Why Improving Productivity Starts With Understanding The Work

David Rice: When a C-suite leader looks across the organization, what they see is an org chart and a process doc. Role A finishes something, hands it to role B, B hands it to C. It's clean, simple, and almost entirely fictional. On today's show, I'm talking with Heidi Farris, CEO at ActivTrak, about the gap between how leaders think work happens and how it actually happens.

The real workflow is the sum of the handoffs, the workarounds, the informal rerouting that never makes it into any document. Heidi calls the good version of this tribal knowledge. It's not the I-know-a-lot-about-the-business kind, but the small daily adaptations that keep things functioning. People make work better every single day is what Heidi says.

The problem is it doesn't really show up in documentation. So today we're covering why the documented process rarely matches how the work actually gets done, the three buckets for deciding where AI belongs in a role, what happens to organizational throughput when one team accelerates and its neighbors don't, and the real signs that a business has moved from AI adoption to AI impact.

I'm David Rice. This is People Managing People, and if your organization is measuring AI success by usage metrics instead of what happened to the capacity that it created, this conversation shows you what you're missing. So let's go.

All right, Heidi, welcome to the show. It's good to have you.

Heidi Farris: Thanks, David. I'm very happy to be here.

David Rice: Well, where I wanted to start our conversation was, you know, when a C-suite leader looks across the organization today, I think one of-- This is one of the common frustrations a lot of people have with leadership at work, right? It feels like they don't really know what goes into the work.

And I'm curious, you know, like, when a C-suite leader looks across the organization, what do they typically misunderstand about how work actually gets done?

Heidi Farris: I think it's what a lot of organizations are struggling with today as you look across the landscape and everyone's talking about redesigning work or applying AI to work today, and what leaders see is an org chart and a process doc, right?

And, you know, role A finishes something, they pass it to role B, and then B to C, and, you know, it's supposed to be a nice clean handoff like that. And we tend to oversimplify the way work is actually getting done in organizations. You know, an individual in an organization can easily be navigating, you know, dozens of apps, multiple approval processes, many different stakeholders and workarounds in any given day.

And the wor-real workflow is the sum of the handoffs, the workarounds, the informal rerouting that people deal with that doesn't show up in a process document or on an org chart. And it's really very-- it's invisible, you know, to the top, and, and we get disconnected from how work really gets done pretty quickly in an organization.

David Rice: Yeah. It's, it's-- I think it's kinda natural, right? As you, as you move up the chain, things get reduced to charts and, you know, SOPs and dashboards. I think there's an int-in-interesting distinction between visibility into employees and then visibility into work. They're not really the same thing, right?

Heidi Farris: That's right. That's exactly right. The work itself, you know, is what we care about, right? I mean, look, I think employees instinctively are alway- you know, on the, on-- largely are always putting their best effort forward, right? It, it, it's not... It, it's more about, like, how do we help improve work across an organization or understand how work is happening so we can then improve work or the humans that, you know, can apply judgment and empathy to the work that AI can't do.

David Rice: A-and if you're trying to decide where AI should change the way the work is getting done, I'm curious what do you need to understand about the work first? What, what's really the, the high-value stuff that's gonna help shape that?

Heidi Farris: You need to understand the tasks that are being done by each role, right?

And be able to bucket them in a way, because jobs are collections of tasks, and AI transforms work by changing tasks, not necessarily by replacing entire roles, right? There's certain parts of every role that AI may be beneficial to. And so the first thing is if you can sort of categorize the tasks that any given role does into repetitive, structured, and rule-based tasks, which are ripe for AI, AI-assisted tasks that require, you know, rapid data synthesis or pattern matching or something like that, where AI serves as a co-pilot to a human, where there's collaboration happening, and then uniquely human tasks, right, which require empathy and emotional intelligence and judgment.

That's the first step, and most people just skip over that step entirely and just throw technology at every problem.

David Rice: That's interesting you say that, 'cause I, I was thinking about this earlier, having seen a clip of somebody talking about empathy being the most important thing moving forward, and I think it's something that we haven't taught particularly well.

I think we've often thought of it as a trait rather than a skill. One danger with AI is that it's, you know, its usefulness at the task level can sort of distract from asking whether the task should exist in its current form, and there are really different levels of intervention, right? You can improve the task, which I think is what we've spent a lot of time doing.

You can remove the task maybe, or you can redesign the workflow. You can reconsider the outcome that you're trying to produce. There's a lot of different things here. But I think there's a difference between asking "Can AI do this?" and asking, "Why does anyone have to do this?" And that second one feels more consequential, I guess.

So yeah, the easiest things to automate, in a lot of cases, they're not necessarily the most valuable things to change, right?

Heidi Farris: That's right. I think any time you do an audit... I mean, AI is shining a light on work right now, but this is-- what you're saying has kind of always existed, right? We always-- over time, we accumulate work debt the same way that a tech stack or a, a product accumulates tech debt.

And there are just certain things, to your point, that at some point you have to reevaluate and say, "Why do we even do this anymore?" Right? And doing a full audit, and then understanding, again, the different types of work and how you make them better. Because not everything, again, like I said, is an AI.

You know, AI is not the solution for, for every problem we see with work. Sometimes it's redesigning the workflow, sometimes it's better enablement for people, sometimes it's the wrong, wrong technology altogether, right? And you have to really understand how work is happening and what tasks are being done to then be able to kind of pull apart the work and say, "Okay, this is how we actually improve this."

And it may be AI, and it may be something else.

