Directionally Correct, A People Intelligence Podcast

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Sanjeev Sharma, Workforce Innovation, Value Realization & Principal Member of Technical Staff at Salesforce! In this episode Cole Napper speaks with Sanjeev Sharma, Principal Member of the Technical Staff for Workforce Innovation and Value Realization at Salesforce, about why AI is quietly gutting traditional analytics architecture by letting anyone build, Salesforce’s proactive playbook for catching a role before it crosses the 51 percent AI threshold, and why he remains skeptical of both AI doomsday predictions and the surrounding hype. Value realization sits between today’s population and the population needed to stay competitive: it tracks how humans plus AI produce P&L and market impact, treating work as trajectories and capacity compositions rather than static headcount states inside a traditional HCM. The practical stack is a computational data layer plus an intelligence layer, with delivery now possible through conversation, predefined skills, or other guided mechanisms that keep efforts inside the company’s corpus. Inside Salesforce the Chief People Officer is pushing HR from transaction and satisfaction work toward genuine value creation. That requires organizational courage to speak the business’s language of cost and value one layer deeper than simple job blocks, while still validating rather than betting on futures. The shift that “everybody feels like they can be a builder” is real and useful for prototyping, yet Sanjeev warns that unlimited energy without connection to company trajectory is entropy; attention spans shrink and token spend rises unless someone joins A to B. He introduces an equation in which every role’s economic value decays over time; AI accelerates the decay so that the floor can drop before people re-skill. The practical response is preventative: analyze and intervene before a role is more AI than human (the 51 percent threshold), choosing augmentation over pure automation and preserving an irreducible human floor for compliance and judgment. When pressed on hiring three to five years out, Sanjeev and Cole discuss the limits of rigid five-year plans; better framing is AI workforce transformation that keeps options open across multiple future branches while protecting core capabilities. The episode reviews three of Sanjeev’s Substack pieces. “The Pen Still Writes” argues that most future-pay ideas are still sharp proxies for value inside the old structure; real change requires ownership tied to the products one actually builds and greater portability so people are not stuck at a company-controlled discount. “The Cliff Is Here” examines how standardized agent-to-agent protocols make trust portable yet create cliffs when systems write to systems and humans lose sight of the trail; network analysis overlaid on org charts becomes essential to see the liquid organization. “Confidence Laundering” describes the recursive process in which probabilistic outputs are narrated into over-confident deterministic claims; tools must label confidence at the data, semantic, and decision layers rather than simply “toss it into an LLM.” Throughout, Sanjeev models a technical individual-contributor path that joins near-term rent-paying work with longer-horizon innovation, insisting that curiosity is the skill that matters most and that HR’s new job is to quantify capacity, engineer trust, and keep the human floor high enough that value realization remains a choice rather than an accident. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Marina Pearce, Senior Director of Workforce Intelligence at Databricks! In this episode Cole Napper talks with Marina Pearce, Senior Director of Workforce Intelligence at Databricks, about what it really means to create an intelligence layer and how that differs from people analytics of just a few years ago. Marina describes a shift from analysts as reporters who alone held the keys to data and definitions to architects who build foundational ecosystems that let everyone ask and answer questions. The work centers on unifying information, securing and governing it, keeping gold tables reliable, and certifying dashboards, agents, and apps so users trust the outputs without needing a human check. She emphasizes democratizing capability rather than raw data, then moving beyond interaction into helping people automate their own work. Building the team looks different in a lean, high-velocity data and AI company. Instead of traditional specialist roles (data engineer, data scientist, analytics partner), Marina prefers full-stack generalists with high G, deep curiosity, and a strong “figure it out” orientation. Many on her current team did not come from people analytics backgrounds. Context matching matters more than any cookie-cutter playbook: understand the company’s operating rhythms and constraints first, then design the function to remove friction. Once the platform exists, HR partners and recruiters begin building their own models and apps on certified assets; the central team’s job expands to support that speed rather than gatekeep it. Marina