What Finance Really Wants From Your Headcount Plan, According to Roku's Eric Lin
Podcast Overview
-
Eric Guidice Headcount Experts, episode 22. I'm Eric Guidice joined by Chris Mannion, very special guest, not just because he spells his name correctly, but a head of finance, from Roku. Eric Lin joins us for a conversation about the finance side of headcount, and we're so happy to have you as a part of the podcast. Welcome to the Headcount Experts.
Eric Lin
Yeah, thanks for having me.
Eric Guidice I love it. so Chris and I, we take this kind of dual-pronged approach. I'm very much in the weeds of headcount. Chris is an MIT MBA who's graced the headcount world with his supply chain, his military, and his operational background. So we take this, you know, I'm kinda asking the quick and dirty, and Chris is at the strategy level. So we'll kind of bounce between each one. But before we even get into kind of practical headcount finance, I wanted to learn a little bit about you and kind of you know, build the story of who is Eric Lin, how'd you get to where you are today, and learn about how headcount has played a part in that.
Eric Lin Yeah, absolutely. So I think my career has been shaped by very different business models. So I started in MA strategy at KPMG, where the job was to understand a business model under time pressure, the market, the customer, competitive dynamics, and whether the business plan actually held together. From there I was at Pacific Gas and Electric and it was a completely different environment, regulated utility. It was long cap cycle capital decisions. We did a lot of workforce planning where a hiring decision today might not translate into productive capacity for years. At Netflix I was in a mature but highly cash generative subscription business. And then at Google I led analytics in a scaled ads business where the issue was sustainable growth, how to deploy sales and support capacity against the right customers. And now at Roku I'm in a hybrid hardware and platform business. So when I think about headcount, you know, I don't just start with how many people do we have. We look at the business model, the work, the strategy that is required and the capabilities and how long it takes to make headcount actually productive.
Eric Guidice Yeah, I know Chris is already thinking of deep questions to ask about how you balance a headcount decision with kind of the ramp time and the productivity of headcount, which we will get into a little bit later. How far in the future do you plan headcount?
Eric Lin We generally look out three years when we do long term planning.
Eric Guidice As you're looking at a team of 4,000 employees and the size and scale that Roku is, what is the how dominant is the headcount topic in the financial planning conversation? Is it cyclical where you're doing it with reforecasts and annual planning, or is this an active part of the conversation? And if so, yeah, how does that play out day to day?
Eric Lin Yeah, yeah, no, it's active. I think it's something that we look at on a quarterly basis, just looking at you know, the actuals on what we plan versus you know what the forecast was and then looking out if any adjustments need to be made. So yeah, so I'd say it's a pretty ongoing part of the operating cadence right now.
Eric Guidice How do you make a decision about headcount on your own team? So how are you looking at adding a headcount or looking at total seats on your own finance team? Like what are the factors that play into how you decide to add or you know, backfill or expand headcount on your team?
Eric Lin Yeah, I think my philosophy is a good headcount discussion starts with three key questions. Like one is what work is truly required by the strategy? What capability is needed to do that work? And then when does that capability need to be productive? Because you know, obviously you hire someone day one, they're gonna take time to actually ramp up. So I think look at those three questions and then that's when I translate it into capacity and cost. So you know, look at the physical heads versus FTEs. I see a approved requisition as a permission to recruit, but it's not actual expense or current capacity yet until you know the person's actually on board and actually productive. So that's kinda how I think about it for my team.
Eric Guidice And how are you factoring in? So we're gonna you know, we have three sections of the podcast, right? This kind of part about you, the next will be kind of about the practical headcount finance topic, and then a third where we're gonna kind of look at some current events, some current topics. And one of the things, as a hint, a preview, is the financial engineer. And really, what it is is somebody on the finance team who's overseeing the AI initiative within the finance department. But as you balance headcount, agents, tokens, how does that play into the math, right? How does that play into the science of you figuring out whether or not a headcount can be productive? You know, the time to productivity, whether or not it is a headcount. How does the AI kind of developments impact your financial planning as it relates to like that math?
