TechPod Talks: Episode 3 Recap

Why Structure Is the Hidden Variable Behind Performance, Culture, and AI Adoption: Lessons Learned from Dr. Janet Sherlock

Ask a leader why execution is slipping, and you'll likely hear something about hiring, retention, or culture. Structure almost never comes up, and that blind spot is expensive. The shape of an organization determines where decisions are actually made, which teams deliver which outcomes, and whether strategy turns into results or dissolves into coordination meetings.

This episode's guest has spent three decades watching that dynamic play out from inside the room, across industries as unlikely as downstream petroleum and global luxury retail, then returned to the question academically through a doctorate focused specifically on organizational design.

Meet Our Guest

In Episode 3 of TechPod Talks, our CEO, Stephen Koza, sits down with Dr. Janet Sherlock, founder and CEO of Org.Works, an advisory firm built around the observation that organizational structure is one of the most material and most overlooked drivers of company performance today. Janet originally founded the firm to help CEOs and boards see what her research revealed and act on it before costs compound.

Her career spans thirty years and an unusually wide aperture, though she'll tell you it all boils down to one industry. She spent her first fourteen years in downstream petroleum at ExxonMobil and BP, which she cheerfully calls "the least sexy sector of retail that you could possibly be in." 

From there, she led Digital and Omnichannel at Gartner, served as CIO at Carter's, and, most recently, held the role of Chief Digital and Technology Officer at Ralph Lauren, where she oversaw a multi-billion-dollar global e-commerce business and led the enterprise-wide modernization of technology, data, and processes. Then, during the pandemic, she added a doctorate in Organizational Change and Leadership from USC, with a dissertation on the impact of overlapping C-suite roles. The findings surprised even her. The effects were more negative than she'd expected going in, and they've shaped her thinking ever since.

Her TEDx talk on organizational clarity was selected as an Editor's Pick, a distinction awarded to fewer than one percent of TEDx talks globally. After all of this, the throughline across every stage of her career showed the same trend: take retail, e-commerce, and the technology that powers them, and make the teams behind them actually work.

Episode 3: Key Takeaways

• Why vertical organizational layers are often more expensive than horizontal ones, and how to identify the "teams between teams" that quietly absorb value
• A framework (the CED model) for organizing AI capabilities so pilots convert into production, not just PowerPoint decks
• How to decide when to bring in outside expertise without creating dependency or abdicating ownership
• The signals that reveal a performance problem are actually a structure problem, not a people or strategy problem
• Why middle managers may be the most consequential people to keep during AI transformation, and why indiscriminate flattening backfires
• How candor delivered with respect is a form of kindness, and why avoiding difficult conversations damages teams over time

If you're a CTO, engineering leader, or senior operator watching your org add roles, layers, and AI pilots without proportional output, this conversation is a diagnostic most leadership books don't give you. Every insight comes from someone who has built organizations at a global scale, dismantled the parts that weren't working, and studied the underlying patterns academically.

Vertical Layers Are the First Place to Look When Flattening

The default instinct when an organization feels bloated is to flatten it horizontally by cutting middle management, trimming the org chart, and reducing bureaucracy. However, Janet's research pushes back on that instinct, and she makes the case for starting somewhere else entirely.

The backdrop is familiar. Meta, Amazon, Microsoft, Intel, and Expedia have all aggressively reduced mid-level roles as part of what industry observers are calling the “Great Flattening.” The rationale blends cost-cutting with AI integration, and on paper, the logic is clean. But Janet argues the execution is usually wrong because leaders are cutting on the wrong axis. Before touching the horizontal layers, she recommends looking vertically first, specifically at the coordination layers between teams that should have been talking to each other all along.

She provided two examples that landed hard. The first is a customer experience team wedged between marketing and sales. On the org chart, it looks reasonable, even forward-thinking. In practice, it usually becomes a buffer that slows decisions and absorbs value without ever producing any. 

The second example is one she says she's seeing all the time right now, and involves a standalone AI team that sits between the technology function and the business. The intent is to centralize expertise, but the effect is the opposite of what anyone hoped for, because the new layer slows the very thing it was created to accelerate.

These vertical layers are almost always the residue of organizational growth that was never rationalized. Someone got promoted. Two functions drifted apart. A coordinator got hired to bridge them. The bridge calcified into permanent overhead, and nobody ever went back to ask if it was still doing work. Removing those layers does more than reduce payroll. It shortens decision paths and accelerates response time, which matters more in the current moment than at almost any point in recent business history.

