Blog

Steve Ells invented the burrito bowl, and now he thinks people might be suffering from “slop bowl” fatigue. In an interview with the Wall Street Journal, the founder and former CEO of Chipotle talks about his new concept, Counter Service, a sandwich shop that launched last year in New York City and plans to expand to Charlotte and Dallas next year.

Counter Service specializes in slow-roasted meats, homemade bread and made-in-house sauces and relishes. The concept pivots away from an earlier post-Chipotle venture, a vegan restaurant built around robotics. It uses technology that flags incorrectly assembled orders, part of a broader system predicting sales using traffic, transit and weather data.

Ells says the appeal is in the details that most chains skip. “The pieces of meat are not perfectly shaped like the things you see in these chain sandwich places,” he said. “We’re making all of our sauces and relishes and garnishes. It’s difficult to do these things.” After 27 years eating Chipotle nearly every day, he says he now eats at Counter Service just as often.

Steve Ells invented the burrito bowl, and now he thinks people might be suffering from “slop bowl” fatigue. In an interview with the Wall Street Journal, the founder and former CEO of Chipotle talks about his new concept, Counter Service, a sandwich shop that launched last year in New York City and plans to expand to Charlotte and Dallas next year.

Counter Service specializes in slow-roasted meats, homemade bread and made-in-house sauces and relishes. The concept pivots away from an earlier post-Chipotle venture, a vegan restaurant built around robotics. It uses technology that flags incorrectly assembled orders, part of a broader system predicting sales using traffic, transit and weather data.

Ells says the appeal is in the details that most chains skip. “The pieces of meat are not perfectly shaped like the things you see in these chain sandwich places,” he said. “We’re making all of our sauces and relishes and garnishes. It’s difficult to do these things.” After 27 years eating Chipotle nearly every day, he says he now eats at Counter Service just as often.

Most outreach asks for value and offers none in return. The businesses that earn links and citations publish proof first.

Every Monday morning, I open my inbox and find the same email waiting 30 or 40 times. The subject line says “quick question” or “collaboration opportunity.” The body asks me to add a link to an article I wrote three years ago, points to a homepage and offers nothing in return.

I run a digital PR agency called ESBO and publish a handful of industry sites, so I sit on both sides of this trade. I send outreach for a living, and I delete it for a living. The deleting takes less time every year, because the emails all share one flaw: they ask, and they bring nothing an editor can use.

None of this is personal. Muck Rack’s State of Journalism 2026 survey found that 88% of journalists delete pitches that miss their beat, and about half say they seldom or never respond to pitches at all. What interests me is the small group of requests that get a yes, because the reason is usually unglamorous and repeatable: they arrive carrying proof.

Editors link to evidence, not to companies

When I cite a source in an article, one question decides it. Does this page back the claim I am making? A homepage backs nothing, and neither does a services page with a paragraph about passion for excellence.

What backs a claim is a number attached to a method. In the same Muck Rack survey, 40% of journalists named original data or research among the elements they value most in a pitch. That matches what I see in my own inbox so closely it barely counts as a finding.

Imagine two emails from the same moving company. One says it is a trusted leader in relocation services. The other says it compared 2,400 quotes against final invoices this year and found that the typical move cost 23% more than the estimate. The first email is marketing. The second is a source, and I would cite it by Friday.

The data you already have is the asset

You do not need a research department for this. Any business that has operated for a few years is sitting on records nobody else can see: quotes, invoices, support tickets, return reasons, delivery times and refund rates. That pile of admin is the raw material.

An accounting firm could publish how many of its clients file in the final week and what the panic costs them. An online store could break down its return reasons by product category. A recruiter could report how long candidates in one niche stay in their first job. The work is not complicated. It comes down to pulling six to 12 months of records, removing anything that identifies a client and writing up what you found.

Honesty about scale matters here. If your dataset is 300 projects, say it is based on 300 projects. Small and real beats big and vague, and editors can tell the difference in seconds.

One good number outworks a hundred cold emails

There is a bigger reason to do this in 2026, and it is the way answers now get assembled. A Pew Research Center analysis of real browsing behavior found that when an AI summary appears on a search page, people click a traditional result in 8% of visits, down from 15% without one. Clicks on the sources cited inside those summaries happen about 1% of the time.

You could read that as a reason to give up on visibility. I read it the opposite way. If answers are assembled from a handful of sources and almost nobody clicks through, the only durable position is being one of those sources, because your name travels with the number even when the visit never happens.

I have watched this on my own sites. Articles built around one original figure keep getting picked up in roundups, newsletters and AI answers years after publication, while pure opinion pieces fade within months. Something else happens too, and it still amuses me: once you become the source, the emails reverse direction. People start writing to you, asking to be included in the next update.

