Inside AI Marketing Warner Music Grey Goose IAB with Yaffe Martin Giegerich

Leaders from IAB, Warner Music, and Razorfish discuss real-world AI uses in marketing — from voice cloning for legacy artists and AI remixes to an AI-powered “Tailored Toast” chatbot and SEO optimizations for Google AI overviews.

Key takeaways: prioritize consent and IP clarity, start with low-risk experiments that excite leadership, and pair buzzy pilots with long-term foundations like search optimization and brand-safe filters.

0:05 — David Burkwitz:
Welcome to AI Insiders by AI Marketers Guild. I have a dream guest host today, Caroline Gilbert from IAB. She’s assembled an all-female panel to discuss AI in marketing. This session is interactive — please put questions in the chat and you may be invited on camera.

1:15 — Caroline Gilbert:
Thanks, David. I host and lead the IAB Center of Excellence on AI. We focus on best practices and standards for AI in advertising. Today I’ll give a quick intro and then hand it to the panel to introduce themselves.

2:19 — Caroline Gilbert:
We released an AI use-case map with over 200 AI use cases across the campaign lifecycle — audience insights, measurement, creative, etc. Alyssa will post a link. For example, synthetic audience testing uses synthetic audiences to test creative for hard-to-reach segments — useful if a target group won’t take surveys. There are also categories like assurance/compliance for bias and cultural sensitivity, and using AI to detect malvertising and cloaking. Publishers can use agents to detect IP violations. Now I’ll turn it over to the panel; Alicia and Kate, please introduce yourselves.

5:26 — Kate Martin:
I’m Kate Martin, social content engagement strategist at Razorfish; clients include HBO Max. Previously I led global digital marketing for Grey Goose at Bacardi. Today I’ll discuss Grey Goose projects like the Tailored Toast AI chat experience and SEO work for Google AI Overviews.

6:30 — Alicia Yaffy:
I’m Alicia Yaffy, SVP of Creative Marketing at Warner Music Global Catalog (Rhino Records US). My role focuses on expanding audiences for catalog artists and experimenting with new initiatives — a bit like Google X inside Warner. We work on tech-forward projects to grow streaming, physical sales, and artist brands.

7:31 — Alicia Yaffy:
A few projects we’ve done: AI remixes of hits in partnership with Endel, a voice double for Randy Travis so he could “sing” again, and a Notorious B.I.G. performance in Horizon World for his 50th anniversary. Consent was the guiding principle: did the artist or estate want this? For estates we look at the artist’s behaviors and legacy to decide if an AI approach is appropriate.

8:36 — Alicia Yaffy:
When cloning voices, we test multiple platforms (e.g., MyOx, 11Labs). For a current project we blended five model iterations built from recordings across an artist’s 20s and 30s plus a specific demo to match a target sound. It’s iterative and requires close work with management.

11:55 — Alicia Yaffy:
With the Endel remixes (e.g., Roberta Flack, War) the partner’s algorithm creates music to trigger specific brain states (focus, sleep). Those remixes are available on Endel’s app and on streaming platforms. There’s a lot of neuroscience research behind functional music and potential therapeutic applications.

14:04 — David Burkwitz (audience question):
If AI clones an artist’s voice, does the artist retain IP rights?

14:10 — Alicia Yaffy:
Yes — we must license any material not owned by Warner for training and usage. Where Warner owns the training materials (as with Endel work) we retained copyright to the output, and the partner functions like a remixer. These arrangements vary and the space is still legally unsettled.

15:06 — Alicia Yaffy:
Some experiments mix owned IP, licensed IP, and rights from artists; we aim to protect rights and be intentional. We’re an IP protection company fundamentally.

16:11 — David Burkwitz (audience question):
Are trained voices considered derivative works?

16:16 — Alicia Yaffy:
Currently, our remixes are treated as derivative works. Whether cloned voices themselves are categorized separately is an open question; we expect more clarity over the next year.

16:11 — Caroline Gilbert:
I also shared a link to a synthetic audiences platform; there are many options out there.

17:17 — Kate Martin:
At Grey Goose we started by auditing where AI could elevate marketing with light budget lifts: efficiency, optimization, and PR-buzz opportunities. We focused on low-cost, high-impact ideas like an AI chatbot (Tailored Toast) and SEO work for Google AI Overviews. Budget and ROI were constraints, so we prioritized quick wins to get buy-in.

18:15 — Kate Martin:
To get leadership buy-in, we positioned initiatives as first-to-market or performance improvements in the category, plus brand differentiation. Innovation internally and externally helped secure approval.

19:17 — Caroline Gilbert:
Alicia, with artists, management, and leadership involved, how did you get buy-in?

20:20 — Alicia Yaffy:
We start with the artist: explain the idea and tech, ensure alignment with their brand identity and consent. Many artists we work with are already curious about innovation. After artist buy-in, we present budgets and projections to leadership. Experiments also signal that Warner is innovative and protective of music across its lifecycle, which helps attract catalogs and artists.

