White Paper

Can Machines be Creative? How Technology is Transforming Marketing Personalization and Relevance

Gerry Brown
July 2017

IDC OPINION

The most important item on marketing leaders' digital advertising agendas both now and in the future is driving customer loyalty, according to a recent global IDC research study sponsored by Criteo. Marketers believe that a key enabler of customer loyalty is "support for a consistently high quality brand experience" and they plan to deliver this using digital personalization to improve the relevance of their communications to their customers.
Marketers understand the value of digital personalization, using it broadly across all their marketing communications activity. In online advertising, marketers personalize advertising insertions to some degree across all stages of the buyer journey; these personalization activities are showing positive performance results, so marketers now intend to escalate their investments.

Some 60% of marketers plan to do "significantly more" or "more" online advertising between now and 2020, making it a high investment growth area for enterprises.
Marketers are moving away from creative content produced by humans toward content produced by machines.

Advertising personalization is complex and intricate, and requires real-time responsiveness to customer buying behaviors "in the moment" of consideration and buying to be effective. The speed and dexterity required to activate such real-time personalization can only, in practice, be achieved using machines.
As a result, marketers are moving away from creative content produced by humans toward content produced by machines, and plan to use machines to automate the personalized delivery of creative content via online advertisements for individual consumers. 64% of marketers believe optimized message targeting and real-time personalized advertising insertions are key areas where machines will deliver business benefits by 2020.

This interest is reflected in IDC's market forecasts. IDC predicts that spending on AI software for marketing and related businesses will grow at a very fast cumulative average growth rate (CAGR) of 54% worldwide, from around $360 million in 2016 to over $2 billion in 2020.

Although marketers are aware of the value and benefits of machine learning within the context of personalization, few are currently using it in this way. This is due to a lack of internal expertise in machine learning and trust in machine-learning technology to provide the required level of customer data privacy and brand control.

A perceived risk is that machine learning may not remain true to brand with respondents citing being "unsure we will be able to manage and control our brand and design" as a key reason limiting current machine-learning adoption.

IDC believes these brand concerns will gradually evaporate as machine learning becomes more established within online advertising operations, and forecasts that machine learning will become pervasive across all elements of the advertising technology industry supply chain over the next five years to 2022.

Key guidance areas for marketers and agencies include:

• Make personalization a central over-arching business and marketing strategy to deliver superior customer experiences.
• Evaluate how machine learning might be applied to your online advertising personalization activities using this research report as a benchmark. Creative content personalization should be considered in the short to medium term as a major area of investment.
• Take your one-to-one personalization strategies to the next level by targeting different stages in the buyer journey with different creative elements within your advertisements. Machine learning is a key enabler of this.

METHODOLOGY

IDC interviewed 459 marketing executives in January and February 2017. Of these, 30% were chief marketing officers (CMOs), VPs of marketing, or marketing directors; 35% were in marketing management, and the remaining 35% were in digital or online advertising roles. 58% were senior decision makers in regards to online advertising investments, and 33% directly influenced decision making in online advertising.

There was a relatively even regional split of the interviews: 155 were conducted in Europe (France, Germany, the Netherlands, the U.K., Spain), 154 were conducted in Asia/Pacific (Australia, China, India, Indonesia, Japan, Singapore), and 150 were conducted in the U.S. All the companies interviewed had over 250 employees and over a third (34%) were conducted with very large companies with over 1,000 employees. All the companies interviewed were in the retail or travel sector.

IN THIS WHITE PAPER

This study reveals how the largely manual process of advertising creative and design work will increasingly incorporate machine learning to automate the delivery of mass individualized and personalized advertisements. Programmatic advertising has automated many of the transactional elements of the advertising supply chain, and IDC believes that the next bastion is the creative content and copy process.

This paper explores the current and future trends of using digital for advertising creativity and highlights the drivers and concerns of using programmatic advertising creative content, revealing how machine learning has the potential to transform future advertising and creative methods by delivering mass creative personalization.