introduces “radiating intent”—openly publishing roadmaps, dependencies, and vision so others can poke holes, align, or help. Value shows up in system-level indicators such as operating efficiency per employee rather than isolated study ROIs. Rolling out a company-wide intelligence layer required deliberate change management: clear expectations, delightful tools, and reinforcers, plus a framework of mindset, skills, usage, and impact. An AI champions network, hypercare support, and leadership modeling accelerated adoption once people experienced the platform and became “hooked.” Technical foundations and security remain essential, yet the harder work is changing habits so people prefer the new tools. Looking past intelligence, Marina sees the next horizon as workflow automation that individuals and teams own themselves through job crafting, not something done to them by a central hub. She also reflects on career paths: business exposure (HRBP-style work or rotations) is often the missing piece for people-intelligence professionals who aspire to broader people leadership roles. The conversation covers three pieces: an article on whether workplaces drain curiosity (and the risk of isolation when AI becomes the middleman), Paul Spector’s critique of formulaic academic papers and the incentives that reward complexity over clear findings, and a large tech-worker survey showing burnout, ambivalence about recommending the field, and the enduring importance of managers. Cole and Marina note the gap between apocalyptic public narratives and the day-to-day experience of many practitioners who already feel amplified by the tools. Throughout, Marina models the philosophy of being loud and opinionated internally so ideas improve, while remaining practical about technical foundations, security, and human habits. The episode paints a concrete picture of what is already possible today when a people team treats itself as infrastructure and enables the rest of the organization to build. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this HR Tech Voices Series episode of the #1 people intelligence podcast with special guests, ⁠Jason Murphy, Chief Product Officer and Matt Maguire, Chief Revenue Officer at HRBench! In this episode Cole Napper sits down with Jason Murphy, Chief Product Officer, and Matt Maguire, Chief Revenue Officer at HRBench to unpack what it takes to build a people intelligence product and why people analytics must move from a transactional function to a product-based one. Jason defines the product role as understanding customer problems and delivering business value so users accomplish their work more efficiently. Matt describes the collapsed feedback loop: client conversations that once required weeks of mockups can now produce interactive prototypes in hours or days. Jason takes real feedback, builds code-based prototypes matching the application, gathers reactions the next day, and hands them to engineering—compressing timelines dramatically while improving accuracy. The discussion centers on the architecture of people intelligence. After dashboards fade, only an intelligence layer and a data layer remain. HRBench acts as the extensible middle layer that cleans data from disparate sources, applies methodology and health checks, and securely passes it via MCP connectors into other AI tools. Organizations seek to eliminate dashboard maintenance, surface engagement insights, and score AI impact by mapping company jobs to a proprietary taxonomy that includes tasks, skills, and predicted effects. Jason details reskilling and redeployment tools so leaders identify impacted employees and move them into valuable roles. Matt notes that even advanced teams lack clear answers, creating opportunity for AI-native platforms free of legacy technical debt. Practical examples include headcount planning workflow approvals that replace spreadsheet processes, expanded performance features such as stack ranking and nine-box, richer employee listening, and org charts that combine open requisitions with planning. Tokenomics is rising: capturing agent token spend, assigning annualized costs, and reflecting agents in the org chart so plans show human and machine costs plus hours returned. Jason explains how AI tools transformed product work itself by automating QA test cases and documentation that follows a custom house style, collapsing weeks of effort into hours. Looking a year ahead, barriers between applications dissolve. Users will work inside a vendor interface or pull data through conversational layers; specialized point solutions will fade as large organizations rely on an intelligence middle layer while smaller ones still need accessible platforms. Cole adds that systems of record must rebuild AI-natively and that former products become features inside broader intelligence and workflow layers. The episode reviews Jack Costl’s argument for portfolio management so teams choose high-value problems rather than polishing every request; Ludic Stella’s analysis linking culture signals such as transparency to return on assets; and a HumRRO paper on indexing AI impact within O*NET through task- and KSAO-level scores. Jason highlights the difficulty of tracking current AI capabilities given rapid model improvement. In Cole’s Corner the guests address rapid-fire questions and the near-term future of intelligence products. The conversation stays grounded in real customer problems, the speed of AI-native tooling, and the shift from dashboards to solutions that redesign work. Supporting platforms and leaders who deliver this practical progress is essential as years of roadmap compress into months. