Companies of every size struggle to keep pace with the data that runs everything—especially customer data. Theoretically, it’s possible to know everything about a client or prospect, from the perfume they bought last week to their preferred vacation amenities. This can that get creepy in a hurry, as Target discovered in 2002, when it used data to identify when a woman was pregnant—and therefore likely to spend more money.
But there are even greater challenges for today’s marketing teams:
This is not another blog about how AI is the answer. It’s not. Artificial intelligence can do many things, but not this. What marketing CMOs and their teams must do, fundamentally, is deploy precision output automation. This can provide a container for properly leveraging Artificial Intelligence. AI is best understood as a tool that can help; by no means is it a complete solution.
In our endless arguments over data and AI, the subject of typography is often ignored. This is a mistake. The size, shape, and positioning of words and letters is crucial, whether the medium is print or digital. As Ben Jura wrote for the American Marketing Association, “The way that words are shaped and their letters drawn offers predictable meanings for the people who look at them beyond the meaning of the word itself… [This] makes typography one of the most powerful tools of design and marketing when creating brand meaning and influencing consumer sentiment.” (emphasis added)
Good typography—and good design in general—has other measurable benefits, including improved conversion rates, brand perception, and user experience. But the problem is that most typographic “engines” generating personalized advertising and marketing content remain primitive. A number of common situations can cause these systems to produce mediocre typographic results (at best) or make the ad illegible (at worst):
Of course there is one modern application specifically targeted at emulating traditional, high-quality typography: Adobe InDesign. Originally launched in 1999, InDesign gave digital designers a level of typesetting control that was unavailable in programs like QuarkXPress. More importantly, it offered the means to fully automate content production—without sacrificing typographic quality.
The short answer is no, of course. The probabilistic model used by generative AI can predict the next word in a sentence or identify whether a photo is of a cat or a dog—probably—based on the right quantity and quality of training data. But to personalize an ad or a brochure effectively, the subject data must be specific, accurate, and secure. AI-based predictions/guesswork, however impressive, will always run the risk of alienating customers and prospects.
AI shines at brainstorming and data/content mining: for example, learning from your collection of templates to create many more: still, these are ideally curated by humans. It might analyze 100 possible hero images, giving you the top 12 that meet specified goals in the light of lessons from past campaigns, yet a human is needed to make the final choice. AI might analyze content to validate correct pre-press settings and content integrity, yet the results should be warnings, validated by a human. A minor print job error can be extremely costly, and it’s dangerous to let AI (at least in its current state) automatically make changes to subtle configurations that can jeopardize production.
AI is a tool, a feature, that ideally exists inside of more traditional applications. The bulk of intelligent marketing automation requires a deterministic data approach—driven by rules-based matching and transparency. Such an approach can include a designer’s intent (to create the original content), as well as the marketer’s insight into customer preferences (to create real-world segmentation rules). Contrary to current mythology, artificial intelligence has neither intent nor insight. It guesses at them, but deterministic programming can rigorously express them and constrain AI to be more consistently successful. A hybrid approach is required for optimal content generation.
InDesign provides the ideal basis for a deterministic data approach to automation. In the hands of capable designers and marketing professionals, it can take a well-planned visual campaign and segment it according to existing customer data without sacrificing visual quality. InDesign Server or IDS, introduced in 2005, has demonstrated this on a large scale, which is the subject of another blog. Products such as Silicon Publishing’s Paginator utilize IDS for high-volume, data-driven document and ad production. InDesign is extremely well-exposed to automation, so it is easily connected to any modern AI solution.
One-to-one marketing, based on volumes of individual customer data, has been the holy grail of marketers for decades. In the short term, automating the process of personalized ads has reduced labor costs while also increasing advertising frequency: AI can improve this, but is often used recklessly. Unfortunately, the unintended consequences have included customer fatigue and/or resentment at the volume of junk advertising.
To counter this problem, marketing teams need to increase the quality of their human-specific abilities, not just add AI to the mix. Designers need to improve their creative skills, including their facility with tools like InDesign and AI in the context of automation. AI alone will not solve content automation challenges, as it only thrives on top of a solid, deterministic, solution.
Marketing pros also need to step up their game, creating customer segmentation rules based on their actual knowledge of customers and the differences among them. If they don’t have the technical expertise to do this, their companies need to find people who can help. The advent of AI and the maturity of software bring potential tactics that were never possible before.
Tools like IDS and Paginator give marketing teams the ability to create effective, scalable content and campaigns. AI adds substantial value to both of these products, which are easily connected thanks to the extreme extensibility they offer. But only human knowledge and intuition—not AI—can take full advantage of modern tools. AI provides huge value: it is a great servant, but a poor master.
By presenting text in a consistent, visually appealing, and recognizable way, good typography supports positive brand awareness, increased conversion rates, and contributes to an overall positive user/consumer experience.
Generative or probabilistic AI can generate words and images based on predictions derived from existing training data, but it cannot understand a customer’s thoughts or needs. Systems based on deterministic data models can extend the capabilities of human designers and marketing professionals who can understand and act upon their knowledge of customer needs.
When an ad or document contains individual-specific information, it can be more relevant or helpful to that individual, provided the information is accurate and obtained/used with the recipient’s consent. The content itself must also be relevant and timely. The risks of over-use or misuse of personalization include frustration and anger at the originator, ultimately leading to loss of business. Automating this process will reduce labor costs in the short term but, if done indiscriminately, poses the risk of accelerating the recipient’s disengagement, and long-term reputational damage to the marketer.
In the long run, AI can absolutely save money, and this will be increasingly the case as AI matures. However, initially there is a significant investment, and a substantial learning curve. The promises of instant gratification are generally false. It is a myth that AI can currently take over content automation on its own. It shines in the context of hybrid solutions using deterministic software, AI, and human direction and validation.