The easiest number to copy wins the citation

Packaging decides whether any of this gets used. Put the main finding in the title, then repeat it in the first hundred words as one plain sentence someone can lift. Explain the method at the bottom. Refresh the numbers once a year so the citation stays current and earns a second round of pickups.

And do not gate it. I have lost count of the promising reports I abandoned because the figure I needed sat behind a lead capture form (abandoned is generous; I closed the tab in five seconds). An editor on deadline will not fill in a form, and neither will a language model.

The hit rate is nothing to brag about. Most of these assets get ignored, and in my experience roughly one in three earns real pickup. The one that lands pays for the other two many times over, which is still a better return than any cold sequence I have ever run.

I still send outreach every week, so this is not a sermon against asking. Asking just stopped working on its own. If you want links, mentions and a place inside AI answers, give people something that survives the delete key. Publish one number this quarter that nobody else has, then watch what starts showing up in your own inbox.

Most outreach asks for value and offers none in return. The businesses that earn links and citations publish proof first.

Every Monday morning, I open my inbox and find the same email waiting 30 or 40 times. The subject line says “quick question” or “collaboration opportunity.” The body asks me to add a link to an article I wrote three years ago, points to a homepage and offers nothing in return.

I run a digital PR agency called ESBO and publish a handful of industry sites, so I sit on both sides of this trade. I send outreach for a living, and I delete it for a living. The deleting takes less time every year, because the emails all share one flaw: they ask, and they bring nothing an editor can use.

When we started building what became the Amazon Web Services (AWS) Database products, plenty of smart people believed enterprises would never trust the cloud with their databases. Today, that market generates more than $107 billion in annual revenue.

The experience taught me something I have carried into every leadership role since. If you wait for consensus, someone else will build the future while you are still debating it. That does not mean making reckless bets. It means learning how to recognize evidence before everyone else does and having the discipline to act on it. Here are four lessons that changed the way I think about innovation and leadership.

Look for constraints, not trends

Most companies spend their time chasing the next technology trend. I have found it is far more valuable to look for the problems customers have quietly accepted as part of doing business.

When we were building AWS database services, we realized enterprises were investing enormous amounts of time, money and engineering talent managing database infrastructure. They were provisioning servers, maintaining licenses, planning for failover and keeping systems running. None of that work created a competitive advantage. It was simply the cost of keeping the lights on.

That observation became the opportunity. If we could remove that burden, engineering teams could spend more time building products instead of maintaining infrastructure. That is why you need to know where your customers are investing time and resources without creating meaningful value. Those constraints often reveal the most important opportunities for development.

Build conviction with evidence

One of the biggest misconceptions about innovation is that successful leaders have extraordinary confidence. My experience has been the opposite. The best decisions come from gathering evidence, even when the market hasn’t caught up yet.

There was tremendous skepticism about whether enterprises would trust a cloud provider with something as important as their databases. We didn’t ignore those concerns. We studied early customer behavior, watched how managed services changed engineering productivity and kept testing our assumptions against real-world results. Every customer who succeeded strengthened our conviction because the evidence kept pointing in the same direction.

That is why I believe evidence is what separates conviction from stubbornness. Stubbornness ignores facts that challenge an idea. Conviction becomes stronger because it keeps collecting evidence before making bigger commitments. Before making your next major investment, spend less time gathering opinions and more time studying early adopters. The people already experimenting with the future will teach you far more than the people predicting it.

Sell the problem before the solution

Many leaders struggle to gain support because they start by pitching a bold vision.

We learned that internal alignment became much easier when we started with the problem instead of the solution. Before asking anyone to believe in managed database services, we made sure they agreed that the existing approach was becoming unsustainable. Once people acknowledged the problem, conversations about a different future became much more productive.

This approach works inside every organization. Before presenting your next proposal, ask whether everyone agrees the current state deserves to change. If the answer is no, spend your energy building alignment around the problem first. Then present your solution as the logical next step.

Move faster than consensus

One lesson stands out more than any other. Waiting for certainty feels responsible, but it often becomes the biggest competitive risk.

When people look at AWS today, they often assume our biggest advantage came from being early. It didn’t. Our biggest advantage came from learning earlier.

Every customer who adopted AWS database services gave us feedback we could never have created inside the company. We saw real workloads, real failure modes and edge cases that no amount of internal testing would have uncovered. Every lesson made the product stronger and helped us make better decisions.

By the time the market broadly accepted managed database services, we had years of production experience behind us. That institutional knowledge wasn’t sitting in a document. It was built into our architecture, our operational playbooks and the instincts of the engineering team. Competitors were trying to catch up with years of accumulated learning.

That is why I tell leaders that speed to market is really speed to learning. Revenue comes later. Learning starts on day one, and every customer interaction expands your advantage.

A practical way to build that advantage is to separate reversible decisions from irreversible ones. If a decision can be changed later, make it quickly and learn from the outcome. Reserve longer discussions for the decisions that truly reshape the business. The faster you learn, the harder you become to catch.