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, ⁠Yuyan Sun, Former Head of Workforce Analytics, AI Transformation and Technology at Motive and Co-Founder at The Zeroth! In this episode Cole Napper and Yuyan Sun explore the future of people analytics and how practitioners can position the practice inside AI-forward organizations. Yuyan argues the field has leaned too long on altruistic voices without rewarding those who give back. Community leaders who open calendars after layoffs, write freely, and share research build the profession, yet vendors often capture the value. She and Cole stress the need to pay it back as well as forward by thanking those who helped you and supporting them when they commercialize expertise so the community can sustain itself. Yuyan left her role as head of workforce analytics, AI transformation, and technology at Motive to co-found The Zeroth with Dan Riley. The Zeroth gives senior AI transformation leaders a quiet peer space to exchange authentic insights on real challenges free from vendor hype and investor noise. At Motive she applied tools through rapid experimentation such as benefits chatbots that scaled into broader solutions. They address AI economics head-on. The technology is expensive and most organizations will lose money in the first six to twelve months of serious experimentation, just as earlier digital transformations required more specialized staff rather than fewer. Real value requires redesigning work, structures, skills, and learning programs while preserving human judgment and creativity that differentiate a business when competitors use the same models. Dashboard and report work is already disappearing or being automated. Practitioners will serve as AI ambassadors inside HR and then help organizations measure capacity and intelligence as AI becomes part of organizational capability. Yuyan and Colby Nesbitt called this a midlife crisis because field growth was tied to headcount expansion during easy-capital booms that are now slowing. Cole contrasts early trailblazers who wrote the playbook with later button-pushers. As AI automates the commodity side, a second renaissance of trailblazers is needed across HR. The function must decide whether to remain a risk-mitigation shop or become a value-creation engine willing to ruffle feathers. They also examine risks. Personality assessments that are not strictly job-related can create monocultures, and AI-generated resumes screened by AI systems amplify the problem. Assaf Shahar’s ideas on cultural flattening highlight the danger of outsourcing judgment to LLMs, producing quick but unaccountable outputs and cultural debt. Cole and Yuyan note the field’s origins in talent analytics were always about performance through people. The sadness is real, yet the answer is a compelling vision of people intelligence rather than nostalgia. In Cole’s Corner rapid-fire questions reveal Yuyan’s entrepreneurial drive, her wish to visit an outer-planet rock, and her expectation that the typical analytics professional three years from now will have automated most data wrangling and reporting with agents then focus on defining the right problems and metrics. The episode closes with a personal exchange about the value of authentic conversation and the hard work of producing content while the field struggles to commercialize the contributions of its community leaders. Supporting those leaders and sponsors like HRBench is essential if the profession is to mature. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, ⁠Ann-Marie Clayton Johnson⁠, Senior Director, Human Resource Business Partner at Microsoft & Former Salesforce People Analytics Leader! Cole has known Ann-Marie since she was a Salesforce people analytics leader he looked up to. Then she became an HR business partner at Microsoft, and this episode asks why. Analytics was the highlight of her last decade: curiosity, access, and a team that shaped strategy by defining the metrics, work she credits in part to Ernestine. She left before seniority locked the door. Companies hesitate to trust a senior analytics person in a role they have never held, even though she would take that risk herself. Cole ties the move to advice from the CHRO practice at Heidrick & Struggles: business partner experience is often the barrier between people analytics and the top job. What an HRBP does depends on the company. Smaller firms fold in operations. At Salesforce, partners did communications and employee experience. At Microsoft, Ann-Marie supports senior leaders in SMB and channel, a high-growth business whose sellers carry hundreds of accounts rather than enterprise’s one-to-one model. Her weeks are leadership