AI in Ad Creatives: Automating Performance Without Losing Creativity

Industry & Competitive Context

The advertising industry is undergoing a structural transformation driven by generative artificial intelligence. What began as isolated experiments in copy optimization has matured, within half a decade, into an enterprise-grade capability deployed across creative production, performance testing, and audience personalization. According to McKinsey's State of AI 2025 report, 78% of organizations now use AI in at least one business function, with marketing and sales consistently ranking among the highest-value domains for AI deployment. McKinsey estimates the economic potential of generative AI across 63 enterprise use cases at $2.6 trillion to $4.4 trillion annually, with marketing representing one of the largest pools of addressable value.

The global generative AI market, valued at approximately $20.28 billion in 2024, is projected to reach $189.65 billion by 2033, reflecting a compound annual growth rate of 28.2%. Within the media and advertising sub-sector specifically, the generative AI market was valued at $2.6 billion in 2024 and is expected to reach $7.48 billion by 2028. These numbers represent more than investment appetite — they reflect a fundamental re-architecture of how brands produce, test, and deploy creative content at scale.

The competitive pressure driving this shift is structural. In a multi-platform media environment where a single campaign must produce assets across social video, display, search, email, and out-of-home formats — each with distinct dimensions, durations, and audience expectations — traditional creative production models have become bottlenecks. Human creative teams, constrained by time and cost, cannot generate the volume of variants required for meaningful A/B testing at scale. AI dissolves this constraint, enabling brands to generate hundreds of asset variants, test them simultaneously, and iterate on the highest performers in near real-time. The central strategic question for brand leaders is no longer whether to adopt AI in creative production, but how to do so without eroding the emotional and cultural resonance that distinguishes great advertising from merely efficient advertising.

Brand Situations Prior to AI Adoption

The case for AI-driven ad creatives gains clarity when examined through the lens of specific, documented brand situations. Three cases — JPMorgan Chase, Coca-Cola, and Unilever — provide the most publicly documented evidence of AI integration in creative strategy and ad performance, each representing a distinct industry, creative challenge, and implementation model.

JPMorgan Chase, one of the largest financial services companies in the world, faced a problem common to large enterprises: the inherent conservatism of financial services marketing, combined with the enormous volume of copy variants required across products, channels, and customer segments. The bank's marketing team recognized that human copywriters, relying on subjective judgment and brand experience, were producing copy that was competent but not necessarily optimal. The question was whether a data-driven approach could surface language patterns that outperformed human intuition.

Coca-Cola, a brand with over a century of creative heritage and among the world's most recognizable visual identities, faced a different challenge. As digital channels fragmented audience attention and younger consumers increasingly expected participation rather than passive consumption, the brand needed to find a way to scale consumer co-creation without surrendering brand control. It also needed to establish a credible position on AI — not as a threat to human creativity, but as a tool that amplifies it.

Unilever, operating over 400 brands across 190 countries and managing an advertising supply chain of extraordinary complexity, faced an operational creative challenge. The sheer volume of assets required for localized, personalized campaigns across its Beauty & Wellbeing portfolio — including brands such as Dove, Vaseline, POND'S SKIN INSTITUTE, and CLEAR — far exceeded the throughput capacity of conventional agency relationships.

Strategic Objectives

Across these three cases, the documented strategic objectives reflect three distinct but related applications of AI in creative advertising.

JPMorgan Chase's objective was performance optimization at the copy level — using AI to identify which specific words, phrases, and emotional triggers drove the highest click-through rates across digital ad formats, and to replace subjective creative judgment with data-driven language selection.

Coca-Cola's objective was brand participation at scale — using generative AI to invite consumers into the creative process while maintaining governance over how iconic brand assets were used, thereby deepening emotional engagement and cultural relevance without requiring traditional production budgets for each piece of content.

Unilever's objective was operational efficiency without creative compromise — using AI to accelerate asset production across markets and languages, enabling faster campaign deployment and more agile iteration while maintaining the brand standards of its premium beauty portfolio.

Campaign Architecture & Execution

JPMorgan Chase and Persado (2016–2019)

In 2016, JPMorgan Chase began a pilot with Persado, a company that uses AI, data science, and computational linguistics to generate marketing copy. Persado's platform — which it calls the "Message Machine" — draws on a database of over one million tagged and scored words and phrases organized by emotional and motivational categories. The system generates copy variants not by randomly assembling language, but by applying machine learning models to predict which emotional registers will drive engagement with specific audience segments.

During the pilot, Chase applied the Persado platform to its Card and Mortgage business lines, testing AI-generated copy across landing pages, direct mail, display advertising, and social ads. The results were sufficiently compelling that in July 2019, Chase announced a five-year enterprise-wide deal with Persado, publicly disclosed through a joint press release and carried by the Nasdaq news distribution service. As part of that announcement, JPMorgan Chase CMO Kristin Lemkau stated: "Machine learning is the path to more humanity in marketing. Persado's technology is incredibly promising. It rewrote copy and headlines that a marketer, using subjective judgment and their experience, likely wouldn't have. And they worked."