Eric Lin Yeah, so I think we're still in the early stages of that. But we're not at least as of yet making headcount decisions based on that. So I think we're seeing that people are becoming more productive and there's a lot more I guess use cases, but we're not seeing that, because we can deploy agents that we're expecting headcount to go down or stay flat, at least in the foreseeable future.
Eric Guidice Last question before I know Chris is gonna start to dive in on some of this stuff in more detail, but one of the more, you know, prominent debates, at least in my ecosystem of LinkedIn, is whether or not AI or an agent is a headcount or should be treated the same as a headcount. Do you have an opinion on that? Does that, you know, trigger you in one way or the other? But you know, do you consider an agent working on behalf of a person in any department a headcount or no?
Eric Lin At least right now I don't see that. I mean it could evolve into something like that as the agents become maybe more autonomous and you have people who are just directing swarms of agents. That might be one way to kind of structure, I guess, the financial analysis, but I don't think that we're at that point yet to separate it that way.
Eric Guidice Yeah, and it's an ongoing conversation. I know that Chris and I were just talking about this before you joined, is you know, we're as we use different parts of the AI ecosystem, like the coding is going great, but generating an image, I don't think I'm replacing my graphic designer anytime soon. But Chris, we have a kind of four topics that we want to dive into. And the first one's around the operating cadence. So you mentioned earlier that you look at it quarterly, and I'm sure there's a you know, and with a three-year kind of long-term planning. And I assume that there's a annual headcount planning or workforce planning exercise that happens. But I'm curious to know, how do you prepare for 2027? How do you reforecast for the end of H2? So Chris, kind of jump in there. What what do you want to know about the operating cadence from Eric?
Chris Mannion Yeah, I'm actually really interested in the transition Eric that you made from Energy into media and how in the energy space there's very long time horizons I have a background in aerospace engineering and you know we were looking at 30 year planning cycles to figure out how do we actually invest and develop things and I know it's very similar in energy but sitting over from that to stream in a media where the time horizons are much shorter. How has that thinking from that earlier experience actually influenced the approach that you take now and how you're actually planning that out where Roku is a, you have hardware and software. Like how are you thinking about the planning cycle and the importance of forecasting and getting accurate on those forecasts in order to make those decisions?
Eric Lin Yeah, so it was it was definitely like you said, very different. So when I was in energy, I think the important insight was some critical capabilities just had multi-year training and ramp requirements. So if you waited for the vacancy to appear in the budget, you're already too late. And so while I was there, we kind of discovered that we had this huge gap because you know, the grey tsunami where a bunch of people were gonna retire but we hadn't been recruiting enough in the pipeline to be able to replace that. And so our strategy was that hey, we can't just hire a bunch of apprentices that are going to take five years to train. We just had to go and poach from other competitors, experienced linemen, to be able to address that gap. And so that was a whole strategic effort that we had that we launched that was critical to making sure that the workforce would be able to match the needs of the strategy. So at you know in tech it's not you don't have necessarily that five year lead time for someone to being productive. But I think it's similar concepts though that you do need to keep in mind that you know you plug someone day one they're not gonna be able to be that effective right away. And so I think similar concepts but I guess you don't have to plan as far ahead as you do if the lead time is that long.
Chris Mannion That makes a lot of sense. We actually saw something similar. So, in aerospace, there was that tsunami of experienced aerospace engineers leaving the field. And so we actually had to start going to schools and educate school children on the importance of engineering in the UK as part of that program. As you went and got to Wayfair and we were looking at campus recruiting, we had the same problem with data scientists in that we were all trying to hire data scientists from PhD programs but every other tech company was trying to do the same thing. Are you seeing that now with AI within your space? And to what extent is this kind of need to recruit the same people turned into a financial and a capital allocation decision as you're thinking through? If we hire people from our competitors for an increased comp, and then they hire our people by increasing comp, there's almost kind of a race to the top there. How do you think about that from a capital allocation and maybe different strategies that you might take moving forward.
