- Start with outcome: What is the specific business result you're trying to achieve? (Reduce churn, increase basket size, improve brand recall among Gen Z, etc.)
- Map your competitive advantage: Where does your company genuinely win today? Brand, product, speed to market, distribution? Is the initiative you're working on connected to that?
- Ask the transformation question: If you were designing this workflow from scratch today, with no legacy systems and no existing process, how would AI change what's even possible, not just what's faster?
- Identify the 70%: Technology is usually 30% of the challenge. Where are your process gaps, change management needs, and literacy barriers? Who needs to be convinced, and how will you build the case?
- Check for executive cover: Is there a senior sponsor who genuinely wants this outcome, not just someone who said yes in a meeting? Without it, don't start.
Why 95% of AI Projects Fail and What Nike's Former VP of Marketing Data Did Differently
When Nike needed to solve the fairness problem with sneaker launches, where people camped overnight and paid others to queue, they didn't just automate the old process. They reimagined it entirely using machine learning, cutting demand forecasting error by 44% and proving that AI's real power lies in transformation, not just efficiency.
Linda Cereda is a former VP of Marketing Data at Nike, where she spent 16 years building a career that spanned retail, digital commerce, data partnerships, and AI transformation. She was the first GM of Nike's Sneakers app, led the company's first-ever connected marketplace data partnerships (linking Nike membership data with Dick's Sporting Goods and JD Sports), and built Nike's Consumer Marketing Academy for 500+ employees. She now advises brands and tech companies on AI strategy, drawing on hard-won lessons from enterprise-scale execution.
Her core conviction: 90% of AI projects fail to deliver value because companies confuse automation with transformation. Real competitive advantage comes not from doing existing work faster and cheaper, but from using AI to strengthen the things you're already best at.
This Week's Big Idea: AI Transformation vs. AI Automation
The winning formula isn't "let's use AI to cut costs." It's using AI to unlock outcomes that reinforce your actual competitive advantage.
Linda draws a sharp line between the two modes:
AI Automation takes your existing workflow and asks: How can I make this faster and cheaper? You apply AI tools to individual steps: research, brief writing, audience targeting, content creation, and the process gets more efficient. That's useful, but it's table stakes. "If everybody's running faster, nobody's winning," Linda says. Eventually, the whole industry catches up and automation becomes the baseline, not the differentiator.
AI Transformation starts from the outcome and works backward. Instead of asking how to streamline your seasonal campaign process, you ask: What is the business result I need, and what's the best possible AI-native workflow to achieve it? You're not putting AI on top of existing chaos. You're redesigning the process entirely around what AI makes possible.
The clearest example: Walmart's "Trend to Product" initiative used AI end-to-end, from social listening to trend identification, to product design sketches, to marketing language and launch, compressing a process that once took months into weeks. That's not efficiency. That's competitive reinvention.
Why it matters: Most companies are rushing to automate because it's visible and measurable in the short term. But Linda's insight is that short-term cost savings rarely create long-term moats. True competitive advantage comes from applying AI transformation to the area where you already win, whether that's product, brand, or customer experience.
Key Takeaways
1. Start with the business outcome, not the technology
Linda's framework for every AI initiative: What is the outcome I want? Then design the workflow around achieving it with AI, rather than fitting AI into an existing process.
At Nike, when the goal was "reduce churn," the question wasn't "how do I automate my email team?" It was: "What's the best possible decisioning system to keep customers engaged, and what data and models do I need to build it?" That reframe led to a multi-armed bandit / reinforcement learning approach that personalized outreach at the individual level, not the segment level.
2. Data as a strategic asset only works when tied to a real business problem
Linda's data partnership work linking Nike membership data with Dick's Sporting Goods wasn't launched because "data partnerships sound innovative." It was launched because Nike had a fundamental blind spot: millions of customers who bought Nike at wholesale partners were invisible to them. Once identified, non-member customers were worth 50% less than known members, because Nike couldn't personalize.
The partnerships solved a specific problem. That's what made them strategic, not the data itself.
3. The hardest part of AI is never the model. It's the people.
The Sneakers app ML model was technically impressive. But Linda's biggest challenge wasn't building it. It was convincing a brand manager in Japan to trust a black-box machine over her own instinct about Gen Z women in Tokyo.
Her solution: invest heavily in literacy. Make learning relatable, not technical (she explains multi-touch attribution using a soccer goal analogy: defense to midfielder to striker). Use gamified micro-learning so people practice instead of passively watching videos. Tie data directly into decision-making processes so it becomes habit, not theory.
Without executive sponsorship and ground-level literacy, even the best AI initiative will stall.
4. AI decisioning is the future of personalization, but it requires a mindset shift
Traditional A/B testing for journey optimization is painfully slow. Testing one variable at a time, waiting for statistical significance, then moving to the next, optimizing a full journey can take over a year.
AI decisioning (reinforcement learning / multi-armed bandit) lets a machine trial-and-error its way to the optimal combination of content, timing, channel, and offer for each individual user, not for the 80% majority. Linda saw this work best at mid-size companies with enough data density but without the organizational complexity of large enterprise.
The catch: it requires probabilistic thinking, sufficient data volume, and it fundamentally conflicts with how most marketing teams are organized around seasonal, functional calendars. The org change is harder than the tech change.
5. The competitive edge in AI belongs to mid-size companies right now
Large enterprises have data density and budgets, but they're slow, political, and often can't experiment freely because procurement processes block early-stage vendors. Startups move fast but lack the data and resources for meaningful AI.
Mid-size companies are the sweet spot: enough data to train models, enough organizational agility to redesign workflows, and enough pressure on resources to actually need AI to multiply output. Ironically, "having too much money can be a deterrent for change."
Try This Today
Run the AI Transformation Audit
Take 20 minutes and run this exercise for one key marketing initiative:
If AI is only making your existing process cheaper, you're automating. If AI is enabling an outcome that literally wasn't achievable before, that's transformation.
Expert Spotlight
Linda Cereda is a former VP of Marketing Data at Nike, where she spent 16 years leading cross-functional initiatives across retail, digital commerce, data partnerships, and AI transformation. As the first GM of the Nike Sneakers app, she pioneered machine learning for demand forecasting and fairness scoring. She built Nike's first connected marketplace data partnerships and led the company's Consumer Marketing Academy for 500+ employees. She now advises brands and tech companies on AI strategy, bringing enterprise-scale execution experience to organizations at every stage.
Connect with Linda on LinkedIn
Next Week
We're diving into how to move beyond generic AI prompts and build content marketing strategies that actually convert. Andy Crestodina, Co-Founder and Chief Marketing Officer of Orbit Media Studios, shares his expertise on leveraging AI for content marketing, SEO, and conversion optimization. Andy will walk through how to build effective AI workflows, harness persona-driven strategies, and use AI to gain a real competitive edge in digital marketing, not just produce more content faster.
Stay tuned for "Beyond Generic AI: The Workflows Turning Content Into Competitive Advantage" with Andy Crestodina.
Found this useful? Forward it to your marketing team.
Listen to the full episode:
Want exclusive access? Join our Discord community to watch these podcasts live as they happen, ask questions directly to our guests, and connect with other AI marketing pioneers who are navigating this shift in real time: https://discord.gg/7eYazKy6nY
Ready to build AI-native workflows? Enrich Labs helps leading companies create real-time listening systems that break the executive echo chamber and inform strategic decisions with fresh market signals. See how teams like Nike's are using AI-powered intelligence to compound learning across their organizations. Learn more at enrichlabs.ai