HOW MARKETING COMMUNICATIONS ARE BEING PERSONALIZED

Personalization, or "a segment of one," has for long been the holy grail for marketers. Personalization has the power to increase the relevancy and potency of marketing communications, customer sentiment, and advocacy toward the brand, propensity for conversion, and brand loyalty. These attributes maximize the brand's opportunity for customer retention and life-time customer value (LTCV) to assure enterprise revenues and cash flow.

Programmatic advertising has automated many of the transactional elements of the advertising supply chain, and IDC believes that the next bastion is the creative content and copy process.

Personalization is therefore clearly critically important to most brands. The section below describes how the concept of personalization is being applied in digital marketing and advertising operations and the business benefits gained.

FIGURE 1

Digitizing Marketing Communications for Personalization is Now Table Stakes

Q. To what extent does your company currently use digital technologies to personalize its marketing communications?

The largest segment of our sample, 34%, took the middle road of "to a moderate extent" to describe their efforts to personalize their marketing communications. This is perhaps understandable. Personalization is hard to implement and it is difficult to measure the relative success of personalization against industry norms and competitors.

n = 459
Source: IDC Online Marketing Survey, Criteo, February 2017

Almost the same number (32%) of marketers took a more aggressive view of "to a great extent" and 10% took the extreme view of "completely." Typical examples of retail personalization include sending emails to customers with offers to celebrate their birthdays and communicating the availability of new products that match customer preferences to encourage website visits. Retargeting is also a common retail application.

U.S. companies were almost twice as likely to respond "completely" (15%) than respondents in Asia/Pacific and EMEA (8%), which reflects a higher level of personalization maturity in the U.S. For most retail and travel brands, personalization is a core business and marketing strategy.

However, nearly a quarter (23%) of brands globally have yet to take personalization seriously, saying they personalize their marketing communications only "to a little extent" or "not at all." This could be a dangerous road to take given the potential marketing competitive disadvantage in the longer term. Investment in personalization is "table stakes" for effective competition in the modern marketing environment.

FIGURE 2

Personalization of Ad Insertions is Currently Focused on Conversion

Q. To what extent does your company personalize its digital advertising insertions to drive the following five stages of the customer journey?

Source: IDC Online Marketing Survey, Criteo, February 2017

We wanted to understand if brands are taking different approaches to online advertising personalization in different stages of the buyer journey using a traditional five-stage consumer buying model of awareness, consideration, evaluation, purchase, and post-purchase.

Most effort is currently placed at the end of the buyer journey to facilitate conversion. This often takes the form of an extra personalized communication touch, especially for high-value customers, to increase loyalty and push them over the conversion threshold. Similarly, personalized retargeting is particularly prevalent in a retail ecommerce transaction context of website cart abandonment.

"Buyer journey" offers great opportunities for machine learning that can "learn" the content optimization required at each customer journey stage.

However, marketers make little differentiation between the level of personalization applied for each of the five buying stages, with less than 10% difference between the highest level (stage 4: purchase) and the lowest level (stage 2: consideration). Asia/Pacific marketers are ahead in recognizing the value of personalization in the perception of the modern customer. Brands in Asia/Pacific rate their level of personalization as being higher than in the U.S. across four of five buying stages, while EMEA lags significantly across all stages. Market penetration of online advertising personalization is higher in Asia. Nearly all Asian brands use personalization in online advertising to a moderate extent or more, which is proportionately less likely in the U.S. or EMEA.

FIGURE 3

Digital Enables Automation of Creative Content Personalization

Q. Is your company currently using or planning to use digital technologies to automate the personalization of creative content?

n = 459
Source: IDC Online Marketing Survey, Criteo, February 2017

A significant proportion, over 50%, of our sample uses digital technologies to automate the personalization of creative content in marketing communications and over 30% plan to use digital technologies to personalize creative content in the future.

