Marketing is the business activity of communicating the value of a product such that a consumer recognizes that value and performs a value-generating transaction for the business. Advertising is the vehicle by which marketing communicates that value. Marketing and advertising analytics, thus, involves conducting analysis to support the marketing function and make recommendations that improve advertising performance. The information and recommendations are used by managers to make decisions about what to do next to run the marketing function, more efficiently allocate spend, or make decisions about marketing investment, both tactically and strategically. The ecommerce analyst’s job may include analyzing the impact of marketing spend overall on revenue, brand value or equity, specific types of campaign channels and their performance, and even down to the impact of individual ad units. Customer analytics, which is discussed in Chapter 9, “Analyzing Ecommerce Customers,” applies to marketing and advertising analytics.
To do marketing analysis effectively, the ecommerce analyst needs to understand the demographics of customers, the social forces guiding consumer behavior, the economic forces at play within an industry, the competitive landscape, and the technology that supports marketing execution and, more increasingly, marketing automation. The analyst must use technique to segment behavior, audiences, customers, and advertising channels and units into similar groups and categories.
By conducting and compiling new analysis within a company, the analyst is said to be conducting primary analysis. When an analysis uses existing data and repurposes to prove or disprove a hypothesis, the analyst is conducting secondary analysis. For example, if an analyst is attempting to understand whether a particular shopping cart flow is effective for conversion, the analyst is doing primary analysis. If the analyst is examining research conducted by others to make recommendations influencing the redesign of a shopping cart, she is doing secondary analysis.
The type of analysis companies will demand of their ecommerce analysts will vary over time, with secondary analytics helping to inform early-stage product development and design, and primary analytics guiding the evolution of the ecommerce experience after a site is launched. It’s feasible and possible for an ecommerce company to do primary research to prove out an idea before launching a site—and subsequently prove out the hypothesis learned with more primary research and/or secondary research. As an ecommerce company grows and evolves, both primary and secondary analysis will function together to guide development and understand the effectiveness of operations.
Marketing and advertising analytics can be further categorized beyond primary and secondary analysis into quantitative analytics and qualitative analytics, which may be combined in an analytical deliverable. For example, an advertising analysis may indicate how much was generated in sales (quantitative) and also describe the attributes of the customer who bought via the advertisement (qualitative). As you’ve probably figured out, quantitative analysis uses numerical methods in which the data can be summed, analyzed, and modeled using mathematics and statistics (data science!), whereas qualitative analysis makes use of non-numerical methods and techniques derived from traditional research methods where text, verbatims, and categorical data are perhaps more common than numerical data. The analyst is the person who will make the decision about what analytical approach to use for research. The decision on which approach to use (or both) can be based on whether the data can be quantified via data collection or acquisition, or whether unquantifiable or abstracted ideas are needed for the analysis.
Ecommerce marketing analytics includes activities such as campaign analysis, search engine optimization (SEO), search engine marketing (SEM), e-mail analytics, social media analytics, affiliate and reseller analytics, and marketing mix modeling. Marketing analytics requires interpreting data about the brand and messages that impact the brand. Advertising for an ecommerce site requires assessing the impact of advertising on sales and requires knowledge of the advertising lifecycle and the types of advertising available. Advertising analytics can include audience and demographic analysis, unit analysis, and reach and frequency analysis. It can also include analysis of the customer’s journey across advertising channels from a temporal perspective as well as from the perspective of the advertising’s impact on the other advertisements seen over that time (i.e., attribution). In this chapter we will explore these concepts, describe these types of analyses, and explain how to take the next steps for executing them at your company. Please note that I covered “competitive intelligence” in my book Building a Digital Analytics Organization. Additionally, ecommerce companies don’t typically have a sales force, so this book doesn’t cover the analysis of the sales team. In some companies, sales analytics may be part of or separate from marketing analytics.
The goal of most analysis is to uncover meaningful information that can be used as knowledge to guide business decisions that reduce cost or increase revenue. Marketing and advertising analysis in ecommerce companies is no different. The highest ideal goal for marketing and advertising analytics is to help a company drive financial performance based on the data collected. Because marketing can be a large cost center, analysis that can help improve how marketing spend is allocated and how a marketing campaign performs can help a company’s bottom line.