22:21 — Caroline Gilbert:
For tech approval, we use a rigorous review: ensure platforms do not train on unlicensed material. Some platforms have company-wide deals; others go through internal review boards and approvals can take time.

25:34 — Kate Martin:
Security and compliance were key for Grey Goose: LDA compliance and 21+ guardrails. Our innovation and tech teams were closely involved. For smaller projects we moved fast; Tailored Toast took about four weeks total.

26:37 — Kate Martin:
Tailored Toast overview: an AI chat experience powered by ChatGPT with a Grey Goose brand filter. Users answer questions (occasion category — big milestone, small victory, social gathering; then specific occasion like wedding, promotion, adopting a pet; sentiment) and Tailored Toast generates brand-appropriate, witty toasts while blacklisting inappropriate content and discouraging overconsumption. We measured average time spent, number of completed toasts, and completion rates. It performed well organically and drove significantly longer engagement on-site than typical visits.

29:39 — Kate Martin:
Brand-safety work included testing and filtering in the backend to ensure the voice felt fun and premium. We ensured the experience aligned with LDA rules and the brand voice.

30:42 — Kate Martin:
On SEO and Google AI Overviews (AIO): AIO is a Google Search feature using generative AI to provide quick answers at the top of search results (launched May 2024). We had optimized Grey Goose for mobile voice search (Siri, Google Assistant) with FAQs and editorial content that matched top consumer queries. Because these optimizations were live ahead of AIO, our content was rapidly surfaced as sources for AI overviews. Key steps: audit site content, research top consumer queries and gaps, optimize content consistently, and pair organic SEO with paid search. KPIs are impressions and clicks; ranking in AIO can be long-tail.

32:48 — Kate Martin:
We used BrightEdge and agency support to identify queries and whitespace. We aimed to be a source for broader beverage queries (not only vodka) — e.g., what is bottle service, what is a martini — to expand industry reach.

33:51 — Kate Martin:
Zero-click search: users may read the AI overview and not click to a website, reducing referral traffic for publishers but offering brands new influence opportunities. Track impressions, clicks, and monitor shifts in organic referral traffic.

36:11 — Caroline Gilbert:
There are platforms (e.g., Profound, GumGum-style tools) that test how your brand shows up across different LLMs and give recommendations. That can help you optimize for each model’s weighting.

37:00 — Alicia Yaffy:
Measurement priorities for us: did we grow streams and sell more records? Engagement is crucial — we aim to convert discovery into long-term fans. Experiments can be about serving superfans or creating culture moments that prompt people to press play again. New technologies and fan engagement help expand monetization of sound recordings.

39:03 — David Burkwitz:
What’s the difference between AI pilots for PR/buzz and integrating AI into the long-term consumer journey?

39:18 — Kate Martin:
If budgets were unlimited we’d run both. In reality, brands often start with buzzy consumer-facing pilots (TikTok filters, PR) to excite leadership and secure funding, then invest in longer-term projects like search optimization or dynamic personalization that deliver sustained performance.

44:18 — Alicia Yaffy:
Agreed. Leadership tends to approve buzzy experiments, which also position us to attract catalogs and artists. From my perspective our duty is to protect artists’ legacies while keeping their music discoverable across evolving platforms and technologies.

45:19 — Kate Martin:
The long-term work (SEO, search, foundational optimizations) is less flashy but critical for performance and discovery.

46:23 — David Burkwitz:
Would you ever use synthetic influencers or virtual models (Vogue-style controversy)?

49:35 — Kate Martin:
Personally I’m hesitant. There needs to be clear transparency. Vogue’s lack of clarity caused backlash; transparency is critical.

50:35 — Alicia Yaffy:
Replacing human models is problematic. If a synthetic influencer is transparent about being synthetic and is used as a creator-driven entity (e.g., an established synthetic character with a known creator), that can be acceptable. We would not present synthetic creators as real people or use them to mislead.

51:41 — David Burkwitz (audience question, Jason):
How are brands handling consumer pushback against AI-generated content?

52:51 — Alicia Yaffy:
Pushback centers on misleading or replacement uses. We focus on using AI to expand access to art, not to replace artists or mislead audiences. Storytelling and transparency are key.

53:55 — Kate Martin:
AI should enhance, not replace, human creativity. Include a real element and be transparent. For Grey Goose, the brand bottle content is always protected as physical IP; backgrounds can be augmented case by case.

54:57 — Kate Martin:
Advice for marketers: know your leadership team. Start with an idea that excites them — ideally low-risk with clear KPIs. Pair buzzy pilots with foundational tactics to demonstrate both innovation and performance.

56:01 — Alicia Yaffy:
My advice: ensure AI serves your goals and values; integrate AI where it extends your brand rather than making AI the objective.

57:03 — Audience / panel wrap:
AI is a journey not a destination. Thank you to Caroline, Alicia, and Kate. We’ll share the IAB use-case map link and follow-up resources. Thank you everyone — have a great week.

(End of transcript)