development, off-sites, talent walks, succession, and org design, including SMB-specific roles inside functions built for enterprise. The sharper turn is producer to consumer. Any partner worth their salt should use analytics. Her background helps her see the system of data. On a low-performance steerco she pushed upstream: what predicts low performance, where the hot spots are, which managers deserve a double click, before a case opens. Connect the points so the story is visible. An agent or a dashboard does not change the world. She has been at Microsoft only since February. People analytics had gone tools-forward and help-yourself. It did not work, and consulting is coming back. Nate Schultz joined from Salesforce, aligned to her skip level, with a team supporting sales. She wants those consultants senior enough to know statistics and the business. A leader who wanted people back in the office became the test. About 19 people were going in, not enough to know whether attendance helped. The answer came back as not statistically significant. She needed practical significance, and a partner who will say the variables are wrong. Embedding fails unless people are held accountable in writing. If data use is not in Connects goals, it will not show up. She wants a culture that still asks how do you know. People analytics is not dead, but the hypothesis work she loved is fading, while the technical layer stays because AI needs people who can structure and choose data. Anyone too far from the business will not last. Cole’s inquisitor and change agent frame matches: ask, analyze, act. The field over-indexed on the automatable middle, and people intelligence flattens the rest into a data layer and an intelligence layer. I-O training let her skip years across centers of excellence. She shares the resume-bullet framework and the S-curve: leave while the work is energizing and build breadth instead of waiting on a pyramid of junior analysts. She is a military spouse. Ryan went to West Point and deploys often. She helps transitioning service members and is PTA president for a year. Cole’s Corner covers teaching, a Texas shore, and Harry Potter. As a future CHRO she would not lose the employee experience that pulled her in after Cisco. They take on Andreessen on talent and workplace bullshit versus lying. Change comes from winning companies. Find her on LinkedIn as Ann Marie Clayton Johnson. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, ⁠Kelly Monahan⁠, Future of Work & AI Advisor and recent Author of “Reclaim the Plot”! Cole Napper sits down with Kelly for a wide-ranging conversation that moves from the uncanny to the practical and back again. Kelly recounts spending six hours in a recording studio so an AI digital avatar could learn her likeness, then months training it, only to watch that digital twin land in the Wall Street Journal. Her first reaction was that it was creepy: a likeness that was her and yet not her. Over time she began using it to open keynotes and to show that the AI conversation is larger than large language models, though she is candid that it is still more novelty than something she would send into a meeting the way Reed Hoffman does. That experiment sits inside a larger argument. After years in tech, Kelly concluded we have lost the plot on the pursuit of AI. The technology has existed since the 1950s; the current wave is real and transformative, especially agentic systems, but treating early ChatGPT as an excuse to hollow out knowledge work without redesigning the work itself is a mistake. Productivity gains show up, she says, citing McKinsey and BCG, only when work redesign accompanies the tools. Only about 21 percent of companies have done that redesign. Without it, people simply send more email and sit in more meetings. AI becomes an expensive way to illuminate how poorly designed the work was in the first place. Kelly walks through a simulation she built in hours for her forthcoming book’s fictional company, Driftwell—work that would once have taken a team ninety days and tens of thousands of dollars. A parent agent spun up specialists, connected to HubSpot, and orchestrated the whole thing. She defines an agent as a system that takes action toward a goal rather than merely chatting. The same pattern appeared in her four years at Upwork around the November 2022 “D-Day” of ChatGPT: the freelancer market entered an existential crisis, disruption arrived slowly rather than overnight, early AI users earned more and bid on larger projects, and inequity widened along a power-law distribution. The lasting finding was that this is a massive skilling moment, especially for transactional content work. The conversation turns to leadership. Most senior leaders mandate AI without using it themselves, either because they believe strategy is too elevated for tools or because they fear the C-suite itself will shrink. Intelligence and expertise are being commoditized; what remains is change management and the messy human middle that agents cannot yet own. Accountability becomes the billion-dollar question: if two hundred agents gather the inputs, can an executive still be accountable for a decision they cannot fully verify? Span of control, entry-level learning-by-doing pathways in consulting, and the dignity of physical work all get examined. Kelly worries we told a generation to abandon the physical world even though those jobs often carry meaning, community, and contribution that leisure-only futures may not replace. In Cole’s Corner they cover growth mindset (Kelly argues the label matters less than accompanying behavioral interventions and human relating), the 1975 “folly of rewarding A while hoping for B,” broken performance management, and eye-watering revenue-per-employee figures that raise questions about reinvestment, customers, and long-term social consequences. Kelly points listeners to LinkedIn and reclaimtheplotbook.com. