Digital marketing is a term that has emerged to describe the usage of the Internet and digital media to support marketing. By using digital technologies, marketing teams achieve marketing objectives defined in the organization. As more consumers, customers and prospects are present and active on different online platforms, digital marketing needs to encompass different kinds of media channels. Paid media, earned media and owned media needs to be considered in the competitive and complex buying environment of today’s market.
Paid media are channels where investment is put forward to pay for visitors through ads or search ranking. Earned media describes the use of editorial content published by advocates of the brand on social media, blogs or publishers. Owned media is a channel owned by the brand itself and consists of social media profiles, websites and mobile apps. Digital marketing on Internet channels also enables a different kind of interaction with customers. While traditional media are push media where the marketing material is broadcasted from the company to a large audience, the Internet enables the customer or prospect to initiate contact with the brand. Inbound marketing describes a market where the customer actively seeks information about a particular brand or item that requires the company to be present with digital marketing material to converge the prospect in their quest for information. Interactive marketing also enables individualized marketing based on intelligence collected and prepared from the large amounts of available data online. Marketing teams need to embrace an analytical and data driven approach to marketing to be able to reach its customers and prospects with a tailored made and individual marketing message.
The marketing budget of a company is still rising, and in 2016, companies are using 12% of their revenue on marketing endeavors. 57% of the respondents in the Gartner CMO Spend Survey for 2016 to 2017 expects an increase of expenditure on marketing initiatives. In addition, the CMO is on track to spend more on marketing technology than the CIO. Labor is where the largest chunk of the marketing budget is spent, but in second place is technology. Marketing uses 3.24% of overall revenue on technology in comparison to the CIO who allocates 3.6% of revenue on technology. In addition, marketing is spending most of their budget on the web, digital advertising, and digital commerce, proof that digital marketing is a key method for an organization to reach its customers and prospects.
In this post, the focus will be on how digital marketing needs to adopt an analytical mindset to be able to utilize digital marketing and the vast amounts of available data on the Internet that can aid in the quest to reach and retain customers. An overview of key topics is provided followed by a discussion of key findings and a conclusion.
Digital marketing, consumers, data and algorithms
This post will elaborate on digital marketing, consumer behavior, big data and machine learning. Through the investigation of these topics, this post will elaborate on the importance of a data driven and analytical approach to digital marketing.
Digital Marketing Digital tools are in daily use by a growing amount of potential customers and existing customers of a brand. Digital marketing is a key tool for a brand to reach its customer base and prospects. By utilizing digital technologies, digital marketing is used to reach customers on the platforms where they are active. Through their 2016 survey of CMOs, Gartner found that companies used 12% of revenue on marketing initiatives. CMOs are also on the path to be spending more on technology than the CIO, with 3.24% of revenue being spent by CMOs on technology against 3.6% of revenue spent by the CIO. This calls for the creation of the chief marketing technologist, or CMT, to bridge the gap between marketing and technology to ensure a good fit of technology and marketing initiatives. 67% of marketing departments are saying that they will spend more on technology-related activities between 2014 and 2017 than they have before.
Consumer behavior Successful companies try to take a position in the market and offer services or products that serve a target set of customers. The company must stay unique and personal to obtain the competitive advantage in its market segment. The assumption to this premise is that consumers are making deliberate and rational decisions based on their own needs and desire. A good company strategy must then figure out and respond to the conscious logic that underlies the consumer’s decision to purchase specific goods or services. However, the human brain is not an analytical machine making decisions based on objective data and thought out reasoning. The noisy data that surrounds the consumer is not thoroughly analyzed, but rather the missing information is filled in by the individual based on experience and intuition.
Big Data Big data is described using the three V’s: large amounts of data (Volume), data that is being generated at high speed (Velocity) or data that comes in different formats (Variety). The data collected is also often unstructured and contains a lot of noise, but patterns and knowledge can be found using algorithms and analytics. Research has also shown that data driven organizations are 6% more profitable than competitors while being 5% more productive.
The introduction of objects and products that are able to sense their environments and report its status to a cloud, massive amounts of data will be generated. The data can be utilized by consumers and companies to improve value creation and daily life. Big data enables companies to measure and obtain deeper knowledge about its operation, leading to improved decision making. Large amounts of data about customers are available through different sources and can easily be consolidated, creating a complete profile of the customers buying habits and transactions. A team in a bank combined transactions from ATM’s, queries made online, customer complaints and other data to create a 360-degree view of their customers to be better equipped to provide customer service.
