AI's Unforeseen Value: Growth From Failures and Augmentation

This conversation from Marketing School reveals how seemingly simple AI implementations can unlock massive, unexpected business value, shifting the focus from mere cost-cutting to strategic growth. The core insight isn't about automating tasks, but about augmenting human capabilities and identifying entirely new revenue streams from the "failures" of automation. Companies that embrace this augmentation, rather than just reduction, stand to gain a significant competitive advantage. This episode is essential for leaders and strategists who want to move beyond basic AI adoption and understand how to leverage it for exponential growth, reskilling their workforce, and identifying market opportunities others miss. It offers a clear roadmap for transforming AI-driven efficiency into a powerful engine for expansion.

The Unforeseen $1 Billion Business: From Chatbot Failures to Design Dominance

The narrative surrounding AI often centers on efficiency and cost reduction. However, this discussion highlights a more profound implication: AI's ability to uncover entirely new business lines by analyzing its own shortcomings. IKEA's deployment of the "Billy" chatbot serves as a prime example. While it successfully handled 57% of customer service inquiries, the true innovation came from studying the remaining 43% that Billy couldn't resolve. These unresolved cases pointed directly to unmet customer demand for interior design services. Instead of simply tweaking the chatbot, IKEA leveraged this insight to launch a design consultancy, reskilling its existing customer service staff and integrating AI to power this new venture. This approach generated approximately 1 billion euros in its first year, demonstrating that "automation plus augmentation equals exponential growth." This strategy starkly contrasts with the common corporate impulse to simply cut staff when automation is introduced.

"I love that because a lot of companies are thinking about how to address sacking employees and saving expenses. The right methodology is how you can start thinking about how you can reuse those people, train them on new things, and actually create more new revenue streams for the business. That's a win-win situation."

This approach reframes the AI conversation from one of layoffs to one of strategic reskilling and revenue generation. The implication is that the "failures" or limitations of AI, when analyzed correctly, can be the most potent source of innovation. This requires a systems-level view, recognizing that customer interactions, even those deemed "unsuccessful" by an automated system, contain valuable data about unmet needs.

Creative as the Ceiling: Scaling Through Content Velocity

The conversation pivots to the critical role of creative content in scaling paid advertising, particularly in a world where AI can generate vast quantities. A company in the medical space scaled from $1.5 million to over $20 million per month within three months, driven by the AI-generated output of 4,500 creatives per month. This illustrates a powerful feedback loop: when the return on investment (ROI) for advertising is high, the ability to scale rapidly increases. The key insight here is that once a profitable advertising strategy is identified, the bottleneck often shifts from ad optimization to the sheer volume and variety of creative assets.

"You need more creative for your business because at a certain point, creative is the limiting factor. It's not even the ad optimization anymore; it's the creative."

This emphasizes that conventional wisdom, which often focuses heavily on campaign structures and bid management, misses the most impactful levers. The discussion identifies two primary drivers of competitive advantage in paid advertising: creative and first-party data. Creative encompasses not just the visuals but also the messaging and the entire user experience, from ad to landing page. The ability to rapidly test and iterate on these elements, powered by AI, allows businesses to find winning combinations that competitors struggle to match. The second lever, first-party data, allows for superior targeting by enabling platforms to find more individuals similar to a brand's most valuable existing customers. This creates a compounding advantage: more data leads to better targeting, which leads to higher ROI, which fuels more ad spend, generating even more data.

The Auctioning of Everything: API Pricing and In-LLM Commerce

Looking ahead, the discussion probes the future of pricing models and advertising within Large Language Models (LLMs). The analogy is drawn between current ad auctions and the potential future of API pricing. As more services expose APIs, and as agents and AI systems interact with them at high frequency, pricing could shift towards an auction-based model. This means that access to services might be determined by real-time bidding, especially when demand is high and rate limits are a concern.

The conversation then delves into how advertising will evolve within LLMs. While some predict more interactive or video-based ads, the more significant shift anticipated is towards keeping users within the LLM ecosystem for transactions, mirroring the strategy of platforms like Instagram and TikTok. The argument is that if a user is searching for the best bank accounts or mortgage rates within an LLM, and the LLM can facilitate the entire sign-up process, banks may prioritize in-platform conversions over driving traffic to their own websites. This is because the LLM can capture all necessary user data, and higher conversion rates on the platform can be more cost-effective.

"If the conversions are 15 to 20% higher because they're still on the platform, so it's cheaper for you, wouldn't you just want more conversions? Keep them on the platform versus them going to your website."

This trend, termed "agentic commerce," suggests a future where transactions are seamlessly integrated into AI interactions. Furthermore, the concept of dynamic, personalized ads that modify copy and video content on the fly, much like personalized ads on YouTube TV, is predicted. This level of hyper-personalization, driven by extensive user data, promises to significantly boost conversion rates by tailoring the message precisely to the individual's context and inferred needs.

The Enterprise LLM Race: Google and Microsoft's Long Game

The debate over who will win the LLM race, particularly in the enterprise sector, is framed by long-term market leadership. While Anthropic currently holds momentum and has surpassed OpenAI in revenue, the prediction is that Google and Microsoft will ultimately dominate the enterprise space within five to ten years. This is attributed to their existing enterprise relationships, vast data resources, and the strategic advantage of offering LLM capabilities as part of their existing, heavily adopted corporate suites.

The argument is that companies already paying substantial fees for Microsoft's or Google's productivity tools will be incentivized to adopt their integrated LLM solutions, especially if offered at a low or no additional cost, rather than paying separately for a third-party service like Claude. This strategy leverages existing customer bases and infrastructure to create a powerful competitive moat.

"And companies are going to be like, 'I could use Claude, which has been great, but this is costing me 5 million a year as a big company. Google's just offering this stuff for free. I'll take the free.'"

While acknowledging Anthropic's innovation and OpenAI's current mind share, the long-term enterprise battle is seen as a play for incumbents with deep pockets and established market penetration. Microsoft's strong enterprise contracts are also cited as a safeguard against potential disruptions from OpenAI's internal turbulence. For consumer use, the generational divide is noted, with younger generations gravitating towards newer, free, or more integrated LLM options over established players like ChatGPT, suggesting a dynamic and evolving landscape.

Key Action Items

• Immediate Action (0-3 Months):
  • Analyze "failures" or unresolved customer inquiries from existing automated systems (chatbots, support tools) to identify unmet needs and potential new service lines.
  • Conduct an audit of current creative production for paid ads, identifying bottlenecks and exploring AI-powered tools to increase content velocity.
  • Review first-party data strategy: assess current data collection, enrichment, and utilization for targeting in advertising platforms.
• Short-Term Investment (3-9 Months):
  • Pilot reskilling programs for employees in roles impacted by automation, focusing on skills that augment AI capabilities or manage new AI-driven services.
  • Develop a strategy for integrating LLM capabilities into customer service workflows, focusing on augmenting human agents rather than purely replacing them.
  • Begin testing in-platform conversion strategies within LLM interfaces or on platforms that encourage on-site transactions, rather than solely driving traffic to external websites.
• Longer-Term Investment (12-18 Months+):
  • Develop a comprehensive strategy for leveraging LLM-driven dynamic and personalized advertising, anticipating a shift away from static ad formats.
  • Evaluate strategic partnerships or integrations with major LLM providers (Google, Microsoft) to ensure enterprise-level AI capabilities are accessible and cost-effective.
  • Invest in building robust first-party data infrastructure to maximize the effectiveness of AI-driven targeting and personalization in an increasingly data-aware advertising ecosystem.