Marketing analytics, of course, is the larger concept in which advertising analysis occurs. Marketing buys advertising. Advertising is a marketing tactic. Thus, we can consider that the same goals for marketing analytics can, in many cases, apply to advertising analytics. It could be argued that in this book I don’t need to make the distinction between marketing analytics and advertising analytics, but I think it’s helpful to make a distinction such that, although the two disciplines are closely related, there’s enough investment and capital being allocated both in general ecommerce marketing and in ecommerce advertising that it makes sense to discuss them individually. Marketing and advertising analytics can be categorized into the following types of analysis:
• Exploratory analysis investigates an assumption about a particular topic. For example, you may come up with a hypothesis about a particular marketing channel, such as social media, based on particular beliefs shared by employees at your company. The hypothesis might be that the profile of the social media audience skews toward conversation being dominated by young females and that males are marginalized in conversations about the brand. As a result, the analytics team may choose to index the text of comments in social media, perform sentiment analysis, map names to gender, and then determine whether the assumption is true.
• Descriptive analysis relates to the act of making the subject of an analysis clear, lucid, and understandable. As the phrase suggests, descriptive analytics focuses on answering the question “What is this?” This type of analysis identifies what is happening and what has happened so that it can be comprehended and communicated. You might notice that every Friday your ecommerce sales increase to their highest point all week. A descriptive analysis would seek to describe how much revenue increased, in what products or categories, what events occurred around the revenue increase, what segments have increased revenue, and so on.
• Predictive analysis is a popular topic in 2016. It references analytics that are executed to predict what may happen in the future. Given certain inputs, predictive analytics will generate an output that represents a prediction of what might happen. Methods for predictive analytics can include regression analysis, survival analysis, various types of simulations, and more. For example, you may want to predict what might happen if the marketing spend for a particular channel is increased at the same time a new line of products is introduced. Predictive analysis, given the appropriate data, can be used to find out the possible answer to what might happen.
• Conclusive analytics attempts to derive a conclusion based on the available data and the relationships between the variables in the data. For example, an analyst might notice that certain products sell when there is a local marketing event, but the data collection doesn’t support directly linking product sales to the event; yet the analyst has seen the causality enough to conclude that there must be a correlation between the event and sales. The analyst would then build a model to explore the relationships between the variables related to sales and the variables related to the event to see whether a mathematically valid conclusion about the relationships can be derived.
• Prescriptive analytics involves determining the right course of action to take given a number of variables and potential outcomes. The idea of determining the right action to take is referred to as prescriptive analytics and is an evolution of predictive analytics. It is one thing to predict what could happen, but it is more powerful to tell people what action to take and the impact a particular set of predictions and actions would have on the rest of the business. As such, prescriptive analytics suggests the best possible beneficial action to take after knowing the predictions, and it identifies the implications of each prediction so that the stakeholder is guided. For example, a predictive model in ecommerce may predict the certain set of products to show to a customer. A prescriptive model will identify the best set of those products to maximize gross profit.
When marketing analytics and advertising analytics are being performed, both qualitative and quantitative analytical methods will be applied. Qualitative analytics are research-based methods that involve collected data from people via focus groups, interviews, and approaches that collect human responses that are not expressed using numbers. Qualitative methods for analytics include these:
• Focus groups in which people are asked directly about their perceptions, opinions, beliefs, and attitudes related to an ecommerce brand or product.
• Interviews in which questions are posed by an interviewer directly to people in the form of conversation. Responses are captured and analyzed.
• Projective approaches in which a person responds to stimuli to help the ecommerce site understand how people are thinking about what has been presented. For example, new media may be shown, a home page design may be addressed, promotional offers may be reviewed, or creative text may be read. The goal is to uncover opinions and emotions that otherwise are difficult or impossible to measure online.
• Panels consisting of a group of individuals whose behaviors, thoughts, beliefs, and actions are sampled and inflated to project to an entire audience or population.
• Surveys with specific questions related to an area of interest sent to audiences to collect primary data based on the responses; or research about prospects and customers purchased as secondary data from research companies.
Analyzing marketing and advertising requires knowledge of the marketing lifecycle. The marketing lifecycle can be expressed as a set of linear phases around which marketing programs are executed. The ecommerce analyst must strive to analyze the data collected during each of the following phases:
Phase 1: Activation. Activation is measured qualitatively via surveys and voice of the customer (VoC) data. Activation in analytics refers to the “awakening” of need in the customer. Activation can be acute (occurring suddenly) or realized (a result of long-term influences). Marketing activities across TV, print, or radio can influence ecommerce. Think of William Shatner or the trivago guy.