The deeper point Janet makes is about who actually understands the work. Middle managers are frequently the only people in the organization who know the processes and workflows well enough to identify where AI agents can genuinely create value. Senior leaders often lack that operational granularity. Individual contributors often lack the strategic context to re-engineer processes around AI. Cutting the layer in between removes the people best positioned to translate ambition into implementation.

"Indiscriminately just removing that layer of middle management may potentially be one of the most consequential reduction-in-force mistakes and strategies organizations can make right now."

That line reframes the entire conversation about flattening. The real work is identifying which layers exist because they earn their place and which exist because nobody ever looked closely enough to ask.

How to Recognize When a Performance Problem Is Actually a Structure Problem

Execution failures get labeled as people problems or strategy problems by default. Janet argues the real culprit is usually structure, and she's specific about the signals leaders should watch for.

She starts with a claim worth sitting with, as structure is a prerequisite for culture. Without the structural pieces in place, cultural problems become nearly inevitable. On the surface, those problems show up as missed timelines, inconsistent execution, frustration, and cross-team infighting. But look closer, and the patterns are almost always structural.

Janet told a story from her own career about omnichannel retail. For years, consumer expectations shifted faster than retail organizations did. Customers wanted to transact seamlessly across channels and platforms, and making that a reality required retail, digital, and e-commerce teams to operate as a single, coordinated group. 

Throughout her career, she structured those teams to work together for exactly that reason. To create faster speed to market, unified customer experiences, and genuine economies of scale. From shared digital assets across retail and e-commerce to product recommendations that flowed from kiosks into point-of-sale systems and in-store displays, these are the things that feel effortless from the customer's side but depend entirely on the teams behind them being structured to collaborate.

The counterexample came from a CEO of a quick-service restaurant chain she was advising at the time. She explained he'd maintained separate teams for restaurant systems and for the mobile app and digital experience, but the moment they tried to launch in-store kiosk ordering, the wheels came off. The kiosks needed a user interface that mirrored the app and website while also reflecting the restaurant's underlying technology. The disjointed architecture was a direct reflection of the disjointed org chart. Different teams fought for control, turning what should have been a straightforward build into a structural fight that took far longer than it should have.

The signals to watch for are concrete. Multiple leaders are engaging in the same decisions without clear ownership. Duplication of effort where teams build toward the same endpoint, like a retail team and a digital team each running their own QA process, while the company still needs a third QA pass just to verify the unified consumer experience. And the real meeting tell? If a CEO needs two or more people to brief them on the same topic over an extended period, the problem is structural.

"Great people and good strategy don't fail on their own. Most of the time, they fail because the structure that was around them isn't doing its job."

That line has the kind of weight that can reframe a quarterly review. It also sets up one of the episode's sharpest answers, where AI initiatives quietly go to die.

The CED Model for Scaling AI From Pilot to Production

Proof-of-concept purgatory is the defining AI complaint in the enterprise right now. Pilots run, demos impress, and then the work stalls somewhere between the excitement of the first result and the reality of actually shipping. Janet's view is that most of those failures trace back to structure, not strategy, and she offered a specific model for solving it that EverOps’ readers, familiar with the AI Opportunity Assessment, will recognize as directly relevant.

She calls it the CED model: Center of Enablement, Federated data science, and Democratized data and insights. Three layers, each with a distinct job, and designed to work as a system rather than compete for turf.

The Center of Enablement typically lives within the technology function. It owns the platforms, data, infrastructure, and governance that make everything else possible. This is the foundation layer. Without common platforms and standards, every team quietly reinvents the wheel in isolation, and the organization ends up paying three times for the same capability.

Federated data science lives inside the business units, where the domain expertise actually sits. These are the teams closest to the high-value use cases and closest to the outcomes the business needs to hit. They lean on the platforms and services the Center of Enablement provides, and they own the problems they solve. It's the structural equivalent of pushing budget ownership to the teams that control the spend, and it works for the same underlying reason that accountability and capability need to be in the same room.

Democratized data and insights form the third layer, opening access to data and personal AI tools across the broader organization. This is where a company-specific GPT and individual productivity tools live. It's where people at every level get to work more efficiently without waiting for a centralized team to build something for them.

"When you have a structure like this, this creates a flywheel, because now you have people who are empowered within their own areas to actually do what's needed for their particular domain."

There are two things that can break the flywheel. The first is weak governance. Without a clear prioritization process, requests pile up at the Center of Enablement, making it the very bottleneck it was designed to prevent. The second is the opposite failure, where teams bypass governance entirely, and a sprawl of ungoverned models and tools takes hold. Janet's fix is elegant. She recommends layering AI prioritization into the existing IT portfolio management process. Most mature organizations already have that muscle, but the real work is extending it. That's a different kind of work than building a new capability from scratch.

