CMOs Guide to AI in Marketing [+5 Case Studies][2026]
Artificial intelligence has moved from an experimental add-on to a core capability shaping how leading brands plan campaigns, write ad copy, and personalize customer experiences. Coca-Cola invited fans to co-create festive brand art using generative AI, Heinz confirmed its ketchup bottle as a true brand icon through an AI image generator, and JPMorgan Chase lifted click-through rates by 450% using AI-written ad copy. Nutella used an algorithm to generate 7 million unique jar labels that sold out within a month, while ATB Financial cut marketing campaign timelines by two weeks using Gemini AI brainstorming. As detailed in this DigitalDefynd guide, these real-world examples show that AI in marketing delivers measurable results when applied with clear intent rather than adopted for novelty alone. This guide walks CMOs through five proven case studies and five practical steps for bringing AI into marketing operations responsibly, from auditing workflows to measuring results before scaling any new AI use case.
Index
CMOs Guide to AI in Marketing: 5 Case Studies
- Coca-Cola: Generative AI campaign lets fans co-create festive brand art
- Heinz: AI image generator confirms ketchup bottle as brand icon
- JPMorgan Chase: AI-written ad copy lifts click-through rates by 450%
- Nutella: Algorithm generates 7 million unique jar labels that sell out
- ATB Financial: Gemini AI brainstorming cuts campaign timelines by two weeks
CMOs Guide to AI in Marketing: 5 Steps
- Audit current marketing workflows before selecting any AI tool or vendor
- Start with one high-impact pilot instead of a full-scale AI rollout
- Build a governance framework covering data privacy, bias, and brand voice
- Train marketing teams on prompt writing and AI-assisted campaign workflows
- Measure results against clear KPIs and scale only proven AI use cases
CMOs Guide to AI in Marketing: 5 Case Studies
1. Coca-Cola: Generative AI campaign lets fans co-create festive brand art
Challenge
Coca-Cola operates in more than 200 countries and relies on an instantly recognizable brand identity built over more than a century. By 2023, the company faced growing pressure to stay culturally relevant with younger, digital-first audiences who were quickly adopting generative AI tools for creativity and entertainment. Traditional advertising formats, while effective for decades, no longer generated the same level of engagement among Millennial and Gen Z consumers who wanted to participate in brand storytelling rather than simply consume it. Coca-Cola needed a way to modernize its visual identity, invite public participation, and demonstrate technological leadership without diluting its iconic imagery, all while producing content fast enough to match the pace of internet culture.
Solution
a. Generative AI creative platform: Coca-Cola partnered with OpenAI and Bain and Company to launch Create Real Magic, a dedicated microsite that gave digital creatives access to a library of branded visual assets, including its iconic bottle shape, script logo, and color palette, which they could remix using generative AI tools.
b. Community co-creation: Instead of relying solely on in-house agencies, Coca-Cola opened the campaign to creators worldwide, inviting them to reinterpret brand heritage in their own visual language. This shifted part of the storytelling process from agency-led production to fan-driven collaboration.
c. Public recognition and amplification: Selected artworks were featured on digital billboards in high-profile locations, including Times Square in New York and Piccadilly Circus in London, giving everyday creators large-scale exposure alongside the brand.
d. Recurring seasonal expansion: Building on the original 2023 launch, Coca-Cola extended the concept into a 2025 holiday campaign that let fans co-create festive AI-generated art, turning a one-time experiment into a repeatable, seasonal engagement format.
Result
The Create Real Magic platform attracted thousands of submitted artworks from digital creators in its initial run, positioning Coca-Cola as an early mover among global consumer brands experimenting with generative AI in marketing. The extended 2025 holiday edition generated more than 120,000 user-generated creations within a matter of weeks and drove a 60% rise in engagement across social platforms. Separately, AI-personalized content from Coca-Cola’s broader generative AI marketing efforts has been linked to a 20% increase in engagement on platforms such as Instagram and TikTok, along with a 15% higher conversion rate on AI-driven email campaigns compared with generic messaging. The campaign also accelerated content production timelines, with work that once took weeks reduced to days or hours, giving marketing teams greater flexibility to test creative concepts in real time.