Three ways to build the next market

If you are trying to build something people don’t fully believe in yet, start here. Stress-test your conviction. Look for evidence from early adopters instead of relying on market opinions. Study what almost caused them to give up, because that is where the strongest insights often appear. Build agreement around the problem. Before presenting a solution, make sure key stakeholders agree the current approach is no longer good enough. Alignment becomes much easier when everyone starts from the same reality. Compress your decision cycles. Move quickly on decisions you can reverse and protect time for the ones you cannot. Speed is less about making perfect decisions than creating more opportunities to learn.

Markets are built by leaders who recognize meaningful problems early and keep moving while everyone else waits for certainty. That is the lesson I learned helping build a business few people believed could exist — and it is one every leader can apply, regardless of industry.

When we started building what became the Amazon Web Services (AWS) Database products, plenty of smart people believed enterprises would never trust the cloud with their databases. Today, that market generates more than $107 billion in annual revenue.

The experience taught me something I have carried into every leadership role since. If you wait for consensus, someone else will build the future while you are still debating it. That does not mean making reckless bets. It means learning how to recognize evidence before everyone else does and having the discipline to act on it. Here are four lessons that changed the way I think about innovation and leadership.

Look for constraints, not trends

Most companies spend their time chasing the next technology trend. I have found it is far more valuable to look for the problems customers have quietly accepted as part of doing business.

Jefferson Simmons was 20 years old and about to be homeless. His entire fraternity house was getting renovated, and every rental in town wanted nothing to do with a group of college guys. 

On a whim, he flipped a Zillow toggle from rent to buy and found a mismarketed three-bedroom house that was actually a 2,700-square-foot property with three extra rooms in the basement. He pitched his parents to co-sign, negotiated the seller down seven rounds to $178,000, and moved his fraternity brothers into the basement. 

Nine years later, he’s walked away from law school, built partnerships with an uncle and a private investor, and grown that first accidental deal into 17 properties and 39 doors. 

Here’s how he built it.

You were a sophomore in college, with no income and no credit. How did you actually get that first house?

I’d saved money since high school from selling firewood and doing livestock projects, and I got a full academic scholarship right before graduation, so I had a nest egg but no income a bank would lend against. 

I went home and pitched my parents using an Excel spreadsheet and a full 10-year pro forma showing rent increases, and they agreed to co-sign. I negotiated the seller down from their asking price to $178,000 over seven rounds of back-and-forth, partly because I knew from the listing agent that the family was highly motivated to sell, and partly because I genuinely had no more room to go higher. 

My mortgage payment has stayed the same the whole time, about $1,300 a month, including taxes and insurance. I rented it the first year for $1,600, and it’s currently leased through 2027 at $3,100 per month.

Your second deal was a foreclosure auction property you bought with your uncle. How did that partnership actually work?

I saw a duplex next door to my first house heading to a bank foreclosure auction, and I had zero money to buy it myself. My uncle, who’d built a portfolio of his own and was a big mentor to me, agreed to fund it as a money partner. 

We could only look through the windows before the auction since we couldn’t access the interior, so we did our underwriting from the driveway over coffee, and he told me we could afford up to $140,000 after repairs. Then he left the country on a trip and told me he’d be completely unreachable, so I was the one bidding live from my laptop. 

I got it to $100,000, and even though it didn’t technically meet the bank’s reserve, they wanted it off their books and took the offer anyway.

You walked away from law school after one semester to go all-in on real estate. What made you pull the trigger?

I sat in my first law school class, and they described the bell curve of graduates, meaning that where you rank determines your salary. I realized I wasn’t going to be at the top of that curve, and I’d be leaving school with over $100,000 in student loan debt for the privilege. 

I’d already closed two real estate deals by that point and had real proof of concept, so I decided I’d rather take on another mortgage that pays me back than debt that doesn’t. I left after one semester, worked as an insurance underwriter making $42,000 a year, got my real estate license on the side, and kept buying single-family homes for years while working two jobs.

You’ve done some creative financing since then, including turning a house sale into a line of credit. Walk us through that deal.

I was working as an agent for a cash-buyer client during an insane seller’s market where every listing was already pending within hours. He was getting frustrated that we couldn’t move fast enough on anything. 

Around the same time, tenants in a house I owned asked to break their lease early to buy their forever home, and I let them out of it. That left me with a vacant house I knew fit exactly what my client wanted. 

Over dinner, I gave him two options: I’d sell it to him for $25,000 more than I paid, or I’d sell it to him at my exact cost if he’d write me a $200,000 private line of credit instead. He laughed, looked at the house with his wife over FaceTime, and agreed to the line of credit. 

Three months later, I used it to buy a $171,000 house, and he wired the full balance the day of closing with no appraisal and no bank fees. I pay him 7.25% interest, which beats his T-bill returns and costs me less than a bank would. We’ve since done several more deals together and become genuine friends.