The framework surfaces a quiet truth about ownership. In the customer service example Janet walked through, the business team owns the model, not the platform team. The Center of Enablement supports this, and the business drives. That single inversion explains why so many enterprise AI initiatives stall, and why the ones that succeed keep moving.

When to Bring in Outside Expertise Without Abdicating Ownership

For senior leaders, the decision to engage a consulting partner usually gets framed as a resourcing question. Janet argues that framing is wrong, and she laid out a four-condition model for thinking about it that will also be familiar to anyone who has worked with EverOps' strategy services.

The first condition is a true capability gap. When a company is implementing something new, a new ERP, an e-commerce platform, an enterprise digital asset management system, the internal team often simply doesn't have the expertise yet. Janet shared that, during her own tenure at Ralph Lauren, when they first implemented the e-commerce platform and the enterprise digital asset management solution, she relied heavily on external resources for exactly that reason. The internal knowledge hadn't been built yet. The goal in those situations is to use outside expertise to build the muscle, then shift ownership inside.

The second condition is objectivity and independence. Certain decisions, system selections, sensitive org design work, RFP evaluations, benefit enormously from a third party who isn't carrying the history or the politics. Janet was specific that organizational decisions, in particular, can be dangerous to leave entirely to internal teams, because the people closest to the outcome are rarely the most objective judges.

The third is speed. Deadlines, strategic pivots, and moments where time is the binding constraint call for external support to accelerate execution across the finish line.

The fourth is value optimization. This is about building an effective resource mix that blends offshore teams, contractors, and specialized partners so that the highest-value internal talent can work on the highest-value problems. Janet was careful to draw a line between this and labor arbitrage. The goal is intentional allocation of work, not the cheapest available execution.

The flip side of the framework matters just as much. Companies get into trouble when they default to external help for things they should own, specifically strategy, decision-making, and core capabilities. Janet has walked into environments where so much had been outsourced that nobody inside the company understood how IT actually worked. That's a structural vulnerability dressed up as a cost efficiency, and it takes longer to unwind than it did to create. 

Stephen framed the right approach through EverOps' three Cs: companies engage partners for competency, capacity, and certainty. Janet's four conditions map cleanly onto that framing.

"The goal isn't to outsource everything. It's to augment your team without abdicating accountability."

For AI specifically, she was even more direct. Internal capabilities matter more right now than in previous technology waves. Over-reliance on third-party tools creates black boxes that can't be explained, and the companies that will win are the ones that own enough of their destiny to understand what their systems are actually doing.

Candor Is a Form of Respect

The conversation closed on leadership, and Janet's hard-earned lesson is one many senior leaders have learned the expensive way.

Early in her career, she avoided challenging conversations. When feedback needed to be delivered, she softened it, sometimes to the point where the message dissolved inside the words around it. She told herself she was being thoughtful. She told herself she was being considerate. What she was actually doing was creating ambiguity, and ambiguity is one of the most corrosive forces in any workplace. It doesn't help people. In fact, it actually holds them in place.

The turning point came with a specific colleague, a CMO she'd worked with a few years ago. Their relationship had grown tenuous, and Janet spent a long stretch frustrated, complaining about him to her husband and to people outside of work. Eventually, she did the thing she'd been avoiding. She sat down with him one-on-one and walked through her perspective, her read on his perspective, and the specific things he said and did that were affecting the working relationship. That conversation completely changed the arc of the relationship. Not every difficult conversation resolves that cleanly, but the experience made her realize how often leaders avoid the exact conversations that would actually fix things.

She now applies the principle broadly. When employees ask her why they weren't promoted or why someone else got a stretch assignment, she tells them directly. Some people filter the feedback out. Most people want it, value it, and use it. But they need leaders willing to deliver it clearly and constructively in the first place.

"Candor delivered with respect is kind, and lack of candor is not kind, and it's just avoiding its avoidance…It does end up creating long-term harm over the long run."

Stephen named the failure mode most leaders recognize at first sight. Someone comes in, you chat about other things, you slide the feedback in somewhere in the middle, you finish with something light, and they leave without ever having heard the actual message. Janet described the same pattern. It's the compliment sandwich, and it's avoidance dressed up as politeness.

Janet now sees this leadership gap across the companies she works with. The cultures that genuinely outperform are the ones where candor is built into the operating rhythm, part of how the team communicates by default, rather than something reserved for crisis moments or annual reviews.

