Today, we’re bringing you the latest in AI-powered marketing and business strategies. Here’s what’s inside:
🚨 AI Top Story: Discover how messy data can quietly wreck your AI campaigns.
🌟 AI Use Case Of The Week: Discover why digital doubles are becoming the new face of storytelling for brands like H&M, Nike, and Cadbury.
🎯 Killer Marketing Prompt: Get five tailored campaign ideas that use digital twins to go beyond influencer hype.
🎥 AI YouTube Resource Of The Week: See how Ray 3 handles motion, characters, and creative direction.
Stop Trusting “Close Enough” Data With AI
Why Bad Data And AI Can Be A Costly Combo
Most marketers are aware their data isn’t perfect. Clicks don’t always match conversions, platforms count sales differently, and customer journeys are often stitched together with more guesswork than reliability. In the past, that was frustrating but tolerable. If the overall numbers looked good, teams could shrug and move on.
That mindset doesn’t work anymore. When AI is making thousands of micro-decisions a second, a missing field or a mislabeled conversion isn’t just a small hiccup — it can send your campaigns down the wrong path at scale. A botched data feed might tell the system to favor one audience over another, or pour spend into channels that look like they’re working when they aren’t. The snowball effect is real, and the costs add up fast.
This is especially true if you’re using AI for media buying, automated bidding, dynamic creative optimisation, or customer segmentation. These tools only know what you give them. Feed them bad signals, and they’ll happily optimise toward the wrong audiences, starve high-value segments, or waste budget on impressions that never had a chance of converting.
So what does better data actually look like? It means standardising definitions across platforms so a conversion in Google Ads means the same thing as a conversion in your CRM. It means building checks that flag suspicious spikes before they steer optimisations. And it means having clear ownership; someone responsible for the integrity of the numbers feeding your AI tools, not just letting them run unchecked.
AI can absolutely sharpen targeting, personalise creative, and improve efficiency, but only if it’s working with information that reflects reality. If the foundation is messy, the outputs will be too, just faster and more convincingly wrong.
The smartest marketers in this new era will treat data hygiene not as a back-office chore, but as the core enabler of every AI-powered campaign. Get the plumbing right, and your tools can finally deliver insights you trust and results you actually want to scale.
👩🎨 Marketers warm to AI, but creative challenges and legal risks still loom - Brands are testing AI in creative, but worries over authenticity, copyright, and trust are holding them back.
🌍 Google Releases 10 AI Policy Gold Standards to Guide Emerging Economies in AI Adoption - Google lays out ten principles to help governments build skills, infrastructure, and guardrails for AI adoption worldwide.
🎥 YouTube adds Ask Studio insights, A/B testing, and easier collaboration in latest Studio updates - YouTube is rolling out new AI-powered analytics, title and thumbnail A/B testing, and tools to make creator teamwork easier.
🛒 Amazon debuts AI ‘creative partner’ to aid with campaign development - Amazon is beta-testing an AI assistant that helps advertisers plan campaigns, create assets, and streamline production.
🧠 New Research Reveals That "Evidence-Based Creativity" Is the Next Must-Have Skill Set for Marketers in the Age of AI - A new study shows marketers will need to blend creative ideas with hard data to succeed in an AI-driven world.
H&M, Nike and Cadbury: Digital Twins and AI Storytelling
H&M, Nike, and Cadbury are showing how AI can push creative boundaries while still keeping the human touch front and centre.
For its spring 2025 campaign, H&M photographed 30 models and then used AI to generate photo-realistic digital twins. Some ads even paired the real and AI versions side by side, with quotes from the models about their experiences. Importantly, H&M made sure all models kept their image rights and were compensated - a transparent move that answered ethical concerns while sparking conversation about what “real” means in advertising.
Nike took a different approach. For its 50th anniversary, the brand used AI to recreate Serena Williams on the court, staging a virtual match between a younger and older version of the tennis legend. By analysing her movements and playing style, Nike built an emotional story about growth and resilience. A powerful extension of its “Just Do It” ethos.
Cadbury, meanwhile, leaned into cultural relevance. The brand digitally recreated Bollywood star Shah Rukh Khan and used geotargeting to help small Indian businesses advertise during Diwali. The result was hyperlocal, socially impactful, and drove a 35% sales lift for Cadbury Celebrations.
How To Replicate Something Similar
You don’t need a global budget or celebrity partnerships to borrow from these playbooks. Start by asking: where could AI help you tell a human story at scale?
For heritage storytelling: Nike’s Serena project shows how AI can mine historical footage or archives to bring legacy stories to life. Marketers can apply the same idea to brand milestones, customer history, or even product evolution.
For local activation: Cadbury’s campaign shows the power of blending AI with cultural context. You could use voice cloning, facial mapping, or hyperlocal targeting to personalise creative for specific communities, regions, or events.
The key is balance. AI should enhance storytelling, not replace it. When brands stay authentic, transparent, and audience-first, AI becomes a powerful creative co-pilot that can scale emotion, relevance, and resonance across markets.
Leverage Digital Twins as the New Micro-Influencers
This prompt is built to help you turn the digital twin trend into concrete campaign ideas. Just swap in your brand, product, audience, and goal, and the LLM will generate tailored concepts with storylines, execution details, and business logic you can actually use.
You are an award-winning creative director tasked with designing campaign ideas for [brand/company], which sells [product/service] to [target audience]. The goal of this campaign is [insert specific marketing goal: e.g., awareness, launch, retention].
Instead of celebrity endorsements, the campaign should use AI-generated digital twins of everyday [audience group: e.g., teachers, baristas, students] to serve as authentic, relatable influencers.
Follow these steps:
Generate 5 unique campaign concepts that clearly explain:
The core creative idea
How the digital twins will be used (e.g., video ads, interactive social content, out-of-home, email storytelling)
A sample message or storyline that aligns with [brand voice/tone]
Why this idea would resonate with [target audience]
Ground your ideas in real-world marketing logic. For each concept, explain the potential business impact (reach, relatability, cost-effectiveness, or scalability compared to celebrity-led campaigns).
Make the concepts practical. Provide at least one execution detail for each idea (platform choice, ad format, or content style).
Write the response in a clear, organized format with headings for each campaign concept.How To Use It
Swap in your brand, product, audience, and campaign goal. The LLM will generate five detailed concepts that explain the story, execution channels, and business impact.
Next step: Once you have the campaign ideas, use creative AI tools (e.g., HeyGen, Synthesia) to actually produce the digital twins and bring these concepts to life in video, social, or interactive formats.
Pro Tip
After running the prompt, ask the LLM to also write a second prompt you can paste into your chosen creative tool. This way, you’ll bridge the gap between idea generation and actual production, making it faster to test and deploy your AI twin campaigns.
Is Ray 3 the Next Step in AI Video?
The video walks through LumaLabs’ Ray 3, which claims to be the first “reasoning” AI video model with native HDR, faster draft mode, and improved physics and consistency. The creator tests the reasoning loop that evaluates its own outputs and retries, plus visual annotations for directing motion, and notes that enabling reasoning requires a paid plan.
The UI is praised for iteration and controls, yet progress feedback is unclear, features are uneven across modes, and the final verdict is a B+, promising but not yet best-in-class.
Why talk it out when you can chart it out?