What does your portfolio look like today, and what’s actually driving your growth now?

I’m at 17 properties, 39 doors total, and I own all of them outright except for a minority stake in a 15-unit I hold with a few partners. Altogether, that’s about $20,000 a month in cash flow. 

A big unlock along the way was sweat equity: I helped my uncle renovate a 12-unit he bought in 2019, doing new kitchens, floors, and paint myself, in exchange for a 10% stake, which let me build equity without putting up much of my own cash. I also stopped thinking I could only buy one house a year by saving for the next down payment, since that mindset was actually limiting how fast I could scale. 

Between the family partnership, the private line of credit, and just getting comfortable asking people directly for capital, that’s what let me go from one deal a year to where I am now.

Nvidia CEO Jensen Huang is on top of the world. 

Huang is the leader of the world’s most valuable company, which boasts a market cap of $5 trillion. Nvidia has directly benefited from manufacturing the AI chips at the heart of the AI boom. Meanwhile, Huang has seen his personal fortune swell to $173 billion. He is the seventh-richest person in the Bloomberg Billionaires Index global ranking. 

Now Huang says that AI naysayers need to find new talking points. 

“Listen, if you want to warn the world about the incredible capabilities of this technology, I think it’s been achieved,” Huang said in a recent interview with Axios cofounder Mike Allen. 

He added that some of his peers are scaring the public by making wide-ranging predictions about AI’s impact on jobs. 

“We just all have to be thoughtful and careful,” Huang said. “We’re scaring people.”

Huang thinks leaders should be honest about AI

Huang advocated for balance when it comes to talking about AI. While it is natural to focus on safely deploying AI on a broad scale, the industry also has to explain the benefits of the technology, he said. 

“I think that we ought to be much more enthusiastic about it, help the United States realize that the only way we get left behind is if we don’t apply the technology,” Huang said.

He argued that AI leaders need to be honest with people. They have to start by noting that sweeping job losses have not shown up so far. He added that some of the dominant storylines around AI are more distracting than truthful. 

“Don’t create a story that is made up,” he said. “It is made up that there’s going to be a singularity. It’s made up that somehow we’re living in a simulation. These are all made-ups. I think it’s a fun story. I don’t mind listening to it. And I even enjoy the narrative when it’s told by many of those leaders who are my friends.”

One voice stands out

Huang didn’t call out specific people or companies, even though he clearly thinks some of the stories about AI are hurting public perception. 

Anthropic CEO Dario Amodei is one of the most prominent cautionary voices in the AI world. He warned that the technology could lead to major job losses. Amodei said last year that AI could eliminate as many as half of entry-level, white-collar jobs, taking over tasks like coding, writing and data work within the next one to five years. 

Anthropic is also unique among leading AI labs in that it regularly releases its own research focused on whether its Claude AI system shows early signs of human‑style thinking or feeling, rather than just following instructions and patterns. Even though Huang wants the industry to stick to straightforward, grounded stories about AI, Anthropic leans into more dramatic questions about if AI could ever start to think or feel more like a person.

Nvidia CEO Jensen Huang is on top of the world. 

Huang is the leader of the world’s most valuable company, which boasts a market cap of $5 trillion. Nvidia has directly benefited from manufacturing the AI chips at the heart of the AI boom. Meanwhile, Huang has seen his personal fortune swell to $173 billion. He is the seventh-richest person in the Bloomberg Billionaires Index global ranking. 

Now Huang says that AI naysayers need to find new talking points. 

Skip to content

What’s New?

Updated 38 seconds ago

Latest

Inside our company, I have built what I think of as an AI executive copilot. It is integrated into how I manage the business every day, not something I check in on separately or run alongside my normal work. It is part of my normal work.

It connects to our CRM data, internal documentation and operational workflows. It monitors initiatives across divisions, tracks KPIs, flags risks and surfaces opportunities I might otherwise miss. When I need to push seven sales initiatives forward simultaneously, I prompt it to draft the relevant communications for the right people and move them to review. When I need visibility into treasury, finance and operations at the same time, I have it. I am not stopping to sequence through conversations one at a time. I am directing multiple things simultaneously in a way that was not possible before.

What surprised me is not what the system can do. It is what it demands from me as a leader.

The constraint is never the technology

There is a persistent misconception that deploying AI at an operational level is a technical challenge. It is not. The constraint is how well you understand your own business. If you do not understand your product, you cannot prompt correctly. If you do not understand your customer, you cannot prompt correctly. If you do not understand your workflows, you cannot prompt correctly. And if you cannot prompt correctly, the system cannot help you.

This is why I believe a strong operator with deep business knowledge is more valuable in this environment than a technical expert. You have to be able to educate the system on what your organization does, how it does it and who it serves. Then you have to be able to prompt toward where you want to go, without telling the system how to get there. That second part is harder than it sounds. Most leaders are trained to prescribe solutions. Here, you have to educate the system by feeding it information and prompting it, and then you have get out of the way.