The architecture of execution was notable for what it preserved as much as what it automated. Human marketers defined the campaign objectives, audience parameters, and brand guardrails. Persado's AI operated within those parameters to generate and rank copy variants. The creative team reviewed and approved outputs before deployment. This human-in-the-loop model was critical to maintaining brand voice while scaling creative experimentation.

Coca-Cola's "Create Real Magic" (2023)

In March 2023, Coca-Cola launched "Create Real Magic," a consumer-facing generative AI platform developed in partnership with OpenAI and Bain & Company. The platform, hosted at CreateRealMagic.com, combined the capabilities of GPT-4 and DALL-E to enable digital creatives worldwide to generate original artwork using Coca-Cola's archived brand assets — including the contour bottle, the brand's typography, the Haddon Sundblom-originated Santa Claus imagery, and polar bear characters.

The campaign architecture was deliberately governed. Rather than opening brand assets to unmediated AI generation, Coca-Cola built a structured sandbox in which users could interact with a curated library of brand elements. The platform ran initially as a limited-time experience to create urgency and manageability. A group of 30 creators selected from all submissions were invited to the Coca-Cola global headquarters in Atlanta for the "Real Magic Creative Academy," a three-day workshop held in partnership with the company's Global Design and Creative teams. Selected artworks were subsequently featured on billboards in Times Square in New York and Piccadilly Circus in London.

Internally, Coca-Cola's AI governance structure involved a senior leadership Digital Council and a central AI team led by Pratik Thakar, Global VP and Head of Generative AI. Any employee seeking to deploy a generative AI tool for a project submitted a formal request to the central team for review, ensuring legal compliance and preventing duplication. This governance-first architecture was a deliberate strategic choice, not an afterthought.

Unilever's Beauty AI Studio (2024–2025)

Unilever began working with The Brandtech Group in 2023 and started testing the Pencil AI platform in the first quarter of 2024. The resulting system, known as Beauty AI Studio, was developed in collaboration with in-housing specialist Oliver and deployed across Unilever's Beauty & Wellbeing power brands. The system covers the full content lifecycle — from ideation to personalization — across priority markets. As part of its rollout, Unilever provided AI fluency training to 25,000 staff members.

The execution model integrated AI into the asset production pipeline rather than replacing the creative brief or brand strategy process. Human creative teams and brand managers defined the strategic direction and brand standards; AI then accelerated the production, adaptation, and performance optimization of assets within those parameters.

Positioning & Consumer Insight

Across these three cases, a consistent positioning logic emerges that has significant implications for how AI is integrated into creative strategy. In each instance, the brand explicitly positioned AI not as a replacement for human creativity but as an amplifier of it. This positioning was both authentic and strategically necessary.

JPMorgan Chase's disclosed approach emphasized that AI-generated copy often produced language that human marketers "likely wouldn't have" conceived through subjective judgment — framing this as a benefit rather than a displacement. Coca-Cola's entire campaign architecture was built around the premise that AI unlocks human creativity by democratizing access to brand assets and tools, rather than substituting for human artistic vision. Unilever's internal communication around Beauty AI Studio emphasized that AI enables creative teams to "quickly double down on what works" — positioning the technology as an accelerant for human creative decisions, not a replacement for them.

The consumer insight underpinning this positioning is a documented tension in the market: consumers are simultaneously drawn to the novelty and personalization enabled by AI, and concerned about the loss of authenticity and human craftsmanship in brand communication. Brands that have navigated this tension most successfully have done so by designing AI into a role that is visible but clearly supportive — enabling human creators rather than replacing them.

Media & Channel Strategy

Where publicly documented, the channel strategies in these cases reflect deliberate choices about where AI-driven creative performs most effectively.

JPMorgan Chase's AI-generated copy was deployed across landing pages, direct mail, display advertising, and social media ads — essentially every performance marketing channel where copy variation and click-through optimization are directly measurable. This channel selection was strategically sound: performance marketing channels generate the data feedback loops that allow AI models to improve, and the success metric (click-through rate) is unambiguous.

Coca-Cola's "Create Real Magic" campaign leveraged a purpose-built microsite as its primary activation point, with social media serving as the amplification layer for user-generated content. The out-of-home placements in Times Square and Piccadilly Circus provided high-visibility earned media moments for the most celebrated creative outputs, translating digital participation into physical brand presence.

Unilever's Beauty AI Studio was deployed across digital channels relevant to its beauty brands, with performance measured through Video Completion Rate and Click-Through Rate — metrics that reflect both engagement quality and direct response effectiveness.

Business & Brand Outcomes

The documented outcomes from these cases provide the strongest publicly available evidence for the performance claims made by AI creative platforms.

In the JPMorgan Chase case, the officially disclosed outcome of the Persado pilot was a click-through rate increase of up to 450% on ads generated by the AI platform, compared to a baseline range of 50–200% for human-written ads tested across the same channels. This result was disclosed in the official joint press release issued by JPMorgan Chase and Persado on July 30, 2019, distributed via Business Wire and carried on the Nasdaq news wire. The significance of this outcome was sufficient to justify a five-year enterprise-wide commitment — the disclosed scope of the deal, though financial terms were not made public.