Related: Role of CMO in Startups
2. Heinz: AI image generator confirms ketchup bottle as brand icon
Challenge
Heinz Ketchup had spent more than 150 years building one of the most recognized packaging designs in the food industry, but that same heritage created a growing risk of the brand appearing outdated to younger shoppers. By 2022, Heinz noticed its affinity scores among younger consumers had started to decline for the first time in years. Millennials and Gen Z audiences were flocking to newly launched text-to-image AI generators, using them to create imaginative visuals from written prompts and sharing the results widely online. Heinz needed a way to insert itself into this emerging cultural conversation, reconnect with a tech-savvy younger audience, and reinforce its position as the definitive ketchup brand, all without a large media budget to compete with bigger campaigns.
Solution
a. AI brand test: Heinz worked with agency Rethink to feed the prompt “ketchup” into DALL-E 2, an advanced AI image generator that was not yet available to the general public at the time, to see what the model would produce without any mention of the Heinz name.
b. Consistent brand recognition: Regardless of how unusual or specific the prompts became, including requests like “Ketchup Tarot Card” and “Renaissance Ketchup Bottle,” the AI consistently generated images featuring the Heinz keystone label shape and signature red color, reinforcing the idea that the brand and the product had become synonymous.
c. Fan-driven expansion: Heinz then invited social media followers to suggest their own ketchup-themed prompts, turning the experiment into the first campaign built entirely from AI-generated visuals co-authored by fans, which were then produced as special-edition bottle labels.
d. Real-world and digital amplification: The AI-generated artwork was displayed on out-of-home billboards worldwide, showcased in physical and metaverse art galleries in Toronto, and rolled out across digital and social channels to maximize organic reach.
Result
The A.I. Ketchup campaign generated more than 1.15 billion earned impressions worldwide, a return worth over 2,500% more than the brand’s media investment. Social engagement rates ran 38% higher than Heinz’s historical campaign benchmarks, and the concept drew coverage from major outlets including Fast Company, Bloomberg, TechCrunch, and Forbes. The campaign also won a Clio Gold Award for Product/Service in 2023 and an Integrated award from The Drum, and it prompted other brands, including Ducati and Sportsnet, to join in with their own AI ketchup mashups. Heinz proved that even a machine trained on internet-wide imagery associates the shape, color, and label of a ketchup bottle with one brand, reinforcing its market position at a fraction of the cost of a traditional global campaign.
3. JPMorgan Chase: AI-written ad copy lifts click-through rates by 450%
Challenge
As one of the largest banks in the United States, JPMorgan Chase produces enormous volumes of marketing copy across card, mortgage, and wealth management divisions, all competing for attention in crowded digital channels. Traditional copywriting relied on the subjective judgment and experience of individual marketers, a process that was difficult to scale consistently across millions of customer touchpoints. Chase needed a way to test and generate higher-performing messaging at scale, without simply guessing which words or phrases would resonate most with different customer segments, while still maintaining brand voice and regulatory compliance standards expected in financial services marketing.
Solution
a. AI copywriting pilot: In 2016, Chase began testing Persado’s Message Machine, an AI platform built on a database of more than one million tagged and scored words and phrases, to generate copy variations for its card and mortgage marketing campaigns.
b. Data-driven language selection: The AI analyzed which words, emotional tones, and phrasing patterns performed best with specific audience segments, then generated headlines and calls to action that a human marketer using subjective judgment likely would not have produced.
c. Head-to-head testing: Chase ran AI-generated copy directly against human-written copy across landing pages, direct mail, display ads, and social ads to measure real performance differences before wider rollout.
d. Enterprise-wide expansion: Following successful pilot results, Chase signed a five-year, enterprise-wide deal with Persado in 2019, extending AI-generated marketing creative across personal banking, home lending, wealth management, and digital advertising, and later exploring the technology for internal communications.