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Kevin Trowbridge, Director of Compensation and HRIS at Kimray! In this conversation Cole Napper sits down with Kevin to unpack what it means to lead compensation and HRIS as one combined function at a 78-year-old Oklahoma manufacturer that had no dedicated role for either until 2024. Kevin landed in the work almost by accident: a math degree, a banking career that burned him out, and a friend who spotted a utility job that wanted someone who could handle numbers. That first administrator seat taught the unglamorous truth. You run the reports, match the jobs, build the structures, and learn that compensation is finance with a human being on the other side of every figure. The work still looks more like an apprenticeship than something you can absorb from a book. WorldatWork’s CCP typically takes three to five years because the classroom only sticks if you are already living the messy decisions. Comp ratios, percentiles, internal equity, and survey matching all sound like people-analytics language, so Cole asks why compensation and people analytics so rarely sit in the same room. Kevin’s answer is blunt. Pay makes people nervous. Companies that lag the market do not want that exposed. Culture can offset dollars, and at Kimray it often does. Tenure of 40-plus years is still common. Leaders teach culture to other businesses through a foundation. Frontline supervisors get executive-level leadership training. People stay because the values match, not because every offer sits at the 75th percentile. Kimray manufactures pressure regulators, valves, sensors, and level controllers for oil and gas, made in the USA. Until May 1 of this year it was family owned. It is now an ESOP, so employees become beneficial owners through a company-funded retirement plan that sits on top of the 401(k). That ownership story collides with another unusual design choice: the CEO refuses traditional sales commissions. He does not want credit-stealing or split-store fights. Sales grew from roughly 100 million to the 230–250 million range with a company-wide EBITDA bonus instead. The team is now testing a group accelerator so everyone rows toward the same outcome. AI is the thread that keeps Kevin up at night. Tools like Claude already save his analyst hours and even teach him Python. HRBench turned a ten-minute request for three years of departmental turnover into a clean, trusted slide. Workforce planning lets finance and leaders add, subtract, and cost positions in the same conversation. Kevin still asks the moral question: if AI swallows the entry-level report running and employee-relations grunt work, who develops the next generation of compensation and HRIS professionals? In ten years the seniors will retire unless companies keep the apprenticeship intact. The episode also lands in a 60–65 percent blue-collar workforce. Extreme heat changes time spent at work not working. Air conditioning in a machine shop is a retention variable, not a perk. Kevin lives that contrast personally: diesel truck, red Oklahoma dirt, 11 acres, pond, chickens, a log-cabin house next to the in-laws, and a history nerd’s dream of Trowbridge, England. Rapid-fire Cole’s Corner covers banking as the road not taken, Neo from The Matrix, and a three-year forecast in which agentic tools run basic analysis in the background while compensation, analytics, and core HR finally stop living in separate Excel dumps. Attract, reward, and retain only work if the numbers, the culture, the systems, and the next generation of talent are treated as one problem. This episode is the field notes. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Denise Teo, Former Head of Analytics & People Science at Grab! Denise spent nearly a decade at Grab, the Southeast Asia super app that began in 2012 to make rides safer where that was not a given. Today it is a hyper-local platform for rides, food, financial services, and B2B work. She joined in 2016 when Grab was still small, first in operations analytics, then commercial and product, before switching into people. She is not from HR or I-O psychology. She came from tech analytics and found the place where quantitative rigor meets the messy applied world of people. She built the function from zero to one, then rebranded it as People Science: stop stopping at insights and get better at effecting change through real partnerships. Denise