Eric Lin Yeah, it's an interesting question. At PG and E we also did something similar with trying to work with schools to develop the curriculum. But even then that's still, you know, really very long lead time to actually getting to the headcount that you need. I guess the way we think about AI, or at least the things that I notice, is that you really want the AI native people, who you know, they just they're just immersed in the tools, they know how to leverage everything. So that's gonna be more and more critical. I think as we make new hires, that's definitely something that that we're looking for, that someone who's already well versed in the tools. And so to the extent that colleges and other education institutions are providing that, I think that's what will need to have the supply because you're right that if everyone's just going after the same small subset of people then you're not gonna be able to afford the comp that results from that.
Eric Guidice What's your reaction to the vocational schools being put up by the big AI companies to try to speed track people with the skill set to build the data centers? Is this something that we'll see adopted? Is this a finance strategy that you think other companies of smaller scale would be able to adopt, or is this kind of for the big players to try to speed the skill adoption up?
Eric Lin Yeah, so I actually am not too familiar with that. So what are they doing with the vocational schools?
Eric Guidice So they're paying for folks to go through like a vocational school on the technology and skills needed to build data centers to fulfill the demand of building all of these data centers. So for the people shortage, they are you know sponsoring or giving folks scholarships to these vocational schools to learn those skills. And so that investment coming out of the company to go get the talent in an accelerated timeline versus you know. I'm sure energy and other things have that longer time horizon applies for a number of reasons, but with AI being this race, that investment is worth the while of these companies so they see it. But it's just kind of a grow at all cost strategy that I'm familiar with from my Uber days, but I don't know if you think that'll apply or if that will be extended or those folks will co sponsor things for other companies of smaller scale. Like I think it's something. I just don't know what it is.
Eric Lin Yeah, no, it's interesting. Yeah, I feel like smaller companies probably wouldn't be able to necessarily move the needle on that. I mean maybe there would be some consortium that different companies can invest into that can help. But yeah, I mean if that's what the hyperscalers are doing, I think that that's great because that definitely is a bottleneck, along with regulatory constraints with getting these things rolled out.
Eric Guidice Yeah, it's called I just popped it up. It's not I was gonna have this maybe on our other section, but it's called America's Workforce Academy, and it's five weeks and it guarantees a job on the other side, and it gives you the NCCER, the OSHA trade standards, and you can do electrical, HVAC, fiber optic, etc. And that's you know, they're doing it in Louisiana, Indiana, and Texas. So it's just a I thought it was a super interesting, you know, when Chris and I talk about workforce planning, there's two things which competing over limited supply, increasing the supply is one argument, and then the other is like speed to skill, speed to ramp. So, you know, creating net new headcount supply and then figuring out how to decrease the ramp time, which is another curiosity I have from your experience level, is like, you know, if you in your last you know year or two years, and then forward looking of workforce planning, has the date to productivity shrunk given the AI boom, right? You know, for Chris and I, it's if we're gonna hire a recruiter, we expect their first hire to be two months after they started, given the time to fill. But for other roles where you might be onboarding to knowledge or onboarding to something else, you know, is that number decreasing? Do you see it broadly decreasing in your world?
Eric Lin Yeah, I think so with I guess recent hires, because of the AI tools, like when you join a new organization, I think one part that just takes time to learn is like who do I need to talk to to get X done? And now it's just easy to just type it in, like, hey, who's the person in charge of X project? So I think that unlock definitely helps a lot in terms of getting people up to speed.
Eric Guidice Got it.
Chris Mannion I'm actually curious to dig into this whole topic a little bit more. Because one of the goals that we have on the podcast is to educate people in HR who are going into these headcount planning roles on how their stakeholders think. And the relationship with FP&A is so important. And from my experience, I came in from the business and my first CHRO had a finance background. And so everything was kind of towards that framework. But I hear a lot of frustration, especially from recruiting leaders who are trying to get investment upfront for a return that they might not see for 18 months. So I was wondering if you could talk a little bit about your decision-making process when you see something where a head of TA or HRBP is asking for an investment this quarter for that long-term horizon. Said you plan out three years, but tied to headcount, how do you actually work backwards from that? And what do you look for as you're looking through those business cases to make sure that it's a good decision for the business.