Result
Chase’s pilot showed that AI-generated ad copy achieved click-through rate lifts as high as 450%, compared with a 50% to 200% lift typically seen from human-written ads. In one direct comparison, an AI-written home equity ad reading “It’s true—You can unlock cash from the equity in your home” drew 47 applications per week, nearly double the 25 weekly applications generated by the human-written version. In another test, an AI-generated card offer produced nearly five times the unique clicks of the equivalent human-written promotion. Chase CMO Kristin Lemkau described the results as proof that “machine learning is the path to more humanity in marketing,” noting that the AI produced copy and headlines a human marketer likely would not have considered on their own. The results led Chase to expand AI-generated messaging well beyond its original pilot scope.
Related: Famous CMO Quotes
4. Nutella: Algorithm generates 7 million unique jar labels that sell out
Challenge
Nutella, the hazelnut spread owned by Ferrero, had built decades of brand recognition around a highly consistent jar design and instantly recognizable logo lettering. By 2017, the brand wanted to deepen its emotional connection with consumers by making each purchase feel personal and collectible, building on an earlier campaign that let customers print custom names on jars. However, producing that level of individual personalization at true mass-market scale, across millions of units sold in a single country, was far beyond what human designers could realistically create by hand. Ferrero needed a way to generate large volumes of visually distinct, brand-consistent packaging designs quickly enough to hit store shelves nationwide without compromising quality or brand identity.
Solution
a. Custom design algorithm: Ferrero partnered with advertising agency Ogilvy and Mather Italy to build a bespoke algorithm for the Nutella Unica campaign, capable of pulling from a database of dozens of background patterns and thousands of color combinations to generate packaging designs automatically.
b. Automated label generation: The algorithm combined shapes, dots, stripes, and other graphic elements in randomized configurations while preserving the brand’s recognizable lettering, ensuring every jar remained instantly identifiable as Nutella despite having a one-of-a-kind label.
c. Verified uniqueness: Each generated design was assigned its own individual identification code, allowing every jar to be authenticated as a genuine one-of-a-kind piece rather than a repeated pattern, adding a collectible quality to an everyday grocery item.
d. Supporting media campaign: Ferrero paired the packaging rollout with an online and television advertising campaign that framed each Nutella Unica jar as being as special and expressive as the individual customer buying it.
Result
The algorithm produced 7 million completely unique jar labels for the Italian market, an output that would have taken a team of 100 human designers working non-stop for roughly 150 years to replicate manually at a typical rate of 20 minutes per design. All 7 million jars sold out within one month of launch in Italian supermarkets. The campaign generated significant international media coverage and was later expanded to additional markets across Europe, the Middle East, and Africa. Nutella Unica demonstrated that an algorithm could take on a traditionally human creative role, mass-producing individualized packaging design at a speed and scale that made large-scale personalization commercially viable for a fast-moving consumer goods brand.
5. ATB Financial: Gemini AI brainstorming cuts campaign timelines by two weeks
Challenge
ATB Financial, a crown corporation and the largest Alberta-based financial institution with more than 62.3 billion dollars in assets under management, relied on marketing teams that often had to wait on input from subject matter experts before developing new campaign concepts. This dependency created bottlenecks in the campaign development process, slowing down the pace at which marketing teams could move from initial idea to finished creative work. In a highly competitive banking sector where marketing teams are expected to produce relevant, timely content across multiple channels, ATB needed a way to accelerate early-stage campaign ideation without compromising the quality or accuracy of the subject matter expertise built into its messaging.
Solution
a. Enterprise AI deployment: ATB Financial rolled out Gemini for Google Workspace to more than 5,000 team members, following a successful pilot program in which hundreds of employees used the AI to draft marketing materials, analyze data, and summarize content.
b. AI-assisted brainstorming: Instead of waiting for subject matter experts to weigh in before shaping a new campaign idea, marketing teams began brainstorming directly with Gemini, using it to generate and refine early concepts on their own timeline.
c. Workflow reorganization: The shift in how ideas were generated inspired ATB’s marketing teams to rethink how they collaborated internally, prompting broader experimentation with new ways of working both within and across teams.
d. Secure enterprise foundation: ATB emphasized that Gemini’s enterprise-grade security and data protections allowed teams to experiment with generative AI capabilities while maintaining the trust and data integrity expected in the banking industry.