Phase 2: Exposure. In a state of activation, the potential customer (the lead) sees and perceives the brand and its associated physical and psychological properties and qualities through paid, owned, or earned media. Marketing activities such as online advertising can generate exposure for ecommerce companies.
Phase 3: Awareness. In an activated and exposed state, the lead who is exposed to the brand becomes cognitively aware of the exposure. Similar to when your mother told you, “You might hear me, but you aren’t listening,” you can understand how awareness results from exposure, but not all exposure creates awareness. Another case in point is the online advertising industries emphasis on a “viewable impression,” which suggests a similar relationship between exposure and awareness. Search engine marketing (SEM) can generate awareness.
Phase 4: Differentiation. This is the process of evaluating a brand and its product or service qualities against competitors and substitutes. A lead compares attributes against each other to determine how to work through the infoglut of advertising. During this phase, the narrative of the advertising and exposure is accepted, viewed as aberrant, or resisted by the lead. Content marketing can help differentiate an ecommerce brand.
Phase 5: Consideration. A lead considers the brand against competitors and substitutes based on a judgment of the fit of the brand’s perceived qualities against the customer’s perceived needs. Consideration is where the many are slimmed to the few—and where it is most likely a customer will seek and be exposed to brand messages from multiple channels. Search engine optimization can drive consideration.
Phase 6: Acquisition. This phase involves accounting for the many paid, owned, and earned media across which your leads may have been exposed. For an ecommerce site these channels may have been used for earlier phases in the lifecycle. What drives acquisition is the fundamental question that attribution analysis tries to solve.
Phase 7: Conversion. Having transitioned from acquisition source and attributable marketing channel, the customer buys online, and a conversion occurs. The customer searches for products, adds them to a shopping cart, buys them, and generates a profitable conversion. Factors that influence conversion include the overall ecommerce experience and specific flows like the checkout process.
Phase 8: Retention. Although customer conversion, and thus the transition from “engagement time” to “revenue dollars,” is an impressive achievement, it is even more impressive and profitable when the customer comes back to buy again. The business activities of nurturing and having a “customer relationship” are crucial to measure. Incentives, promotions, and discounts sent over e-mail and promotions offered via social media can drive retention.
Phase 9: Loyalty. A customer can be considered loyal after a second purchase. Loyalty analysis involves understanding why and how the customer purchased again from your company. Even more critical is how—post purchase—the customer again became activated and moved through phases 1 to 5. In both Retention and Loyalty, customer churn and lifetime value analysis are important. Customer reward programs, periodic events (like seasonal sales), and newsletters can help create loyalty.
If you are an advocate of the linear funnel, such a theory for linking advertising to conversion in ecommerce might read something like this:
1. Creates awareness through differentiation.
2. Positions the brand and buying experience such that it evokes favorability.
3. Reaches enough people so the advertising strengthens the brand and supports or maintains brand equity.
4. Informs and compels a person’s purchasing behavior via a certain frequency of exposures.
5. Leads to a person visiting an ecommerce site or mobile app via an advertisement directly or indirectly through brand exposure.
6. Creates a frictionless buying experience that enables a person to easily find the product for which they are looking and buy it.
7. Generates or sustains loyalty and reactivation during the next cycle of realization of the intent to purchase by using targeted, customer-specific marketing and advertising.
The analyst’s job is to ensure that business questions can be answered about the activities within the marketing lifecycle. To do so, the analyst must collect data within and across all of these phases, coordinating with the marketing team, IT, and supporting teams and ensuring data collection and availability for future analytical projects. Then the analysts must, of course, analyze the data.
Ecommerce marketing tactics are used in the marketing lifecycle to move people through each phase. The ecommerce analyst will measure the performance and outcomes of the advertising and campaigns of different marketing types such as these:
• Activation marketing can include programs that try to activate the brand in the mind of consumers by putting advertisements wherever they can be bought. But as Angela D. Nalica, professor of statistics at the University of the Philippines, says, “Marketing activation usually entails a universal blast of information to all consumers. Often, only a small proportion of the consumers react positively to such activation, resulting in waste in marketing expenses. If a circle of influencers can be identified for certain events or phenomena, then such activities can be focused into a group of factors or individuals, thus optimizing the outcomes.”