is self-critical about the traps people analytics teams fall into. First, caring more about methods than outcomes. Complexity becomes a proxy for skill even when the business is not ready. Firms advertise the coolest methods, then drop people into sanitized work that never uses them. Cole argues there is little correlation between method sophistication and business value. Second, not knowing who the users are: employees, leadership, and the people team. User research should be easier because users sit in the same office, yet teams still skip discovery. Many analytics products exist only to paper over bad HCM interfaces. Third, failing to bring users along. Nascent functions outpace readiness; partners who are not data-native retreat to Excel when the model arrives. Southeast Asia makes the work harder and richer. Cultures and languages differ sharply. Qualitative text was nearly inaccessible before large language models unlocked an asset that used to sit unused. Denise sees the field growing fast in Asia. She simplifies the practice into thinking, doing, and communicating. The doing layer is what most people enjoy and what AI is hollowing out fastest. What remains is judgment about the right problems, design of guardrails, and translating insights for brains that do not think like yours. You move from the factory floor to the control tower. Cole maps this to his inquisitor-and-change-agent framing and adds orchestration of agents. Both worry that a thousand flowers blooming can scale noise as fast as signal. That shift hits identity. Denise is leaving full-time work at Grab to try fractional practice across more organizations. She describes the hollow feeling many analysts report: routines vanish, mastery of SQL or Python feels threatened, and purpose thins when the tasks that defined the job are no longer yours. Cognitive surrender is uncritical acceptance of AI outputs in place of your own reasoning. Cole contrasts people intelligence with workforce intelligence. The first treats AI as a tool that serves humans. The second risks putting people in service to the machine. Ethics and privacy have cooled even as data use expanded. Denise’s heuristics are symmetry, transparency matched by action, and consent plus agency. Value has to flow back to employees or the exchange stays extractive. Cole tests the principle at the limit with genetic data; Denise holds to those heuristics rather than a yes or no. In Cole’s Corner she would have been a fusion chef, most wants to see Iceland, and picks the zebra from Racing Stripes who believes he can race with thoroughbreds. On fractional leadership she says budget constraints already push organizations toward smaller units of exchange. They close on Andrew Merritt’s output-gap piece and John Sullivan on brilliant jerks. Denise is on LinkedIn. Hire her. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Roxanne Bisby Davis, Head of People Intelligence at Cisco! Cole Napper talks with Roxanne about a title they almost share, how people intelligence gets built inside a large technology company, and why trust has to sit at the center of listening and analytics. Roxanne has been at Cisco more than twenty-six years. She arrived in 2000 as a contingent worker, converted to full time later that year, and did not set out to run this function. She started in recruiting, was pulled into HR marketing because she knew PowerPoint, learned to simplify executive messages so people across the company could understand the intent, covered a leave and inherited early meeting surveys, and stayed as Cisco invested in listening, research, data science, and storytelling. People intelligence stuck as the name because the work outgrew an annual census or a narrow analytics shop. Her team now includes researchers, data scientists, and storytellers looking at the full depth of humans at work, not just the average. She keeps coming back to Todd Rose’s point that nobody is an average. Leaders need the range if decisions are going to serve more than the middle. Storytelling starts with listening more than speaking, asking questions that pull color and context, reading the room, and answering the “what’s in it for me” so people feel seen. Trust has followed her for years because it is so individualized and so fragile when teams and leaders keep moving. Psychological safety was a major finding in Cisco’s 2016 best-team research. Reciprocal trust matters: employees trusting the institution, and the institution extending trust back instead of treating listening as one-way extraction. She welcomes skeptics because skepticism sharpens transparency. Cisco tested survey load before a full census in 2006, ran annual surveys into the mid-2010s, then moved to quarterly plus continuous listening so they would not have to ask every day. Daily thumbs-up surveys can work in cultures that close the loop fast. In a company of Cisco’s size, the cost of attention can make daily asks feel like asking for asking’s sake. Cole compares that to Buffett’s weighing machine versus voting machine: a snapshot is not the same as a relationship that changes behavior. Roxanne also traces ITSG, the IT Survey Group that spun