There are seven creative elements that respondents either automated or planned to automate in their personalization activities. The most obvious of these "images" (for example, the picture of a car previously viewed on a website visit) came top. "Images" are perceived as the most effective personalization method across all stages of the five buyer journey stages, and had top usage for awareness, evaluation, and post-purchase, with stand-out usage for evaluation.

These overall results are like those of the five-stage buyer journey in Figure 2 in that only small differences in personalization are made across the seven creative elements. The marginal exception was "call to action," which lagged with 46% usage. However, this produced by far the highest level of satisfaction of respondents as it increased the likelihood of conversion. 36% plan to invest in call-to-action content personalization in the future.

Using contemporary technologies, it is difficult to identify and act in real time on the multitude of marketing communications creative content combinations available. Machine learning helps to continuously uncover the nuances of creative content performance across each of the seven elements above, so that optimized and personalized creative combinations can be delivered at each customer buying stage.

FIGURE 4

Many Business Benefits are Gained from Digital Personalization

Q. What benefits does your company receive or expect to receive from digital personalization?

Source: IDC Online Marketing Survey, Criteo, February 2017

Brands are satisfied with the results of their personalization efforts to date, which will spur more future investment in personalization. Brand awareness/positive brand associations is the most important benefit brands receive or expect to receive from personalization. However, it is clear from the clustering of responses that many benefits are evident, with six items garnering over 50% response. For those that are still not invested in personalization, this graphic provides conclusive evidence that they risk being left behind in terms of their marketing, advertising, and overall business performance.

CURRENT AND FUTURE USE OF MACHINE LEARNING FOR PERSONALIZATION

Previously we discussed the current digital personalization strategies being pursued and the need for an overarching personalization business strategy. In this section, we review the context and use of machine learning in digital personalization efforts.

Machine learning is a form of artificial intelligence that enables computers to learn without explicit programming. Machine learning uses algorithms that learn from data, such as continuously improving the prediction of future consumer behavior, with increasing levels of forecast accuracy as the volumes of data increase. Therefore, machine-learning algorithms can learn, think, and iterate far beyond the scope and capabilities of traditional simplistic rules-based systems.

Machine learning opens up a completely new operating paradigm for advertisers and marketers, devoid of much of the time-consuming and painstaking administrative work that characterizes the back office of marketing. But machine learning also opens up opportunities to execute marketing activities that have not previously been practical. Personalization of marketing communications and advertising at scale, the core subject area of this paper, is one such application.

FIGURE 5

Familiarity with Machine-Learning Applications is High, but Usage is Low

Q. Which of the following best describes your familiarity with machine learning as a technology that could potentially be used for communications personalization?

n = 459
Source: IDC Online Marketing Survey, Criteo, February 2017

Although 83% of our sample are familiar with machine-learning applications for communications personalization, only 14% are using it today. Early machine-learning applications in retail include website applications and push notifications. Hence we are at the very early stages of "machine learning for personalization" market development. However, we will shortly enter a dramatic growth phase — 33% of marketers are planning to invest in machine-learning technology for communications personalization, which suggests that latent demand is strong and the market will grow significantly in the coming years.

Marketers believe that machine learning will have the highest usefulness in media planning and media execution, executing multichannel campaigns, and creating personalized advertisements. Many see opportunities for machine learning in churn predication and life-time value modelling.

FIGURE 6

Machine Learning Delivers Applications Today, with More to Come by 2020

What communications applications do you think machine-learning technologies can (1) deliver today or (2) will in addition be able to deliver in 2020?

n = 379–381. Note: Percentages do not add up to 100% as respondents could respond both to (1) today and (2) in 2020.

Source: IDC Online Marketing Survey, Criteo, February 2017

Two-thirds of respondents believe that machine-learning technologies for personalized headlines and advertising copy; personalized advertising design formats such as layout, color sets, and sizing; and personalized advertising creatives are deliverable today. The use of real-time personalized advertising insertions and optimized message targeting were perceived more as machine-learning technologies of tomorrow. Hence, machine-learning technologies are perceived to have wide applications in online marketing and personalization both today and in the future.