Eric Lin Yeah, I think you know what I've seen work best in my career is to make sure that there's a shared decision artifact for each meaningful role or role group. So it should state the work that's gonna be done, the outcome expected in the first year, the required capability, the date the person needs to be productive, and then you know, looking at the fully loaded costs and what are the alternatives. I think where I've seen it work is the business leader is accountable for the outcome. Finance can challenge assumptions and compare the role with other use of resources. I think the people team can challenge the organizational design, what the levels are right, and then recruiting should assess does that talent exist at the needed location and compensation and timing. And then the manager is the one who owns the interview process. So I think finance does administer the budget, but I don't think it's helpful that finance is just this police force for every hiring decision. I think it should really be shared in the way that I described.
Chris Mannion And what kind of advanced collaboration or kind of working relationship would you expect from that individual? I'm coming to you and I'm trying to get investment for a new pipeline that we want to open for talent, let's say opening an apprenticeship program like we just discussed where we're actually going to invest for the next two years. And then by year three, we're actually going to see an increase in this kind of critical pipeline of talent. What would you want to see from me? How early on would you want to see that? Like what is the data that is needed in order to support that decision?
Eric Lin Yeah, so I think in that scenario, I'd want to see, yeah, like what is the gap, workforce gap that we're trying to address with building this and I guess also pressure testing. Like is this solution that you're coming, is that the optimal solution? What other things are you considering? And putting that on paper so that everyone is clear on, you know, this is the rationale for why we were gonna do this. Those are some of the high level things that I'd want to see.
Chris Mannion Got it. So, the core business reason, framed as a business, kind of cost, and then all the alternatives that you've explored, and then why the recommendation is the one that you want to go with, as opposed to, let's wait three years and then hire people from the market and kind of show the breakdown of what that would actually cost in that time horizon.
Eric Lin Yeah, yeah, I think being able to quantify that and frame the business problem that way, it makes it a lot easier to green light a certain request.
Chris Mannion Got it. Sounds good.
Eric Guidice Do you ever has it ever been a time where you've you've declined something that you shouldn't have or approved something that you shouldn't have where it's like that framework did or didn't work? And like, what you know, for some of the finance folks who don't have your level of experience, they're gonna learn some of these things the hard way the first time. So if you could give preventative advice during this kind of like headcount finance workforce planning process, what are what are things to look out for? What are things that you know that smaller scale or that super scale up startup finance leader should be looking for? When folks are vying for these yeah, either longer term projects or competing over limited resources.
Eric Lin Yeah, you know, I think when I think back in my career what one mistake I made on headcount it was assuming that if the analysis is strong enough then the resource decision follows. But I think what I missed was a headcount plan creates implied promises. So leaders may want to protect a requisition because they're afraid that they'll never get it back. Recruiters have invested all this time building a pipeline. You know, teams may see a role as a validation of their importance. And so not that anyone's acting irrationally, but these are some of the incentives that are happening. And so I think the important thing is to establish the rules before any conflict happens. Like what makes a role committed versus contingent? What kind of evidence releases it? Does a vacancy automatically trigger a backfill or is it a fresh decision? And then can a leader release capacity early without being punished in the next planning cycle? Because if it's like, well, every annual planning you get your base workforce plus, you know, maybe an adjustment, then you're incentivized to try to hold on to all the heads that you can even though it may not be optimal for the company. And so when those rules are explicit, then the conversation becomes less political and more about the actual work.