Result
The shift to AI-assisted brainstorming reduced marketing project timelines by up to two weeks. Approximately 40% of ATB team members reported using Gemini on a daily basis, with individuals saving an average of nearly two hours per week on routine tasks. In the earlier pilot phase, 60% of participating employees said they could accomplish more work in the same amount of time or complete their tasks faster. ATB’s data science team noted that the results reinforced the importance of early, hands-on experimentation with generative AI to keep pace with the evolving demands of financial services marketing.
Related: Surprising Facts About CMOs
CMOs Guide to AI in Marketing: 5 Steps
1. Audit current marketing workflows before selecting any AI tool or vendor
CMOs who skip a workflow audit often adopt AI tools that duplicate existing spend, with one 2026 CMO Survey finding that technology adoption is outpacing organizational readiness across marketing teams.
Before evaluating any AI vendor, marketing leaders need a clear map of where time, budget, and headcount are actually going across campaign planning, content production, media buying, and reporting. Many marketing organizations discover during this process that a significant share of team hours go toward repetitive tasks, such as reformatting reports, manually compiling campaign data from multiple platforms, or rewriting briefs for different stakeholders, rather than strategic work. Without this baseline, CMOs risk purchasing generative AI tools that address problems the team does not actually have, while leaving genuine bottlenecks untouched. A useful audit typically involves interviewing each function within the marketing team, from brand and content to performance and analytics, to document specific pain points, average time spent per task, and where subject matter experts create delays for other teams.
ATB Financial, for example, discovered through this kind of internal review that its marketing team was frequently blocked waiting on subject matter experts before developing new campaign concepts, a specific bottleneck that later shaped its choice to deploy Gemini for early-stage brainstorming rather than for content generation alone. That precision, identifying the exact stage of the workflow causing delay, is what separates a useful AI deployment from a generic one, and CMOs should treat this audit as an ongoing exercise, revisiting it every two to three quarters as channels, team structures, and campaign volumes evolve.
2. Start with one high-impact pilot instead of a full-scale AI rollout
ATB Financial’s phased pilot approach, where hundreds of employees tested Gemini before a 5,000-person rollout, delivered average time savings of nearly two hours per week per employee.
Rather than deploying AI tools across an entire marketing organization at once, CMOs get more reliable data and lower risk by selecting one specific, measurable use case first, such as campaign brainstorming, ad copy testing, or performance reporting, and running it with a smaller group before expanding further. A narrow pilot makes it possible to isolate what is actually working, since a full rollout across every team and channel simultaneously makes it far harder to attribute results to any single tool or workflow change. JPMorgan Chase followed this exact model with Persado, testing AI-generated ad copy against human-written copy in a controlled pilot across card and mortgage marketing before signing a five-year, enterprise-wide deal once results, including a 450% lift in click-through rates, were confirmed.
Pilots also give marketing teams room to build internal confidence and skill with new tools before the pressure of company-wide deployment. In ATB Financial’s case, the pilot phase showed that 60% of participating employees could accomplish more work in the same amount of time, evidence that helped justify the subsequent expansion to more than 5,000 team members. CMOs should select a pilot use case with a clear, quantifiable outcome, such as time saved per week, click-through rate change, or campaign turnaround time, so results can be compared cleanly against a pre-AI baseline established during the workflow audit, before committing to a broader organizational rollout.
Related: Famous Female CMOs
3. Build a governance framework covering data privacy, bias, and brand voice
A 2026 study in the Journal of Business Research on agentic AI in branding, drawing on Coca-Cola, Netflix, and Unilever, found that brands with strong governance frameworks derive the most value from AI-driven personalization.
Before scaling any AI tool across marketing functions, CMOs need clear written guardrails covering what customer data the AI can access, how outputs are reviewed before publication, and how brand voice consistency is maintained across every AI-generated asset. Without this structure, marketing teams risk publishing AI-generated content that misrepresents the brand, mishandles sensitive customer data, or reflects biased outputs inherited from training data, all of which can damage consumer trust far more than any efficiency gain is worth. Governance should define specific approval checkpoints, such as mandatory human review before any AI-generated claim about pricing, health, or financial outcomes goes live, since these categories carry the highest regulatory and reputational risk.