• Exposure and awareness marketing can include public relations releases and communications; advertising on television, on the radio, and in print; online advertising; e-mail; and paid and organic search. The goal of awareness marketing is to drive home the message of activation marketing (if done). Its primary purpose is to get the brand or product recognized in the minds of potential customers.
• Consideration marketing appeals to consumers who are nearing readiness to buy. They are aware of the brands and products and are actively considering alternatives and options. Content, social reviews, events, and promotional offers sent by ecommerce companies can help to promote consideration. Social ads, social reviews, blogs, user-generated content (sponsored or unsponsored), and media such as webinars, videos, and other multimedia content are used in consideration marketing.
• Acquisition marketing focuses on driving traffic to an ecommerce site. Depending on the goal of the program, the acquisition may be broad to appeal to all possible consumers who are in the target market, or narrow to focus only on consumers who have certain attributes or are within a particular phase of the customer journey. Targeting and retargeting, online media, e-mails, promotions, direct mail, and focused advertising are used to bring prospective customers to the site.
• Conversion marketing drives addressable audiences toward the completion of a conversion activity, such as the purchase of a product. Conversion optimization, discussed in Chapter 8, “Optimizing for Ecommerce Conversion and User Experience,” is a type of conversion marketing. Conversion marketing also includes paid search, targeted display ads with direct offers, onsite merchandising, and promotions.
• Retention and loyalty marketing focuses on the critical need of companies to sell more to customers who have purchased in the past. Promotional programs, loyalty programs, and targeted marketing across different channels can help to promote customer retention and loyalty. Community forums, user groups, social networks, blogs, newsletters, special promotions, targeted incentives, and offers to repurchase are all part of retention and loyalty marketing.
Marketing uses tactics and techniques for creating a brand, driving brand awareness, targeting customers, and then compelling and guiding them through a customer journey that results in a purchase. The phases of the marketing lifecycle can be thought of as supporting a potential customer as he moves through a journey from not having bought to buying. To analyze marketing and advertising, it’s important, as always, to consider the business goal of the analysis, the type of the analysis you are going to do, the customer journey, and the types of marketing within phases of the journey. By considering all of this information, you can create and execute a plan for ecommerce marketing analytics, such as these fundamental types of analysis:
• Campaign analysis uses the campaign code as the primary dimension for understanding performance. Campaign codes are human-readable or encoded parameters in a query string. These parameters are name/value pairs that are meaningful to people, and usually understood by machines. Campaign codes typically specify the campaign name, the type of marketing campaign (i.e., paid search, display, e-mail, and so on), and other information, such as the placement, the version, the variation of the campaign, or the creative in it. It is a good idea to standardize on a specific set of naming conventions and valid names and values for campaign codes. These conventions should also fit the requirements necessary for automatically processing campaign data using campaign codes.
• Search engine marketing (SEM) traditionally refers to the business and technical processes and work required for bidding on paid search keywords and managing them to maximize performance. The work can include keyword research in which analysis is done into the words and phrases most likely to be used by people to search for the ecommerce site or its products. Ads are created, including the creative text that appears in the ads and is displayed in the search engine. The content in the ad is controlled. Most important, search engine marketing allows for the ranking of the ads to be specified (up to the cap or the available budget). SEM enables detailed and specific targeting of the keyword to specific devices, browsers, geographies, time periods, and more.
• Search engine optimization (SEO) describes the process for ensuring that the content on an ecommerce site is displayed in organic search results in the desired ranking for a specific keyword. Like SEM, SEO involves keyword research to identify important and relevant phrases for the ecommerce site and its products. But it also involves onsite technical work to determine whether the site’s content is indexable by search engines and contains the specific keywords and relevant content to rank in search results. SEO also requires backlinking from other sites to the ecommerce site and an assessment of how sufficiently search engines are crawling and indexing site content. Other tools for SEO assist an analyst with understanding the factors that influence search rankings, the keywords used to find the site, the performance of organic search keywords toward KPIs, and other helpful information.
• Affiliate analytics is a way of analyzing the effectiveness of affiliates and affiliate networks in teams of conversion, revenue, and profitability. Affiliates are, of course, sites that send traffic or, optimally, customers who engage in transactions of buying products. In exchange for the revenue-generating customer or the traffic, the ecommerce site will pay for performance at an agreed-on rate. Analysts must set up the right tracking, measurement, and reporting to be able to analyze the performance of affiliates in order to assist with decision making. Affiliate tracking involves the customer and the ecommerce site, but also the affiliate (publisher) and the affiliate network in which they are a part. When tracking affiliates, you must use campaign codes specific to each campaign per affiliate, per affiliate network, and per offer—and you must be able to track these dimensions through behavior that leads to conversion.