Eric Guidice Yeah, I was listening to a diary of a CEO podcast the other day, and I forget the guest, but they were talking about how like the military will just dump gas in the ocean because if they don't, they're not going to get the budget for it next year. And I feel like somebody somewhere in some company has held on to a role when they probably could have given it back, or vice versa. So it's a very common challenge. And you know, there's two things that like I've tried to develop in my career, and one is like the impact of variance. So I'm very curious to get your take on how you look at the evolution of a headcount plan over the course of the year and what information do you take out of what did we think was gonna happen versus what actually happened. So you know variance tracking and what's important to you, what's not important to you and why is it important. So that's kind of one bucket. The next bucket is kind of the idea of that the same thing that you're talking about is that empire building or how do you how do you control that variance and what should folks be putting in place to have some sort of check and balance against, whether or not someone's straying too far away from what you intended to happen and what's the consequence? So long question, two part of the first is about variance. What do you look for and why is it important to you?
Eric Lin Yeah, I think I guess for variance I'm just looking at okay what the plan was and then what actually was achieved and though I think what I look at is yeah what's that delta and then how does that influence what we should be forecasting for the next quarter and beyond. Because if certain department is always under hiring their target, then I think we can maybe assume that that's gonna continue so that we can free up resources otherwise you definitely see that you just have this chronic under hiring situation and you're not actually deploying the resources that you're hoping to.
Eric Guidice Well quick follow up there? Do you view under hiring as a savings or a loss of production?
Eric Lin I see it as a loss of production.
Eric Guidice Okay. I think that's like the most common question we'll get from our finance customers is like how to quantify a missed hire. And so when you look at plan versus reality, there's the missed hiring aspect and there's this 50-50, I would say fifty-fifty. Like some of our finance leaders are like, we intended for that to happen so that we can have a budget savings without having to necessarily restrict anybody directly. And others are like, no, we wanted to hit a goal, we missed it. The second part of the question was really about like corrective action from headcount variance right? Like what stands at like what do you need to correct if you start to see variance happening quarter to quarter? And what's a priority for you?
Eric Lin Yeah, the revenue producing roles. I think if we have specific initiatives that we have that require these headcount, and we aren't able to hire the people we need that'll kinda delay everything. And so you wanna kinda look at those vacancies as being top priority.
Eric Guidice Chris, what you think? I know that there's some questions around your mind around you know, when we talked when we were prepping for the podcast, about how to plan for scenarios like this, in conjunction with a finance leader. What's on your mind as far as you know, looking at variance and then planning for it? Especially, you know, last week we just recorded our best practices podcast just Chris and I about how to do an H2 reforecast. A lot of our customers and also companies in our ecosystem are kind of reforecasting for the end of the year. What's on your mind, Chris about the scenario planning side of things?
Chris Mannion So the debate and I think building off the point around is a missed hire, productivity gain, productivity loss or a profit gain. A lot of things that I'm hearing from people I've been working with are the kind of frame of dynamic resource allocation. So I'm actually curious, kind of leaning more into whether variance is kind of a good thing or a bad thing in that respect how you're going to be accounting for variance in a world where productivity gains becomes more of a focus, especially at sea level.
Eric Lin Yeah, so I think you know, one way to be proactive about these variances is to make sure you have good different scenarios that you've run. And so I'd say that there should be like a demand upside scenario for your business where a trigger might be to release a pre-approved group of roles and accelerate recruiting. If it's a downside case, then you know you have your agreed upon actions to pause contingent roles, reduce contractor usage. And so I think having those scenarios up front can help I guess inform the business on what the plan is if any variances happen.
Eric Guidice This next section of the podcast is a couple of LinkedIn posts that I've seen trending. One, I think I met you through, where here's the latest that we've heard in the market about headcount. We want to get your live reaction. So I'm gonna share my screen, I'll get a couple going, and we can hear from your perspective on some of this stuff. The first, coming from Felipe. He's a finance manager at music.ai. I think this is where you and I met. He's talking about AI doing things faster, but potentially not as accurate. And that being okay. What's your thoughts here? How is AI and the scenario planning going to influence the future of work for finance leaders?
Eric Lin Yeah, so I think my initial reaction to that is that it needs at least for finance it needs to be accurate depending on what thing you're talking about. So there's certain financial processes where you can't afford to just be eighty percent correct even though it's faster. I think so depending on what it is, I think I'd answer that question differently.