from the old Mayflower consortium so tech companies could share norms and get practical help without cold-calling peers. She still values consortiums that pull you out of your own echo chamber. She has always tried to make order out of chaos and treat messy organizational problems as pattern puzzles. If she were not doing this work she would be a professional organizer. She most wants to visit New Zealand and she would be Mary Poppins, walking around with a bag of tricks to connect people and make things better. Looking ahead, she argues the field only grows where capability is paired with radical transparency. New tools make surveillance cheap. That is why informed consent, employee choice, and the human premium matter more, not less. Psychology in the loop is not the same as a vague human-in-the-loop slogan. AI slop flattens people toward the average. Automation works when you design human-automation-human chains instead of sprinkling potential onto isolated tasks. Skills inference only works if employees keep voice and agency, because a skill you can do but hate will not produce engagement. People intelligence is synthesis across individual motivation, team conditions, process, and business outcomes, not a single data layer pretending to be the whole story. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.

Thanks to HRBench for powering this episode. To find out more about the company building the future of people intelligence, reach out to book a demo at hrbench.com/directionallycorrect! Check out this episode of the #1 people intelligence podcast with special guest, Jack Buckley, Chief People and Systems Officer at Roblox! In this wide-ranging conversation, host Cole Napper sits down with Jack to explore his unconventional journey from Navy nuclear engineer and tenured statistics professor to running a federal agency and now leading both the people and systems teams at the gaming platform company. They dig into what it really means to hold a dual CHRO and CIO role, with Jack arguing that success depends more on the right organization than a uniquely special person. At a mid-size, engineering-heavy firm like Roblox with around 3,500 employees, combining people processes with internal tools and enterprise systems creates powerful synergies that might not work elsewhere, especially where the systems side focuses less on pure technical leadership and more on how work actually gets done day to day. Jack shares how he arrived at Roblox through the rapid acquisition of his game-based assessment startup during the pandemic, just weeks before a planned sale to McKinsey, bringing deep expertise from redesigning the SAT at the College Board, leading research and psychometrics at the National Center for Education Statistics, and pioneering innovative measurement approaches at places like the American Institutes for Research. Listeners will hear about the frontier of game-based assessment for hiring early-career software engineers, product managers, and data scientists, including why explainable scoring rooted in evidence-centered design and item response theory still matters even as AI reshapes technical evaluations, and how the team moved existing technology onto the Roblox platform itself while maintaining rigorous, non-black-box methods. The discussion turns to AI enablement across the business, from safe agentic coding tools embedded in engineering workflows to thoughtful automation that avoids black-box risks in HR, legal, and finance, plus practical lessons on verifying outputs, building deterministic harnesses, and focusing on real workplace problems rather than hype or vendor pitches. Jack also reflects on building people analytics and employee listening from scratch when the company was far smaller, the value of succession planning that exposed him gradually to every part of the function, the necessity of humility when moving from specialized analytics into broader service leadership, and even using Roblox itself for multi-day new-hire orientation, game jams that teach both play and Studio creation, and cooperative team-building experiences. Along the way the conversation wanders into favorite pizza styles from New York to thin-crust Chicago tavern style, the sudden midlife appeal of birdwatching with AI feeders, cooperative games like Helldivers 2 that reward communication over competition, and the midlife questions facing people analytics as a field, including whether it is experiencing a true crisis of trailblazers versus button-pushers or simply needing new S-curves of applied problem-solving drawn from neighboring domains like labor economics and operations research. Packed with insight on measurement, talent strategy, technology enablement, and leadership that blends rigorous science with practical problem-solving, this episode captures the spirit that defines directionally correct people intelligence while showing how one career can connect nuclear engineering, educational testing, serious games, and the dual people-and-systems mandate at a major platform company. If you like this episode, you’d also love exploring prior episodes—visit colenapper.com for the full archive and show links.