In the Coca-Cola case, the "Create Real Magic" platform generated thousands of pieces of user-generated content during the initial campaign period in March and April 2023, according to publicly available reporting. The platform was noted to have resonated particularly strongly with Gen Z audiences. Coca-Cola's official financial filings for Q1 and Q2 2023 documented net revenue increases of 5% and 6% respectively, though the company's public disclosures do not attribute these results specifically to the "Create Real Magic" campaign, as multiple business drivers were active in that period.

In the Unilever case, the company's own official communications — published on Unilever's corporate newsroom in September 2025 — disclosed that the Beauty AI Studio enabled asset production up to 30% faster than prior workflows, while Video Completion Rate and Click-Through Rate both more than doubled across the Beauty & Wellbeing brands using the system. These disclosures represent unusually specific and authoritative public documentation of AI creative performance outcomes.

No verified public information is available on the internal cost structures, agency fee savings, or team headcount changes associated with any of these programs.

Strategic Implications

The evidence across these cases supports several strategic conclusions that extend well beyond the specific implementations described.

First, AI's most documented advantage in ad creative is not imagination but iteration. The creative breakthroughs in these cases — whether JPMorgan Chase's emotionally resonant copy variants or Coca-Cola's consumer co-created artworks — were not produced by AI acting alone. They emerged from AI's ability to generate, test, and optimize at a volume and speed that human teams cannot replicate. The strategic lesson is that brands should deploy AI where iteration velocity creates direct performance value: performance copy, A/B testing frameworks, and multi-market asset adaptation.

Second, brand governance is the prerequisite, not the afterthought. Coca-Cola's "Create Real Magic" architecture — a curated asset library, a structured platform, a human curation layer, and a senior Digital Council — demonstrates that AI-scale creative production and brand integrity are compatible only when governance is designed in from the start. Brands that launch AI creative programs without equivalent governance frameworks risk brand dilution that is difficult to reverse.

Third, the human-AI creative relationship is most productive when AI operates within human-defined parameters rather than replacing the creative brief. In every documented case of successful AI creative deployment, human teams defined the strategic objectives, the brand guardrails, and the success metrics. AI operated within those constraints to optimize execution. The failure mode — not yet documented in these cases but implied by the governance investments made — is AI operating outside defined parameters, producing outputs that perform on narrow metrics while degrading brand equity.

Fourth, the organizational dimension of AI creative adoption is material and underappreciated. Unilever's disclosed investment in AI fluency training for 25,000 employees reflects an understanding that the technology itself is only one component of the capability being built. The human skills required to prompt, review, curate, and improve AI creative outputs are distinct from both traditional creative skills and traditional data science skills — and they must be deliberately developed.

Fifth, the measurement architecture for AI creative requires careful design. Click-through rate, the primary metric in the JPMorgan Chase case, is a narrow measure of performance that does not capture brand equity, long-term customer relationship quality, or downstream conversion. Brands deploying AI in creative must ensure that the optimization function — the metric the AI is trained to maximize — is aligned with actual business objectives, not merely with the metrics that are easiest to measure in digital channels.

Discussion Questions

1. JPMorgan Chase's partnership with Persado produced a documented 450% increase in click-through rates, yet the WARC commentary at the time questioned whether optimizing for CTR alone constitutes sound marketing strategy in financial services. How should CMOs in regulated industries design the measurement architecture for AI-driven creative programs to ensure short-term performance metrics align with long-term brand and relationship objectives?

2. Coca-Cola's "Create Real Magic" campaign required a significant governance infrastructure — a senior Digital Council, a central AI team, a formal approval process, and a curated asset library — before any consumer-facing AI content was produced. To what extent does this governance model represent a scalable template for large CPG brands, and what modifications would be required for brands with less codified visual identities or smaller internal marketing organizations?

3. Unilever's Beauty AI Studio was built in partnership with an external technology group (Brandtech) and an in-housing specialist (Oliver), rather than through a traditional agency relationship or internal build. What are the strategic trade-offs between this partnership model and either full agency outsourcing or fully internalized AI creative capability, particularly given Unilever's stated goal of reducing its global agency roster?

4. Across these cases, AI consistently outperformed human-only creative on narrow performance metrics (CTR, VCR) while operating within human-defined strategic and brand parameters. Does this evidence support the hypothesis that AI will ultimately displace human creative strategy, or does it instead suggest a durable model in which AI and human creativity are structurally complementary? What conditions would need to change to alter your answer?

5. The three cases presented represent large, well-resourced global brands with the organizational capacity to invest in AI governance, staff training, and technology partnerships. How should mid-sized brands with significantly constrained marketing budgets and smaller creative teams think about the sequencing and prioritization of AI adoption in their creative workflows, and what evidence-based criteria should guide that sequencing?