL ANALYTICS 549 segmentation based on geography, day of the week, time of day etc. origin; channel: direct traffic, referred traffic, other FIGURE 12.4 The ABC model Company: Value proposition, activities, etc. Segments Acquisition A unique visitors, number of sessions, % new visits Behaviour B Bounce %, exit %, pages per visit, time on page/site Conversion number of (soft) conversions, yield, C conversion % Then make an analysis for each of these important segments, keeping your knowledge about the company in mind. Only then can you acquire real insights. Level C in relation to B You look at which visitors successfully convert (C) and then see if you can find out where they differ in beh