The advertising market is in a state of transformational flux where the role of advertising itself is moving from a standalone activity to becoming an integrated part of the customer experience through the entire purchase cycle. Advertising that isn't relevant or part of a broader customer experience, or is intrusively trying to create awareness, is no longer welcome in many households.

Technologies that facilitate relevance through personalization are high on the list of brands' future investment plans.

FIGURE 7

Data-Driven Advertising Will Drive Customer Loyalty and Brand Relevance

Q. Which of these are part of your company's advertising agenda today? And which will be the most important to your company in the future? Choose the top 3.

n = 459

Today, brands' advertising agenda is focused on generating customer loyalty (cited by 41%), gathering customer data (36%), and remaining a relevant brand that is at the top of consumers' minds (34%). These will continue to be the most important aspects of advertising for retailers and travel companies. Interestingly, though using advertising "as a method of creating social and viral customer conversations" is of lower importance today, it will rise in importance as social continues to exert a greater influence over the marketing mix.

Marketers now believe that advertising should drive customer loyalty and engagement, and staying relevant and creating conversations will supplement customer loyalty in the future. Personalization is the strategy to achieve this goal, and machine learning is a key technology enabler of "personalization at scale."

CHALLENGES/OPPORTUNITIES

The online advertising industry is a young industry, barely 10 years old in its current form, and it has changed beyond all recognition in that time. Real-time bidding (RTB) has transformed the industry, and indeed the whole advertising supply chain, and machine learning is set to transform the industry in a similar way yet again through the delivery of personalized advertisements.

Although the growth and opportunity for both brands and suppliers is enormous, there are machine-learning challenges to be overcome. These include brand adoption of machine learning as an online advertising transactional mechanism and to deliver greater value exchange between the enterprise and its customers. The latter requires brands to shift their focus from digital delivery to customer experience and value-based marketing.

FIGURE 8

Machine Learning Has Yet to "Cross the Chasm" for Personalization

In regards to machine learning within the context of personalization, "lack of internal expertise and/or technical constraints," "lack of trust in managing first-party data security and compliance with automated machine learning," and "unsure we will be able to manage and control our brand and design using machine learning" are the top 3 challenges (each with over 30% of responses).

The biggest concern — "lack of internal expertise and/or technical constraints" — is a common concern for new technology adoption. Indeed, 30% also cited "a lack of internal funding/buy-in of the IT department and other decision makers" as a concern. Monetizing machine learning is also a concern for 30%, while 28% consider machine learning to be still too immature/unproven, and not trusted to deliver on its promise. These are educational and training challenges for suppliers that will likely disappear as the technology gains more widespread acceptance and crosses into mainstream early-majority market life-cycle adoption.

Machine learning for personalization is a subsegment of the online advertising technology category. The latter is itself complex and under-served in terms of resourcing due to the meteoric level of growth of the online advertising industry, and the specialist niche competencies required. Hence, outsourcing of advertising technology services provision to suppliers and agencies is often the norm due to a lack of skilled and specialized talent to resource in-house deployments.

CONCLUSIONS

Machine learning improves the accuracy of targeting over time by learning from the responses to advertisements by individual consumers and by continuously monitoring their online behavior.
Marketers acknowledge that machine learning will be an increasingly important technology for use across the advertising supply chain. Marketers also appreciate that machine learning will improve customer experiences through the delivery of more timely, contextually relevant, personalized online advertisements that deliver more customer advocacy, loyalty, and customer life-time value.

There is a latent need for the automation of creative copy development using machine learning. Machine learning therefore refines personalized advertising creative content delivered over time in a continuous performance optimization process. It also removes tedious data preparation and analysis work to free up creative staff to work on creative ideas, fueled by a flow of relevant, real-time behavioral data. Creative staff will continue to provide the "base" creative content, using machines to deliver relevant personalized communications at scale.