Machine Learning and algorithms Algorithms consist of predefined steps to be taken in order to solve a given problem. Machine learning, on the other hand, tries to come up with rules based on data, statistical methods, and computing power. Instead of telling an algorithm what to do and how to do it, machine learning aids the computer in making sense of data and deriving rules from patterns and correlation. Based on observations, data is collected and fed to a machine learning algorithm which then creates a model that can be used to process data. The ultimate goal of machine learning is to generalize findings based on a subset of data. Machine learning is utilized in a growing number of applications ranging from identifying human faces in an image, clustering customers and creating autonomous cars.
Machine learning is a good tool in the age of big data as it is good at analyzing datasets consisting of a huge amount of rows and columns. Machine learning can generate personalized recommendations that are helpful without giving any information as to why for example users like what they like and how it is possible to change what they like. The aim when utilizing machine learning is to find consistent relationships in datasets that can provide value, and we want to filter out correlations that happen at random.
Big data and advanced analytical models that use demographic, transnational and buying information of customers are perfect tools to create campaigns that are created to give short-term earnings to a company. However, short-term revenue created by sales promotional campaigns will threaten investments that are building the brand in the long run. Chief marketing officer at Subway, Tony Pace, points out that if promotions that are outputs of analytical models are biased against maximizing sales in the near future, the customer will always get promotions to products that they have purchased the most. This can trigger a short-term increase in revenue as the promotions have a high probability of leading to a purchase, but be destructive in the long run. There exists a positive correlation between the number of products from a brands catalog that has been purchased more than once and the customer’s loyalty to the brand. The short-term promotion that only encourages repeated customer behavior, will only be a short time success for the company and its brand. Big data and machine learning is a great supplement for marketers, but trusting the results alone can lead to promotions that contradict the brand and its values. Experienced marketers need to ensure that the brand and the values of the company are safeguarded and not disrupted by data-driven models.
Dellarocas et al. (2007) used publicly available data to build a model to better predict sales and revenue of movies. From historical data and knowledge, the impact word-of-mouth had on how well a movie did in its gross total, Dellarocas et al. (2007) proposed a model where online user reviews were used as input to their model. The best model created was able to predict revenue trajectory on a weekly basis with a mean absolute percentage error of 10%. The volume of online reviews can also be used as an indication for early sales at the box office, and demographics of the online reviews can be used to forecast demand in different segments. As expected, Dellarocas et al. (2007) also found a correlation between the early volume of online reviews and the box office revenues in the opening weeks. The strength of this model is that online movie reviews from movie goers are available online within hours of a movie premiere, thus enabling the marketing team of the movie to act quickly based on freely available data.
For a company to be able to implement analytical endeavors, three important capabilities need to be present in the organization:
(1) Be able to manage and consolidate multiple data sources
(2) Build models for prediction and optimization
(3) Management must champion the initiative throughout the organization so that the models are contributing to better decisions.
It is also important not to build models starting with the data, but rather start with a business opportunity and use data and analytics to improve performance. Consolidating huge amounts of data allows companies to perform statistical and machine learning analysis to identify patterns and knowledge buried in the data. However, if the managers are unable to use this insight to enhance the performance of the business, big data and analytics will provide little benefit. The business needs to approach big data with the mindset of a data scientist, and not only by data mining in an endless search for story in the data.
When creating analytical predictive models, these models can often get complex with many variables used to explain data. The accuracy might be high, but managing and maintaining a vast array of variables is hard over time. Companies should therefore aim at building models that minimize the complexity and at the same time maximize performance and accuracy. Models that are built also needs to be transparent and understandable to the users. A retailer who built a model to optimize return on their investments on advertising failed to utilize the model because key marketers who made decisions did not understand or trust the models being built.
Big data and analytical data driven endeavors will also make it easier to measure marketing campaigns across multiple marketing channels. Historically, different marketing channels such as TV, print ads, and digital marketing has been analyzed in isolation. Big data initiatives enable the consolidation of multiple data sources making it easier to create a 360-degree view of all marketing initiatives across channels. Cross channel analysis will also illuminate the interaction between channels where for example a TV-advertisement initiates a web-search that again leads to the prospect clicking on an ad and purchasing a product.