What used to be programming is now prompting. What used to be execution is now direction. That shift applies to every leadership role in the organization, including mine.

What this actually required me to know

To make this work, I had to get much closer to how the business actually operates. Not how I assumed it operates. How it actually operates. That means understanding the full lifecycle from intake through underwriting, case management, risk management and account resolution. It means knowing which KPIs matter and why, which ones I might be missing and what drives enterprise value beyond internal performance.

It also means understanding the people who use our platform. A paralegal. An attorney. A healthcare provider. A revenue cycle manager. Where are the friction points in their experience? Where are we adding work instead of removing it? What does our sales cycle actually look like at each stage, and where is activity breaking down relative to close rates?

AI does not solve those questions for you. It amplifies how clearly you have already answered them. When your understanding is shallow, you scale mistakes. When it is deep, the system can analyze across more variables than any individual could and surface opportunities you would not have identified on your own. It may find a coding issue, a sales activity gap, an onboarding friction point or a margin opportunity. Depth of input determines depth of insight.

How I actually use it day to day

I still have seven direct reports. I still hold a standing Zoom with my leadership team every morning. What has changed is everything that happens around those touchpoints. I go into the system and prompt initiatives for each leader. I track where things are stalling. I surface anomalies in performance data before they show up in a meeting. I move from managing tasks to directing outcomes, and the system handles the coordination work in between.

The result is that I can advance multiple priorities simultaneously in a way that was not possible when every piece required me to move sequentially. I used to have to stop, connect with one division leader, suggest changes, move to the next one, suggest changes there and then try to get the two of them talking to each other. Now I can send it all out at once. That is not a minor efficiency gain. It fundamentally changes how fast the business can move.

To be precise about what this is: I am prompting the system and it is producing what I ask for. It is not acting independently. But it is allowing me to initiate, track and drive more than I could manage on my own.

The visibility it creates is the real advantage

Every organization has blind spots. Initiatives that go stagnant. Decisions that get delayed because no one has the full picture. Opportunities that never surface because attention is fragmented.

A system like this does not have those constraints. It tracks how long things actually take, not how long we think they take. It surfaces patterns that would otherwise take weeks of analysis to see. It gives me a level of continuous visibility across the business that is genuinely difficult to maintain manually, and that visibility changes how I lead.

My prediction is that this will not stay optional for long. Boards will build these systems. Investors will use them to evaluate performance. External stakeholders will have their own view into what is actually happening inside companies. Leaders who have already built this capability internally will be ahead of that curve. Those who have not will find themselves explaining gaps they did not know existed.

The real constraint is judgment

The question is no longer whether something can be built. It is whether it should be built, how it should work and what it should be directed toward. Those are leadership questions, not technical ones.

The leaders who thrive in this environment will not be the ones who understand the tools best. They will be the ones who understand their business, their customers and their operations clearly enough to direct those tools toward outcomes that matter. That clarity is the competitive advantage.

Inside our company, I have built what I think of as an AI executive copilot. It is integrated into how I manage the business every day, not something I check in on separately or run alongside my normal work. It is part of my normal work.

It connects to our CRM data, internal documentation and operational workflows. It monitors initiatives across divisions, tracks KPIs, flags risks and surfaces opportunities I might otherwise miss. When I need to push seven sales initiatives forward simultaneously, I prompt it to draft the relevant communications for the right people and move them to review. When I need visibility into treasury, finance and operations at the same time, I have it. I am not stopping to sequence through conversations one at a time. I am directing multiple things simultaneously in a way that was not possible before.

What surprised me is not what the system can do. It is what it demands from me as a leader.

A new report surveying more than 300 enterprise FMCG marketers revealed a statistic that should concern every major brand leader: Only 1% of campaign ideas originate through testing-and-learning in public. Meanwhile, 41% still come from quarterly or annual planning cycles, and just 11% are driven by social or cultural insights.

That statistic helps explain why challenger brands continue to outperform incumbents in today’s attention economy. While enterprise organizations are still planning for culture, challenger brands are learning from culture in real time.

Increasingly, that difference is determining who wins.

Enterprise was built for control; culture wasn’t

For decades, enterprise marketing rewarded scale, consistency and risk management. Big brands controlled shelf space, dominated media buying and shaped consumer perception through carefully orchestrated campaigns.

Unfortunately, the way demand is created has fundamentally changed.

Today, discovery happens publicly. According to data from Socially Powerful, more than a third of enterprise FMCG marketers say social media and creators now drive more product discovery in their category than TV or search. At the same time, 86% say brand loyalty is weaker today than it was five years ago.

Consumers are increasingly less loyal by default. They’re influenced continuously by creators, communities, algorithms and online conversations happening at a pace traditional organizations simply weren’t designed to match.