Transparency with consumers matters as much as internal controls. The same 2026 research found that giving users meaningful agency over how their data is used functions as a trust builder that actually improves personalization system performance rather than limiting it, contradicting the assumption that stricter governance slows down AI-driven marketing results. CMOs should resist deploying every available personalization capability simply because the technology allows it, and instead define, in writing, which use cases align with brand values and regulatory obligations before those capabilities are activated across live campaigns.
4. Train marketing teams on prompt writing and AI-assisted campaign workflows
BBVA employees using Gemini in Google Workspace reported saving nearly three hours per week on average, a gain the bank attributes directly to structured training on how to prompt and apply the tool within daily workflows.
Deploying an AI tool without teaching marketing staff how to use it effectively often produces disappointing results, not because the technology underperforms, but because employees default to generic, poorly structured prompts that generate generic, unusable output. Training should go beyond a single onboarding session and instead cover practical, role-specific applications, such as how a copywriter might prompt AI to generate headline variations aligned with brand voice, or how a campaign analyst might use AI to summarize performance data across multiple channels. ATB Financial’s marketing team, for example, moved beyond using Gemini purely as an individual productivity tool once employees learned to brainstorm campaign concepts collaboratively with the AI, a shift that directly enabled the two-week reduction in project timelines.
Ongoing training also matters because AI models and their capabilities change frequently, and skills that worked well six months earlier may not reflect the tool’s current functionality. CMOs should build in refresher sessions, encourage employees to share effective prompts and workflows internally, and set an expectation that AI fluency is now a core marketing competency rather than an optional add-on. Marketing agencies training staff specifically on prompt construction and campaign-specific workflows consistently report stronger adoption rates and measurable time savings compared with organizations that provide access to AI tools without structured guidance on how to apply them.
5. Measure results against clear KPIs and scale only proven AI use cases
JPMorgan Chase measured its Persado pilot against a specific, quantifiable KPI, click-through rate lift, before expanding to a five-year, enterprise-wide deal, a discipline that prevented the bank from scaling an unproven tool based on anecdotal enthusiasm alone.
Every AI pilot needs a predefined set of metrics established before the tool launches, not after, so results can be compared cleanly against the pre-AI baseline captured during the initial workflow audit. Useful KPIs vary by use case, but commonly include time saved per employee per week, campaign turnaround time, click-through rate or conversion rate lift, and adoption rate among team members, all of which should be tracked consistently over the same measurement period used for the pilot. CMOs should avoid scaling any AI use case based on qualitative impressions or vendor-provided benchmarks alone, since internal results can vary significantly depending on team structure, existing workflows, and the specific marketing tasks involved.
Once a pilot demonstrates measurable, repeatable results, such as ATB Financial’s documented two-hour weekly time savings across 40% of daily Gemini users, CMOs have a defensible basis for expanding the tool to additional teams or use cases. Use cases that fail to show clear improvement against baseline metrics should be discontinued or reworked rather than scaled by default, since continuing to invest in an underperforming AI tool diverts both budget and marketing team attention away from use cases that have already proven their value.
Conclusion
The case studies covered in this DigitalDefynd guide, spanning Coca-Cola, Heinz, JPMorgan Chase, Nutella, and ATB Financial, demonstrate that AI in marketing succeeds when it targets a specific, well-defined problem rather than serving as a blanket solution across every function. Each brand identified a clear bottleneck or opportunity, tested AI against that specific need, and measured results before expanding further. The five-step guide for CMOs reinforces this same discipline: audit existing workflows, pilot before scaling, govern data and brand voice carefully, train teams thoroughly, and measure everything against clear KPIs. For marketing leaders considering AI adoption, these examples offer a practical starting point grounded in real outcomes rather than speculation. As AI capabilities continue to evolve, CMOs who follow a structured, evidence-based approach will be best positioned to capture genuine value from AI in marketing, rather than chasing every new tool that enters the market.