Data driven insights and analysis can also enable a company to keep the existing marketing budget and still get up to 10% to 30% higher performance on marketing initiatives according to Nichols (2013). However, to successfully model their business the company needs to gather data about market conditions, competitive activities, marketing actions, the response from consumers and business outcomes. Car manufacturer Ford used predictive analytics to uncover that the marketing budget allocated too many funds on the digital display that lead to less investment in search related marketing initiatives. Big data and analytics enabled Ford to adjust their marketing spending to utilize different marketing channels to the fullest. Ford also uncovered that national and local marketing budgets rarely coincide, and thus needed better coordination to encompass the national and local strategy. The shift from national focus when allocating marketing funds to a local focus has given tens of million dollars in revenue without spending more money than before on marketing.
The ability to allocate marketing funds to the correct channels is crucial for the success of marketing teams. Big data and advanced analytics enable this through real time analysis based on massive amounts of data that can be acted upon on a daily, weekly, monthly or even down to seconds in the digital online space.
A large part of marketing today is based on brands that deliver advertisements to customers in a quest to influence their buying choices. Advertisement money is spent on reminding the customer to pick up a specific product next time they go grocery shopping. At the horizon, bots and smart objects are emerging that will take over the actual procurement of products for the consumer. Smart fridges will let the customer know when he needs to restock, and an order to the cheapest grocery store that is closest to the customer will be sent. When choices regarding shopping are left to bots and algorithms, marketing that targets the consumer to influence his or her buying pattern will be in vain. In the near future, a bank might analyze a customer’s bank account and determine that they could save money by changing their energy provider. If the customer then accepts, a bot will talk to the energy company’s bot to negotiate a better deal on behalf of the customer. The role of marketing is changing and the marketplace will be fast paced in the future. Building brands and creating promotional campaigns might prove to be outdated when it is set into action. The use of data driven initiatives that incorporate advanced analytics and machine learning will be vital for a company to succeed at marketing.
A shift from influencing customers to influencing algorithms might be the future of digital marketing. If bots and algorithms take control of procurement, a brand needs to be the default choice in the algorithm. Loyalty against brands will also be harder to distinguish. Is it algorithmic loyalty or consumer loyalty that is the result of a purchase? Also, as customers are subject to bias and habits, marketing initiatives might have to think differently when purchasing choices are performed by algorithms and bots that do not have the same preconceived view of the world as human consumers have.
Marketing is rapidly becoming a battle based on data and knowledge that can only be won through careful implementation of data driven and analytical endeavors.
Barton, D. and D. Court (2012, oct). Making advanced analytics work for you. Harvard Business Review.
Brinker, S. and L. McLellan (2014, jul-aug). The rise of the chief marketing technologist. Harvard Business Review.
Chaffey, D. and F. Ellis-Chadwick (2012). Digital Markering: Strategy, Implementation and Practice (fifth ed.). Pearson Education Limited, England.
Davenport, T. H. and D. Patil (2012, oct). Data scientist: The sexiest job of the 21st century. Harvard Business Review.
Dawar, N. (2016, oct). How marketing changes when shopping is automated. Harvard Business Review.
Dellarocas, C., X. M. Zhang, and N. F. Awad (2007). Exploring the value of online product reviews in forecasting sales: The case of motion pictures. Journal of Interactive Marketing 21(4), 23–45.
Domingos, P. (1999, dec). The role of occam’s razor in knowledge discovery. Data mining and knowledge discovery 3(4), 409–425.
Domingos, P. (2012, oct). A few useful things to know about machine learning. Communications of the ACM 55(10), 78–87.
Horst, P. and R. Duboff (2015, nov). Don’t let big data bury your brand. Harvard Business Review, 3–9.
Lafley, A. and R. L. Martin (2017, jan-feb). Customer loyalty is overrated. Harvard Business Review, 3–10.
Lantz, B. (2013, oct). Machine Learning in R: Learn how to use R to apply powerful machine learning methods and gain an insight into real-world problems. Packt Publishing Ltd, Birmingham’.
McAfee, A. and E. Brynjolfsson (2012, oct). Big data: The management revolution. Harvard Business Review.
Ng, A. (2016, nov). What artificial intelligence can and cant do right now. Harvard Business Review.
Nichols, W. (2013, mar). Advertising analytics 2.0. Harvard Business Review, 2–10.
Sorofman, J., A. M. Virzi, and Y. Genovese (2016). Gartner cmo spend survey 2016–2017: Budgets climb (again!) as marketers juggle more demands. Technical report, Gartner Inc.
Sweetwood, A. K. (2016, oct). How an analytics mindset changes marketing culture. Harvard Business Review.
Yeomans, M. (2015, jul). What every manager should know about machine learning. Harvard Business Review.