That’s why challenger brands have become so dangerous — 7 in 10 enterprise marketers believe challengers outperform them on speed to market, from faster approvals to quicker creative production and publishing. But speed itself isn’t the real advantage. The real advantage is learning velocity.

Challenger brands test messaging publicly. They co-create with communities. They adjust based on audience response. They learn while moving.

Enterprise brands, meanwhile, are often trapped inside planning structures built around certainty. By the time a campaign survives approvals, legal reviews, stakeholder alignment and production timelines, the cultural moment it was designed for may already be over.

Culture ships daily. Most enterprises still operate quarterly.

Why most enterprise influence keeps resetting

One of the sharpest insights in the report is that enterprise influence still behaves like a burst. A campaign launches, attention spikes, engagement rises — and then everything resets once the spend stops.

That creates a costly cycle where brands repeatedly buy attention instead of building momentum.

The irony is that enterprise marketers already know where cultural understanding lives. According to the research, 81% agree that influencers understand culture and trends better than internal teams. Yet 62% still believe they can remain culturally relevant without fundamentally changing how they work with creators.

That contradiction explains why so much enterprise creator marketing still feels transactional.

Creators are often brought in late, after strategy is finalized, and used primarily for distribution. Challenger brands do the opposite.

They involve creators upstream as real-time intelligence networks that help shape positioning, messaging and product narratives while culture is still forming, allowing for a much quicker change of course should it be needed.

The incentive problem nobody wants to address

The challenge isn’t simply that enterprise organizations move slowly. It’s that most enterprise marketing systems were designed to reward predictability rather than learning.

When a brand manager presents a quarterly plan, success is often measured by how accurately results align with forecasts. Deviating from that plan can create operational complexity, even when the deviation is driven by genuine market insight. As a result, experimentation frequently becomes a side project rather than a core operating principle.

This creates a subtle but important asymmetry between incumbents and challengers.

Challenger brands are rarely expected to be right the first time. They are expected to discover what works through iteration. Enterprise brands, by contrast, often feel pressure to justify decisions before they reach the market. The consequence is that learning happens internally, while challenger brands learn externally.

The irony is that modern consumer behavior increasingly rewards the latter approach. According to Edelman’s Trust Barometer research, people place greater trust in peers, creators and individuals they perceive as authentic than they do in institutional messaging. At the same time, studies from McKinsey have consistently shown that consumers are more willing than ever to switch brands when presented with better value, convenience or relevance.

In other words, the market itself is becoming more dynamic while many enterprise operating models remain relatively static.

This is why the future competitive advantage may not be creative excellence alone, media scale alone or even data alone. It may be organizational learning speed: the ability to observe shifts in consumer behavior, test responses quickly and incorporate those learnings into decision-making before competitors do.

Ideally, brands should seek to combine the advantages of both systems. The proactive side handles traditional enterprise marketing launches, seasonal campaigns, retail moments and long-term brand planning. The reactive side operates continuously through creator partnerships, rapid experimentation, community feedback and ongoing cultural sensing.

The most effective organizations won’t replace planning with improvisation. They’ll build feedback loops into planning itself. Rather than treating strategy as a document that is reviewed quarterly, they’ll treat it as a living framework that evolves alongside consumer behavior. In practice, that means giving local teams more autonomy, shortening approval cycles, embedding creators earlier in the decision-making process and creating mechanisms for small-scale experiments to influence larger strategic decisions.

While allowing a reactive expansion of your brand may lead to some loss of autonomy, as trends are not always congruent with brand identity, it’s difficult to argue that either extreme is a sustainable model.

Acting like a challenger brand at enterprise scale can lead to inconsistency and an unreliable customer experience. Abandoning the challenger mindset entirely, however, is akin to rolling out the red carpet for emerging competitors.

The brands that win the next decade won’t necessarily be the loudest or the biggest spenders. They’ll be the ones capable of learning publicly while everyone else is still waiting for approval.

Right now, only 1% are built to do that.

A new report surveying more than 300 enterprise FMCG marketers revealed a statistic that should concern every major brand leader: Only 1% of campaign ideas originate through testing-and-learning in public. Meanwhile, 41% still come from quarterly or annual planning cycles, and just 11% are driven by social or cultural insights.

That statistic helps explain why challenger brands continue to outperform incumbents in today’s attention economy. While enterprise organizations are still planning for culture, challenger brands are learning from culture in real time.

Increasingly, that difference is determining who wins.

Right now, core financial processes in your organization — approving payments, matching invoices, forecasting cash — are likely running continuously and largely without human intervention. That’s the promise of finance automation, and for most companies, it’s delivering on efficiency. What it isn’t delivering is control.

