AI to Reshape Creative and Measurement, and the Publisher-Advertiser Relationship

The already blisteringly fast pace of AI innovation is accelerating. While “predictive” AI has long quietly underpinned much of our recent advertising and media technology, it often was unglamorous and behind the scenes. But the rise of generative AI is both supercharging long-standing capabilities as well as delivering entirely new ways to reach and persuade consumers.

OpenAI, the major tech giants, and a burgeoning array of startups release so many (often jaw-dropping) innovations each day that simply keeping up is nearly a full-time job.

As AI continues to evolve and improve, it promises to permeate and reshape nearly every element of knowledge work – liberating humans from time-consuming tasks and producing ways to better create content, parse data, communicate, and understand the world. The impending AI wave will undoubtedly and irrevocably alter nearly every industry, but the coming changes will be especially wrenching for predominantly digital industries like advertising and media. But time will not stand still, and companies that ignore the changing tides do so at their own (great) peril.

The AI-powered golden age for advertising and media

Marketers, it seems, are already awake to AI’s potential. A 2024 Marketing AI Institute study revealed 99% of respondents are personally using AI in their personal lives or professional endeavors.

Currently, marketers are harnessing generative AI tools to generate significantly more customized content, which they will advertise to hyper-specific segments. Meta and Adobe are among the many startups that enable people to create AI videos with text prompts and other inputs.

Publicly available tools like ChatGPT, Google Gemini, Microsoft Copilot, Claude, and others can be prompted into effective copywriting, and a new crop of startups is already emerging to automate tasks like copywriting and testing thousands of different taglines with relative ease.

AI is also beginning to revolutionize content creation. Image Generation tools like MidJourney, DALL-E, and Firefly already dramatically lower the cost of prototyping and concept art, and creative studios are figuring out how to use AI broadly to facilitate rapid production, repurposing, and remixing of materials, while significantly reducing overall ad production costs.

Cheaper, more effective ads help not just advertisers but publishers as well – the more value advertisers derive from running an ad, the more they’ll be willing to pay for the ad space to run it. This helps publishers who have a) many attention-minutes to sell, b) distinctive audiences, c) quality audience data, d) compelling content, and/or e) sophisticated recommendation algorithms.

AI’s outsized role in enhanced measurement

While marketers have long had access to robust measurement tools, the real challenge lies in their effective utilization.

• Marketers must design experiments that actually test—and either validate or refute—hypotheses about which creative and media strategies resonate best with their audiences
• Marketers must apply rigorous statistics to glean useful conclusions from measurement data (while remaining humble about what questions their data cannot )
• Marketers must communicate complex results to (often non-technical) stakeholders
• Marketers must put insights into practice. Few marketers are equipped to make 1,000 small changes to a campaign, even when doing so may double a campaign’s performance

Marketing Technology News: MarTech Interview with Keith Kazerman, President of Streaming @ Locality

Each step requires effort and expertise.

AI can materially help with (and in the near future, may entirely automate) each of these steps. LLMs have “read” every textbook on statistics and experiment design; LLMs (particularly “reasoning”-enhanced LLMs like OpenAI’s “O1” model) can manipulate statistical equations and write analytical Python code; LLMs can write compelling, audience-tailored prose, make insightful, actionable presentations, and patiently answer stakeholder questions. And LLMs (together with classic machine learning and DSP APIs) can put campaign optimizations directly into practice.

To be fair, generative AI still requires hand-holding to perform each of these steps. But model abilities will continue to improve rapidly over time. It is wise to focus more on AI’s trajectory than its current foibles.

Real measurement is an evaluation of what works, not an endorsement of what has been done. Embracing measurement requires (often uncomfortable) change, beginning with an acknowledgement that some of the things you’re doing aren’t working. Airtight methodology and transparent, automated analysis can bolster confidence in the conclusions drawn, making it easier to convince stakeholders to adapt, even when those conclusions are challenging to accept.

As cutting edge measurement, analytics, and optimization via AI become broadly available and cheap, there will be little to no excuse for not applying the best tools and demanding the highest standards of performance from every ad campaign. The increasingly antiquated notion of a divide between “brand” and “performance” advertising is poised to dissolve, both philosophically and practically, as teams, agencies, and marketing departments reorganize around a more integrated approach.

Publishers’ role in the AI-everything future

AI will also empower more advanced AI assistants that consumers can use to make purchasing decisions without navigating to publishers’ sites. New search engines like Perplexity (and ChatGPT itself)  are already streamlining the search process–creating expectations for a single, definitive answer to queries, rather than a list of links with multiple options.

Brands that once relied on advertising through publisher websites as a way to reach consumers will likely pivot to focus on offering compelling data and product information that AI agents can use to determine the best match for users’ needs and preferences. This shift will redefine how brands market in an AI-centric world (and drain money from advertising budgets, hurting publisher revenues).

What should companies do?

AI has already disrupted much of the marketing sphere, and the disruption will only intensify from here. The traditional publisher-advertiser dynamic is entering a new phase. While more effective AI-powered advertising will help publishers in the short term by making ad space more valuable, the future will make it easier for brands and consumers to connect without going through publishers at all, thereby challenging the established model.

The rapid evolution and advancement of AI is bound to bring unexpected developments in creative execution, measurement, and the dynamics between publishers and advertisers. And as measurement becomes easy, cheap, and ubiquitous, brands will converge rapidly toward the techniques that work to grow their businesses. Little room will remain for any player in media and advertising that does not add meaningful, measurable value above and beyond what AI can deliver itself.

Every stakeholder in the advertising ecosystem should embrace change and experimentation, recognizing that AI will transform their businesses (with or without their permission.) The future is uncertain, but one truth is clear: no company is safe standing still.

Marketing Technology News: First-Party Data is Still the Right Bet