The Limits of Best Practices in an AI-Driven World

Severin Sorensen
23 hours ago
6 min read

Last quarter, a leadership team gathered to review their go-to-market strategy. The deck was polished: benchmark data, industry best practices, case studies from high-performing competitors. Every recommendation had precedent, and every decision felt safe.

Two weeks later, a smaller competitor half their size launched a new model that undercut them on speed, pricing, and customer experience. No benchmark had predicted it, and no playbook had outlined it.

The gap had nothing to do with intelligence or resources. It came down to orientation: one company looked backward at what had worked, while the other built forward in real time.

That gap is widening, and AI is the primary reason.

Best Practices Are Becoming Obsolete

For decades, best practices represented a genuine competitive advantage; hard-earned institutional knowledge about what worked, what scaled, and what drove results. Today, they are a commodity.

AI has made that institutional knowledge universally accessible. Research indicates that the majority of businesses now use AI in at least one function, and the tools available can synthesize decades of strategic, operational, and marketing insight in seconds (McKinsey & Company, 2025). What once required years of accumulated experience can now be generated on demand.

But there is a deeper structural problem: the half-life of knowledge is shrinking.

In this environment, best practices do not simply lose value; they just become obsolete faster than they can be implemented. What worked last quarter may not work next quarter. What works for one company can be replicated by competitors overnight.

Takeaway: The move for leaders is to treat strategy less like a document and more like a hypothesis: something to be tested, refined, and updated as the market reveals itself.

The Illusion of Productivity

On paper, AI looks like a productivity revolution:

McKinsey Global Institute estimates that generative AI could deliver between $2.6 trillion and $4.4 trillion in annual economic value across 63 enterprise use cases (McKinsey Global Institute, 2023).
Controlled research studies have found that AI tools can increase individual throughput on realistic business tasks by an average of 66%, a figure that dwarfs the average annual labor productivity growth of the prior decade (Nielsen Norman Group, 2024).

But the reality inside most organizations is more complicated. A survey of more than 5,000 white-collar workers from companies with over 1,000 employees found that while executives reported being excited about AI, nearly 40% of front-line workers said the tools had saved them no time at all (Ellis, 2026).

This is the core trap of best practices in an AI-enabled world: most companies are layering AI onto existing workflows rather than redesigning those workflows from the ground up. The result is more tools, more output, and more noise, but not necessarily better decisions or stronger business outcomes.

Takeaway: Best practices optimize the old system. AI demands a new one.

The Real Divide: Builders vs. Followers

The divide is already visible and quantifiable. PwC's 2026 AI Performance Study found that just 20% of companies are capturing roughly 74% of all AI-driven economic returns. These organizations aren't winning because they have more AI tools; they're winning because they've pointed those tools at growth and reinvention, not just efficiency (PwC, 2026).

What are these companies doing differently? They are not applying AI to best practices. Rather, they are replacing best practices altogether. Specifically:

They redesign workflows instead of automating them.
They use AI to generate new strategic approaches, not to replicate old ones.
They treat every process as a living system subject to continuous refinement, not a fixed playbook to be executed.

The PwC research further found that leading organizations are nearly three times as likely to increase the volume of decisions made without human intervention (PwC, 2026).

Meanwhile, the majority of organizations remain in what McKinsey describes as "pilot purgatory,” experimenting broadly but capturing value in only isolated pockets, unable to translate AI adoption into enterprise-wide financial impact.

Takeaway: This is the new competitive divide: not AI versus no AI, but builders versus followers. For executives, the starting point is an honest audit of how your organization currently learns and how fast it can act on what it discovers.

From Best Practices to Intelligence Loops

If best practices are dying, what replaces them? A new operating model is emerging, one built on continuous learning and rapid iteration.

Most organizations still operate on a familiar linear model:

Best Practice → Implementation → Scale

Leading companies have replaced this with something fundamentally different:

Hypothesis → AI-Assisted Execution → Feedback → Iteration → Organizational Learning

Call it an intelligence loop. AI enables faster experimentation, near-instant feedback, and continuous optimization, but only if the organization is structured to use it that way.

Operationalizing the intelligence loop requires leaders to make specific structural choices:

Redefining decision rights so that action can happen closer to the data
Shortening feedback cycles so that iteration is measured in days, not quarters
Empowering teams to test hypotheses rather than simply execute against fixed plans

Takeaway: Static knowledge is a snapshot. Dynamic intelligence is a live feed. The leaders pulling ahead have stopped optimizing the snapshot and started investing in the feed.