The problem isn’t that automation is failing. It’s that it’s succeeding inside structures that were never designed to support it at scale. When automation expands across regions, business units and systems without clear enterprise-level ownership, it amplifies whatever fragmentation already exists, whether that’s inconsistent cash visibility, gaps in controls or capital decisions made on incomplete data.

The window to get ahead of this is narrowing. According to Gartner, 70% of finance functions will use AI for real-time decision-making on operational costs and cash flow management by 2028. The organizations positioned to benefit from that shift are the ones governing it now.

Over 17 years working in solution architecture and pre-sales strategy across global enterprises, I’ve seen automation become a liability, and I’ve seen it become a strategic asset. The difference is almost never the technology.

Most finance automation governance failures share the same root causes. Addressing them doesn’t require a technical overhaul, but it does require deliberate decisions about ownership, standards and visibility. Here’s where to focus:

1. Assign enterprise-level ownership, not functional ownership

Finance automation cannot sit in a gray area between departments. When no single person owns performance, risk and outcomes across the organization, each business unit fills the vacuum with local decisions. The result is a patchwork of workflows and approval thresholds that looks efficient at the unit level and incoherent at the top.

I watched this play out in a global manufacturing organization that had rolled out automation region by region, with each business unit optimizing locally by adjusting thresholds, redefining workflows and customizing reporting. Processing times dropped. On paper, it looked like progress. But the CEO faced a different reality: inconsistent cash visibility across regions, conflicting KPIs and increasing audit complexity. Treasury decisions were being made on incomplete data.

Once the CEO mandated centralized governance, including standardizing processes, aligning KPIs and establishing clear accountability, the company reduced working capital variance within two quarters and significantly improved global cash forecasting accuracy.

The lesson for CEOs and entrepreneurs? Don’t let automation sit in a gray area. Name an owner with enterprise-wide authority and give them the mandate to match.

2. Standardize where it matters most

Effective standardization targets the areas that directly shape risk and capital — cash management, revenue recognition and payment controls — and leaves room for local variation everywhere else. These are the processes where inconsistency creates real exposure, including audit gaps, inaccurate forecasting and working capital surprises.

Siemens offers a useful example of what this looks like in practice. Facing a sprawling network of thousands of decentralized bank accounts across multiple time zones, Siemens Treasury made centralization the foundation of its transformation. It simplified processes first, then automated on top of that structure.

The result was a reduction in bank accounts and cash pools by more than 50% globally, a 70% drop in internal management effort and an automated cash application rate of 80%, contributing to more than $20 million in annual cost savings. The gains came from standardizing the right processes within a governed framework before scaling automation.

3. Tie every automation initiative to a capital outcome

Too often, automation initiatives are evaluated on processing speed. Speed is table stakes. What matters is whether a given initiative improves cash flow, reduces risk, accelerates acquisition integration or expands margins, and whether you can measure it.

According to a Bain & Company survey of nearly 900 automation executives, companies that invested most heavily in automation reduced process costs by 22%, compared to just 8% for laggards. The differentiator is governance, not the technology stack.

In a private equity-backed services company I worked with, the CEO treated finance automation as a growth lever from the start. Automation initiatives were scoped around a specific thesis: faster integration of acquisitions and tighter cash management across a growing portfolio. Post-acquisition integration timelines shortened, and the company improved EBITDA margins by streamlining financial operations across entities.

That’s the difference between automation as a tool and automation as a strategic asset. If an initiative can’t be connected to a strategic outcome, it’s likely adding complexity without value.

4. Build real-time visibility into the system

This is where governance either pays off or exposes its gaps. Real-time cash visibility is a reporting feature, as well as the condition under which every capital allocation decision gets made. Without it, you’re operating on lagging, inconsistent inputs and making investment decisions accordingly.

According to Capgemini’s World Payments Report 2025, inefficient cash management, including poor forecasting and lack of visibility, costs businesses nearly 7% of revenue annually. At scale, that’s a governance problem, and the fix runs deeper than a better dashboard.

It requires treating data as infrastructure — a single, consistent source of financial truth that runs through your automation framework rather than sitting adjacent to it. Governance should be embedded in how decisions are executed, not applied after the fact. When it is, you gain what every CEO actually wants: clear visibility into cash positions, exposures and exceptions across the enterprise, in real time, without chasing it.

Automation shapes decisions as much as it executes them

Finance automation is changing not only how work gets done, but also how your business operates. Done right, it builds durable capability, the kind that supports growth, resilience and long-term value creation.

Governing automation effectively frees your leadership team to focus on what actually drives value — strategy, market positioning and growth — rather than reconciling inconsistencies behind the scenes. At scale, that makes it your concern, not your CFO’s.

Right now, core financial processes in your organization — approving payments, matching invoices, forecasting cash — are likely running continuously and largely without human intervention. That’s the promise of finance automation, and for most companies, it’s delivering on efficiency. What it isn’t delivering is control.