Eric Guidice I yeah, I think what we're seeing in our market, so we port headcount data into whatever Claude model or chat GPT that you want. And what people are using our data for is this combined forecast with other data across their system, and they're just getting that first pass so that they can start a conversation around it, then bringing it to their Pigment, their Anaplan, whatever, to then build out the final scenario. But being able to rapidly experiment with finance data has been a plus. My point in that article when I was interacting with Felipe is like, you know, when I was doing headcount analysis in my internal career, I didn't have the same luxury. Like a guess couldn't really be a guess without it being a little bit more a little bit more correct. Chris, any thoughts there?
Chris Mannion Yeah, I've actually just released a course on exactly this thing for HR because what I found was that we were doing a lot of Claude for Excel work and generating capacity models and headcount plans and things like that. But there are a lot of errors baked in and they were hard to find. So I've actually flipped it and said, well, what if we just teach people how to do the basic model in themselves, then use AI for enhancements or to check the work. So it still accelerates your process, but you're not taking the AI output and then just doing a quick check and moving on because I think to Eric's point, quite often you can't accept something being 80% right, 90% right. It's gotta be 100% correct, especially if you're dealing with people. So yeah, I support both your points.
Eric Guidice All right, next one. Adam Horn, he's a longtime people leader who's been influential in a few different parts of the talent space, leading open org. He says he's looking at org design and seeing four models. So his four models are startup within a startup, and he quotes PostHog. Maximum leverage, zero management. This is the Cursor, which obviously had a lovely exit as of recent, depending on how you think an all-stock deal will work out. And then the organize like a slime wall, the Perplexity. So smaller teams that work where the capacity and the interest is. So is there a shift? I guess is the broad question. I know there's a lot to digest on a reaction post, but is there a shift in org design now that we're seeing a shift in the way work and productivity gets done? Do you foresee the org chart looking the same in the future? Eric Lin, what's your reaction?
Eric Lin Right. My reaction is that yeah, I think it probably would change. If we do have all these autonomous agents that are capable, then I would see the org changing. Not sure exactly you know, which one of those four scenarios would be the one, but yeah, I think this technology is changing so rapidly, I think it wouldn't be too difficult to envision the org needing to be a lot different. Yeah.
Eric Guidice Chris, what are you seeing? New orgs? Same orgs? Same org, different story?
Chris Mannion I don't know. So I read all of those and it just reminds me of like two pizza teams, like the Amazon framework or what we had in the military, like unit command, where you just have small teams that can operate independently. Not a big company is in a successful big company. It's just a number of those. But I feel like all design follows the culture of the organization. If you can trust people to build products individually, then you have this very like unstructured organization, which has kind of small teams. I've not seen a big company at scale have no management. I don't know if that actually works. So I'm very skeptical. I'm certainly not an early adopter of these new processes. I think the two pizza team, the unit command, the like small team approach is optimal for most situations. I wouldn't change it.
Eric Guidice All right, next up here. There's two in one. Parker Gilbert, the co-founder at Numeric, is reposting essentially this finance engineer job posting from Spring Health and another post very similar. I'm seeing a few of these come across. Alana Misco. Sorry if I'm butchering that pronunciation, says the most valuable person in finance will be the finance engineer and it's already happening as a job title. I have seen a few of these out there. A16Z, I did the same thing with this with talent engineer maybe three months ago, and now they have a cohort, right? Of the next group of builders. So the finance engineer seems to be the new target. Eric Lin, what's your reaction? Are you hiring a finance engineer?
Eric Lin Yeah, well did they have the job description of what that actually is?
Eric Guidice Yeah, those the sprint let me see if I could pull up the Spring Health the job description, but my presumption is it's anything finance ops and finance business partner would do, but kind of rooted in AI and having as much automation and integration as possible as kind of like the finance representative of how AI is used within a business.