• Social media analytics is about listening, engaging, and participating in social media through conversation, commentary, marketing, advertising, and other forms of multimedia-based social engagement (video, audio, and rich experiences). The goal of social media can be to build brand awareness, strengthen and inform customer relationships, influence prospective customers during the customer journey, create word-of-mouth virality, and drive consumer engagement and direct response. Text analysis, such as sentiment analysis, concept extraction, and classification, and text mining are useful when analyzing text-based social media data.
• E-mail analytics measures the effectiveness of e-mail as a marketing channel for causing purchases directly as a result of the e-mail or indirectly from latent visits and buying after exposure to e-mail. E-mail analysis includes understanding the customer data in order to create relevant segmented e-mail lists based on known behaviors and/or past purchasing history or other attributes. Analysis of the actual e-mail distribution includes deliverability metrics, opt-outs, bounces, open rates, click-throughs, and e-mail–specific attribution. Links, buttons, images, and other hot spots in e-mail creative can be campaign-coded and linked to specific landing pages, such as individual product pages, and their impact on conversion can be measured and reported.
• Marketing-mix modeling is the term used to describe the application of data science and statistical analysis to sales and marketing data in order to understand the impact of marketing on sales, and then to forecast, estimate, or predict future sales based on maintaining or modifying the mix of marketing. A marketing-mix model estimates sales in two ways. First, base sales are identified, which are the sales that result from natural demand. Second, incremental sales are estimated based on demand generated from marketing. Factors in a media-mix model include the price, promotions, the distribution strategy, competitor impact, and the different types and channels of marketing and advertising. These factors are all analyzed to suggest an optimal marketing and advertising mix and tactics that achieve the desired revenue or profit goals. Media-mix modeling can help determine the return on investment, as well as the contribution of, and effectiveness of, marketing tactics. This information can be used to better allocate marketing budgets, spend marketing investments more wisely, and guide promotional offers and advertising strategies.
• Attribution involves understanding the impact of marketing channels on revenue, engagement, or another quantitative indicator in rank order of their impact. See Chapter 11, “Attribution in Ecommerce Analytics,” for more information.
• Audience analysis is about understanding the attributes of the audience from available data. This data can include age, gender, demographic, psychographic, propensity, mind-set, neurometric, psychological, transactional, behavioral, and other data. This information is used to create narratives describing and identifying the customer. Whether via personas or “buyer legends” or simply segments, clusters, and cohorts, marketing requires the ability to understand the audience across these dimensions in order to maximize lifetime value or reduce the cost of customer acquisition.
• Online display ad unit analysis informs the creative conceptualizing, building, trafficking, and placing of ad units on external sites by an ad server. This type of analysis involves describing and comparing how well ad units contribute toward the desired goals. Some ad units may be for branding; others, for direct response. The impact of exposure of these ads to awareness, favorability, consideration, and acquisition can be studied. In addition, fraud can be detected, investigated, and remediated.
• Customer journey analysis refers to an analysis that tracks prospects and customers as they move through different phases of the customer journey specific to the ecommerce site. Specific metrics can be measured, reported, and analyzed to inform about the count and behavior of prospects and customers with marketing and advertisements targeted to each phase in the customer journey.
One of the simplest marketing analyses is also the most complex and hardest to realize. The Customer Origin report shows the source of customers acquired from the different marketing channels; for example, direct, referral, paid search, organic search, social, and the various types of online advertising. It’s a simple report because many tools will deliver it as a capability “out of the box.” It is complex because it requires the ecommerce company to define how the traffic should be allocated to the various marketing channels and to practice and adhere to those definitions when adding campaign codes and metadata available for reporting. It also requires a specific attribution model to be used. For each traffic source or marketing channel, consider collecting data such as the number of visits by channel (including both prospects and customers), conversion rate, revenue, the number of orders, the number of customers (i.e., visitors who bought), repeat customers, repeat customer rate, average order value, and revenue per customer. From these core metrics and KPIs, you can begin to understand what marketing channels are contributing to performance, conversion, and revenue—and then direct your subsequent marketing and advertising analysis activities toward exploring, describing, concluding, predicting, and prescribing what you have uncovered.