How to Apply These Insights Today

Based on Janet's framework and specific advice, here are concrete actions senior leaders can take this quarter:

1. Audit your vertical layers before your horizontal ones: Map the teams that sit between other teams. For each, ask whether they solve a structural problem or whether they exist because two functions drifted apart. Look hardest at coordination roles, customer experience teams wedged between marketing and sales, and any standalone AI team that sits between technology and the business.
2. Diagnose your execution failures structurally first: The next time a team misses a deadline or fights with another team, resist the instinct to label it a people problem. Ask whether multiple leaders are engaging in the same decisions, whether two teams are duplicating effort toward the same endpoint, or whether you personally need multiple people to brief you on the same topic over time. Any of those is a structural signal.
3. Evaluate your AI structure against the CED model: Check whether your Center of Enablement owns platforms and governance, whether your business units have domain-led AI ownership tied to real outcomes, and whether individual employees have access to personal AI tools. Most organizations have one or two of the three. The flywheel requires all three, along with a clear prioritization process that ties them together.
4. Apply the four-condition test before engaging outside partners: When you're considering a consultant or services partner, map the need to capability gaps, objectivity, speed, or value optimization. If it doesn't fit one of those four, you're probably outsourcing something you should own. If it does, be equally clear about what stays internal: strategy, decision-making, and core capabilities.
5. Have the candor conversation you've been avoiding: Pick the one relationship where ambiguity is costing you productivity and sit down for it this week. Janet's advice is specific: explain your perspective, describe what you believe their perspective is, and name the concrete moments that have affected the working relationship. The conversation may not go perfectly, but the clarity on the other side is worth the discomfort.

Keeping Up with TechPod Talks

If you found this conversation useful, we highly recommend you continue tuning in to TechPod Talks, as we are just getting started. Subscribe now so you don't miss what's coming next. Every episode features a practitioner who has actually built and operated at scale, sharing what works in technology leadership, organizational design, cloud, AI, and modern operations.

Subscribe to any of these to tune in:

Apple Podcasts | Spotify | YouTube | EverOps Podcast Page

Follow for additional clips, behind-the-scenes content, and future episode announcements:

• https://www.linkedin.com/in/stephenkoza/  
• https://www.linkedin.com/company/everops/ 

Books and resources mentioned:

• Dr. Janet Sherlock's TEDx talk "Clarity Before Chaos: Building Organizations That Work": Referenced as a TEDx Editor's Pick, an honor given to fewer than one percent of TEDx talks globally, and a concise articulation of the structural clarity principles Janet covers throughout this episode.
• Dr. Janet Sherlock's doctoral research on overlapping C-suite roles (University of Southern California): The research base behind the CED model and the structural signals discussed in this episode. Her dissertation examined the organizational impacts of overlapping executive roles and found effects more negative than she anticipated.

Connect with our guest, Dr. Janet Sherlock, on LinkedIn today:

https://www.linkedin.com/in/janet-sherlock/ 

Learn more about her advisory firm: https://org.works 

Frequently Asked Questions

What is TechPod Talks?

TechPod Talks is a podcast hosted by EverOps CEO Stephen Koza featuring candid conversations with technology leaders, engineers, and operators. Each episode explores how real teams build, scale, and operate modern systems, with a focus on practical takeaways.

Who is this podcast for?

The podcast is designed for CTOs, engineering managers, DevOps and SRE professionals, platform engineers, and technical operators, especially senior leaders at companies navigating cloud optimization, modernization, organizational design, and AI adoption.

Where can I listen to TechPod Talks?

TechPod Talks is available on Apple Podcasts, Spotify, YouTube, and the EverOps website. Episodes are released in both audio and video formats.

What topics does Episode 3 cover?

Episode 3 features a candid conversation with Dr. Janet Sherlock on organizational design, including why vertical flattening usually matters more than horizontal flattening, how to diagnose execution failures as structural problems, the CED model for scaling AI from pilot to production, a four-condition framework for bringing in outside expertise, and why candor is a form of respect.

Who is Dr. Janet Sherlock?

Dr. Janet Sherlock is the founder and CEO of Org.Works, an advisory firm focused on organizational design. She previously served as Chief Digital and Technology Officer at Ralph Lauren, CIO at Carter's, and as the leader of Digital and Omnichannel at Gartner. She holds a doctorate in Organizational Change and Leadership from USC and delivered a TEDx talk on organizational clarity that was selected as an Editor's Pick.

Can I suggest topics or be a guest on the podcast?

Yes. You can share topic suggestions by reaching out on LinkedIn or through the EverOps website, which includes a guest request form for speakers interested in joining future episodes.

How does this episode connect to EverOps' work?

EverOps helps technology leaders navigate the same structural and execution questions Janet covers in this episode, particularly around AI adoption, organizational design, and knowing when to engage outside expertise. Services like AI Opportunity Assessment, strategy consulting, and embedded operations map directly to the frameworks Janet describes.