Eric Lin Okay. Well yeah, I mean, with that description, yeah, I can see definitely a use for someone like that. I mean that's something we're trying to lean heavily into AI and getting everyone comfortable around using Claude Code and Claude cowork. But there's varying levels of expertise and comfort within the organization. So if there is someone that comes in and basically makes sure that the tools can work for finance, then I think that that would be pretty useful.
Eric Guidice I mean, just by reading the job description, looks like with four years of experience you can earn yourself a nice salary so long as you familiarized yourself with some basic AI coding and automation. So you know, for the will AI create or eliminate jobs? Plus one in the create category. Who knows what happens on the other side of this as it relates to jobs staying or going. But looks like we have time for one more. Let me see if I can find one of the best ones from my most recent. Here we go. I like this one a lot. So this is a post from Helen Russell. She's the chief people officer at HubSpot. And she's saying, the most squeezed role in your organization is the manager. So with AI doing more with less, we're asking managers to take on more. That means more people per manager, more work per manager, creating potentially this idea of some overwhelmed, underserved employee base by way of this manager role that's being abused. I don't know if I'm maybe putting words in her mouth there, but you know, this person still has to deal with the people aspect, but at the same time has to increase their productivity with AI. So what's your reaction? Is the manager the most squeezed role in the organization? Eric Lin, take it away.
Eric Lin Yeah, that's a good question. I don't know I feel like AI would empower them to be more productive is the hope. I don't know why it would make them more squeezed. I guess if you load more direct reports under them, and that's the expectation then perhaps it could be, but yeah, at least my initial reaction was that I don't know if I would agree that that is where the squeeze would happen.
Eric Guidice Chris, what's your thoughts?
Chris Mannion Yeah, there's a lot of talk about this at the WorkCube and Live conference a few weeks ago and how the middle management layer has been really kind of thinned out with all the rifts over the last few years because that's an easy layer to say, okay, if we cut the management layer by a third, we can just increase everyone's span of control and that's not really going have much of an impact. And it doesn't until it does and then the managers leave. And meanwhile, the people that are reporting into them don't want to move up to that position because they see how stressful it is. And there's this like middle manager crisis thing that has been kind of called out. And I think that the focus now is on, well, how can we use AI to mitigate the effects of that squeeze? I don't think it's been planned that way as in, we're going to get more efficiencies because of AI and so we can reduce the manager population. It's a cost saving measure and we also have AI, but nobody's really trying to piece them together, I think in a strategic way. So I think it's a huge problem. I think it's a long way away from being solved. And I'm very curious to see how that plays out. From what I understand, Gen Z doesn't want to move into manager positions. And I'm not surprised there's someone who's a millennial manager here.
Eric Guidice You're picking a fight there, I like it. Yeah, no, I think look, you know, middle management getting stretched and if AI makes more productive, maybe in the moment it is productive, but what I've seen from close friends going through management changes now is like when that middle manager leaves and you have a gap, that's a real that's tough coverage. Anyway, more to come on all things headcount, headcount experts episode 23. Next week we'll be covering some things related to the future of H2 and H2 planning, among some other new analytics that are coming out within Headcount. Eric Lin, it was amazing to have you on the podcast. You have a level of experience that we're very excited to kind of dig into and appreciate you entertaining us with the reactions to the content so thank you so much for joining the podcast follow Eric on LinkedIn any plug any career site any place you want to send any listeners for more engagement or benefit to Roku
Eric Lin No, no, I think yeah, you can find me on LinkedIn. Definitely always happy to chat with people doing interesting things.
Eric Guidice Alright, appreciate you giving me the look after I think I reached out to you on Felipe's post. I think that was the one. And so I really appreciate you giving us a look and coming on the podcast. It was great to have you on. Chris, any final thoughts?
Chris Mannion No, this was really helpful. I hope that a lot of people that are listening picked up lot of things from Eric and reframing how they think about the headcount process when working with FP&A. So thank you so much, Eric. It's been awesome.
Eric Lin
Thanks for having me.