The problem isn’t that automation is failing. It’s that it’s succeeding inside structures that were never designed to support it at scale. When automation expands across regions, business units and systems without clear enterprise-level ownership, it amplifies whatever fragmentation already exists, whether that’s inconsistent cash visibility, gaps in controls or capital decisions made on incomplete data.

The window to get ahead of this is narrowing. According to Gartner, 70% of finance functions will use AI for real-time decision-making on operational costs and cash flow management by 2028. The organizations positioned to benefit from that shift are the ones governing it now.

Every company wants to be seen as an AI leader. That goal has created a bigger problem than many executives realize. I call it AI washing. It’s what happens when organizations spend more time promoting AI than proving its efficacy through measurable business outcomes.

I see it in executive meetings across industries. One leadership team proudly walked me through a long list of AI initiatives inside their company’s customer service operation. Response times improved. Automated routing reduced manual work. Every dashboard suggested the project was a success. Then I asked a simple question: What changed for your customers? At this, the room became quiet.

Customer satisfaction had slipped. Retention was moving in the wrong direction. Employees handled conversations more quickly, yet customers felt they were moving through a system rather than receiving help from people who understood them. The technology worked. The business outcome everyone cared about never improved.

That’s AI washing in action and, boiled down, it’s a leadership issue. The organization invested in AI, but the investment never translated into better customer or business outcomes.

After helping build Amazon Web Services and spending years guiding organizations through digital transformation, I’ve learned that AI creates value only when it improves the experience for the people you serve.

Stop measuring activity — start measuring outcomes

The customer service team I worked with wasn’t failing because the technology was bad. They were measuring the wrong things. Faster response times, lower handle times, and higher adoption rates all looked impressive on a dashboard, yet customer satisfaction and retention continued to decline. That experience taught me an important lesson. You shouldn’t base AI success on how much of it you deploy, but on what changes it precipitates.

Many organizations mistake implementation for transformation. They celebrate new platforms, pilot programs, and adoption rates while overlooking what matters most.

Measure what changed because of the investment

Did customers stay longer? Did employees spend less time on repetitive work and more time solving meaningful problems? Did leaders make faster, better decisions? Those are the outcomes that determine whether AI is creating value or simply creating more activity.

When leaders start with the business outcome instead of the technology, priorities become much clearer. The conversation shifts from “Which AI tool should we buy?” to “What business problem are we solving with this technology?”

Design AI around people

The customer service organization eventually changed its approach by redesigning the experience around customers instead of internal processes. Every decision came back to the customer. Does this change help the customers?

That shift produced far better conversations inside the company. Instead of focusing solely on efficiency, leaders began balancing customer experience with employee experience, leading to stronger business results. AI became a way to remove friction rather than simply automate tasks.

Every AI initiative should answer three questions before it moves forward:

When all three improve together, organizations create lasting value instead of temporary excitement.

Three steps to separate real AI strategy from AI washing

If you’re wondering whether your organization is creating real value or simply keeping pace with the latest trend, you don’t need another strategy session. You need an honest assessment.

Audit your biggest AI investments

Pull your three largest AI initiatives and ask one question. What measurably changed because of this investment? Resist the urge to talk about deployments or adoption rates. Focus on business outcomes. Did customer retention improve? Did employees save meaningful time? Did revenue grow? If you cannot answer those questions, you’ve found where your attention belongs.

Assign one accountable owner

Every successful transformation has someone responsible for the outcome. Review every AI initiative and make sure one leader owns the business result. When ownership becomes shared across multiple departments, accountability usually disappears. Clear ownership turns technology investments into measurable progress.

Talk to customers and employees before talking to another vendor

Ask customers whether they feel more understood than they did a year ago. Ask employees whether AI has made their work easier and more meaningful or simply more complicated. Those conversations will reveal more about the health of your AI strategy than another dashboard ever will. They will also tell you exactly where your next investment should go.

Measure what matters

AI will reshape every industry. That reality is already here. The organizations that benefit most will be the ones creating better customer experiences and delivering measurable business results.

Before approving your next AI initiative, ask yourself what will actually be different because of the investment. If you can answer that clearly, you’re building an AI strategy grounded in outcomes instead of appearances. That difference is what separates real transformation from AI washing.

Every company wants to be seen as an AI leader. That goal has created a bigger problem than many executives realize. I call it AI washing. It’s what happens when organizations spend more time promoting AI than proving its efficacy through measurable business outcomes.

I see it in executive meetings across industries. One leadership team proudly walked me through a long list of AI initiatives inside their company’s customer service operation. Response times improved. Automated routing reduced manual work. Every dashboard suggested the project was a success. Then I asked a simple question: What changed for your customers? At this, the room became quiet.

Customer satisfaction had slipped. Retention was moving in the wrong direction. Employees handled conversations more quickly, yet customers felt they were moving through a system rather than receiving help from people who understood them. The technology worked. The business outcome everyone cared about never improved.