What AI Actually Demands From Leaders

The most counterintuitive finding of the AI era may be this: the primary constraint is not technology. It is leadership. AI can generate insights, analyze data at scale, and surface strategic options. But it cannot decide what matters. It cannot frame the right problems. It cannot exercise judgment under uncertainty. That responsibility rests entirely with leaders, and many are not yet prepared for it.

Too many executives are still:

Requesting benchmark data and waiting for industry proof points
Looking for proven models before committing to action
Seeking certainty in an environment that no longer offers it

Certainty is precisely what AI disruption eliminates. The organizations that will win are those led by executives who:

Reason from first principles rather than precedent
Prioritize speed of learning over perfection of execution
Are genuinely comfortable operating without a complete playbook

Takeaway: The most important thing a leader can do right now is build a culture where learning, testing, and adapting are treated as core strategic competencies.

Companies Already Operating This Way

The following organizations illustrate what it looks like in practice to move from optimization to reinvention.

Klarna: Replacing Functions, Not Optimizing Them

In February 2024, Klarna announced that its AI-powered customer service assistant—built on OpenAI technology—had handled two-thirds of all customer service conversations within its first month of global deployment. The assistant handled 2.3 million interactions, performed the equivalent work of 700 full-time agents, matched human satisfaction scores, and reduced average resolution time from 11 minutes to under 2 minutes (Klarna, 2024).

Klarna did not apply AI to optimize its existing call center model. It rebuilt its customer service function around AI as the primary interface, with human agents handling escalations requiring judgment and empathy. The lesson is not that AI can replace customer service, it is that the model of customer service itself can be reinvented (Klarna, 2024).

Netflix: Continuous Experimentation at Scale

Netflix operates thousands of A/B tests simultaneously, using AI to drive personalization, content recommendations, and even thumbnail selection for individual users based on viewing behavior. They do not rely on industry best practices for what works in streaming. They build and operate a continuous intelligence loop that generates those insights daily.

Amazon: Decision Velocity Over Perfection

Amazon uses AI across logistics, pricing, demand forecasting, and inventory management. Their competitive advantage is not merely the volume of data they hold, it is the speed at which they act on it. Their widely documented internal principle of "disagree and commit" reflects an organization optimized for decision velocity over consensus-seeking. Speed replaces the need for certainty.

Goldman Sachs: Scaling Cognition, Not Just Labor

Goldman Sachs has deployed AI tools to assist bankers with document drafting, data analysis, and insight synthesis. Early results indicate material efficiency gains in high-value knowledge work. Critically, the goal is not to reduce headcount; it is to compress the time required for complex cognitive tasks, enabling senior professionals to spend more time on judgment-dependent work.

The Main Takeaway

The death of best practices is not a loss. For leaders willing to adapt, it is one of the most significant opportunities in a generation. For the first time, competitive advantage is not constrained by what has worked before. Organizations can test faster, learn faster, and adapt faster than at any prior point in business history, but only if they release the assumption that the right answer already exists somewhere in a benchmark report.

In the age of AI, advantage does not come from knowing more. It comes from learning faster than everyone else.

References

Klarna. (2024, February 27). Klarna AI assistant handles two-thirds of customer service chats in its first month [Press release]. https://www.klarna.com/international/press/klarna-ai-assistant-handles-two-thirds-of-customer-service-chats-in-its-first-month/
McKinsey & Company. (2025). The state of AI in 2025: Agents, innovation, and transformation. McKinsey Global Institute. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
McKinsey Global Institute. (2023, June). The economic potential of generative AI: The next productivity frontier. McKinsey & Company. https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier
Nielsen Norman Group. (2024, January 30). AI improves employee productivity by 66%. https://www.nngroup.com/articles/ai-tools-productivity-gains/
PwC. (2026, April 13). PwC 2026 AI performance study [Press release]. https://www.pwc.com/gx/en/news-room/press-releases/2026/pwc-2026-ai-performance-study.html
Ellis, Lindsay. “CEOs Say AI Is Making Work More Efficient. Employees Tell a Different Story.” The Wall Street Journal, 21 Jan. 2026, www.wsj.com/lifestyle/workplace/ceos-say-ai-is-making-work-more-efficient-employees-tell-a-different-story-6613ce9d.