A Framework for Practical Headcount Finance
Headcount is the largest line item in most operating budgets. It is also the one where finance, talent, workforce planning, and business systems teams most often talk past each other. In episode 22 of the Headcount Experts Podcast, Eric Guidice and Chris Mannion sit down with Eric Lin, head of finance at Roku, to close that gap.
Eric's resume reads like a tour of every headcount planning environment that exists. M&A strategy at KPMG. Multi-decade capital cycles at PG&E. A cash-generative subscription business at Netflix. Scaled ads analytics at Google. Now a hybrid hardware and platform business at Roku with roughly 4,000 employees and a three-year planning horizon. The result is a conversation that translates headcount planning into the language finance actually uses to approve, defend, and cut roles. We will not spoil the full framework here, but these are the themes worth listening for.
According to Eric Lin, every headcount decision starts with three questions
Eric does not start with how many people a team has. He starts with the work.
What work is truly required by the strategy? The role exists to unlock an outcome, not to fill a seat, and the business case has to state that outcome up front.
What capability is needed to do that work? Level, skill, and location decisions flow from the work itself, not from what the last org chart looked like.
When does that capability need to be productive? An approved requisition is permission to recruit, not current capacity. Ramp time is part of the math, and finance is counting it even when hiring teams are not.
Headcount plans create implied promises
The most candid moment of the episode comes when Eric describes a mistake from his own career: assuming a strong analysis automatically wins the resource decision.
Leaders protect requisitions they fear losing. If giving back a role means never getting it back, every leader is incentivized to hoard heads, even when it hurts the company.
The rules must exist before the conflict does. What makes a role committed versus contingent? Does a vacancy automatically trigger a backfill, or is it a fresh decision? Explicit answers keep planning analytical instead of political.
A shared decision artifact keeps every function accountable. The business owns the outcome, finance challenges the assumptions, the people team pressure-tests the org design, and recruiting validates that the talent exists at the target compensation and timeline.
Variance is a forecasting signal, not a scorecard
Asked directly whether under-hiring is a savings or a loss, Eric does not hesitate: it is a loss of production.
Plan versus actual should update the forecast. A department that chronically under-hires is holding resources the business could deploy elsewhere.
Revenue-producing vacancies get corrected first. When variance appears, roles tied to committed initiatives take priority over everything else.
Scenario triggers beat mid-year scrambles. Pre-approved upside and downside scenarios, with agreed actions attached, turn a reforecast from a crisis into a checklist.
Featured Headcount Content Creators
Filipe Vizentim | Finance Manager - Music.ai | AI Hallucinates, and I don’t care
An FP&A leader argues that AI hallucination is an acceptable tradeoff because AI now does the invisible, time-consuming modeling work in seconds, and reviewing the output the same way he always reviewed everything gives finance leaders their time and strategic influence back.
Adam Horne | Co-Founder - Open Org | Future Org Design, 4 Models
Four emerging org design models, PostHog's startup within a startup, Rippling's compound startup, Cursor and Linear's zero-management leverage, and Perplexity's slime mould structure, share three common themes: talent over headcount, management as a last resort, and shipped product as the performance framework.
Parker Gilbert | Co-Founder, Numeric | The Finance Engineer
The finance engineer is less a new job than a new capability layer added to every existing finance role, combining the skillset to automate closes, forecasts, and workflows with AI, and the mindset to treat the finance function as a machine to build while keeping the final sign-off human.
Helen Russell | Chief People Officer, Hubspot | Managers are the most squeezed role
AI is compounding rather than relieving pressure on managers, who face rising output expectations from above and overwhelmed employees from below, suggesting the role needs a fundamental redesign toward less IC work, more people orchestration, and larger teams.
Bring finance and talent into the same headcount conversation with headcount365
Every framework in this episode depends on one thing: finance, talent, and business leaders working from the same headcount data. That is exactly what headcount365 provides, a single source of truth connecting headcount plans, variance, and recruiting capacity in one place. Listen to the full episode, then see how headcount365 turns Eric's frameworks into your operating cadence.