David Rice: Let's say AI gives a team 15% of its capacity back, right? We've, we've seen a lot of productivity gains, and we've seen in some cases where it has made admin stuff easier, things like that. But I'm curious is if people are getting capacity back what decision has to happen next?

What's sort of the-- I've been asking a lot of leaders this lately what's your first step in redeploying that capacity with business value in mind? 'Cause if you leave it to employees, it could get pretty interesting.

Heidi Farris: You know, I think the first thing, you have to empower someone in your organization to be a decision maker.

And what I mean by that is, like you said, capacity doesn't redeploy itself. So if no one's explicitly accountable for it, it just quietly gets absorbed- Right? Back into the organization, maybe low-value work, maybe disap- appears into Slack somewhere. You know, like I said, there's that list of work debt that has accumulated over time, and that may not be the most valuable work you want people just defaulting back to.

And the second is deciding, again, which buckets it goes into, right? Is it, is, is, is our goal cost reduction, right? So do we actually want to shrink the function? Is it reinvestment? Do we wanna redirect freed hours back into higher value work, or is it expansion? Do we wanna actually take on an entirely new project or entirely new work in an organization that's never existed before?

Each one has a different financial outcome. Some you tie back to EBITDA, some you tie to growth some you tie to differentiating yourself in a market. And, you know, if you're only celebrating the adoption metric and never check which bucket that capacity landed in, you're never gonna be able to draw a line to the P&L to say, "This is how we did it."

Because at the end of the day, making work better is one half of the equation, but the other half is showing where that res- results in your, in your financial profile for, for an organization.

David Rice: Yeah, and I, and I couldn't agree with more with what you said there about it becoming low-value work. I was actually talking with a leader recently in a, a different media organization.

We were talking about, you know, like some of the, some things getting freed up, and they had somebody who they were really not impressed with some of the ideas that they had brought to the table, and they had started to use AI in their sort of creative planning. And that was freeing up a lot of time, but this person was then being asked to put it all in these spreadsheets, different spreadsheets, and to just manage-- The time is absorbed by managing spreadsheets.

And I'm like, is that a good use of the capacity that was created? 'Cause I think the AI could probably manage the spreadsheets too. But it doesn't automatically mean you don't need that person anymore. You know? I think that's the part of the narrative that we get lost in too much is headcount reduction and...

But there's a lot of other things that it could become. I don't know. It seems like there's maybe a missing management decision between we save time and we created value. Somebody has to decide what that time is for. And I guess my question, my follow-up question here would be is should leaders be deciding how capacity's gonna get redeployed before they automate?

Automate to create that capacity so that you can do X.

Heidi Farris: Yeah, because you should start with your goals, right? Because if your goal is EBITDA to eliminate roles, then you should understand where the biggest opportunity for AI automation exists in your organization first, and start there. As opposed to if you have a bunch of roles in an organization where 80% of their time is work or tasks, made up of tasks again, that require either co-pilot, you know, AI as a co-pilot, or a lot of human judgment, you're not gonna eliminate those roles.

And if you do, you're gonna be hiring them back because you realized, oh no, you know, human judgment, empathy, and ability to kind of pressure test what AI is telling me is important, not just I'm gonna automate something that is already a really well-defined data-driven process. Those are very distinct things, and if a role's made up 80%, you know, of those other things, you're not gonna get the EBITDA you want anyway.

So understanding what your goal is, to your point, understanding what your goal is before you start is very important. Otherwise, you could do a lot of work and, and never actually hit your goal.

David Rice: Yeah, absolutely. Okay, so AI adoption and AI value, right? That's-- One doesn't automatically lead to the other. But I'm curious what do, what do you think leaders should actually measure to know whether AI is creating meaningful capacity- Yeah

and whether that capacity translates into better business outcomes?

Heidi Farris: I mean, the, the important part is you, you kind of mentioned it, but, you know, usage metrics measure adoption, right? Just plain usage metrics, measure adoption. But you need a before and after measure, right? You need to understand what did cycle time throughput, you know, you name it, look like before I implemented a fix, and that could be AI, right?

Is what we're kind of centered on today. And then what did the work look like after, right? Because without the before and after picture, it's really hard to assess, well, how did the new thing that I put in this process actually affect how work is happening? Did it improve capacity? Did it improve throughput?

Where did I gain something, right? And and then where that capacity actually went is what, you know, almost nobody, right, tracks today or looks at today. And if you can't answer what is a team doing, you know, with the time that AI gave back, you don't have an ROI story, you just have an adoption story. We just start...

We just spent money and started using more tools, but we didn't actually tie it to anything meaningful in the company.

David Rice: Yeah, you just said right there what is the team doing? And I think that's gotta be the focus. Individual productivity, right? It's very seductive because it's kind of easy to demonstrate oh, this used to take me 40 minutes, now it takes me 10.

But does the org actually benefit from one person's process getting a little bit faster, right? You can become 30% more productive without the organization becoming 30% more productive, and if I finish my work twice as fast, but then it still sits in somebody else's queue as a bottleneck for three days did we actually improve anything?

So yeah, there's a useful distinction between task efficiency, workflow throughput, and business outcome. I'm, I'm curious, like, why is there often such a gap, though, between how leaders think a workflow operates and, and how it actually gets done? I know we kind of alluded it to it at the beginning, but in your experience and what you've seen, you know, kind of take me through it a little bit.

Heidi Farris: Yeah. I think, you know, aside from what we've already talked about, the process doc is a snapshot in time, and work is a moving target It's, work is always changing, right? And so a document, you, you document a process. You know, we, we say internally a lot, a process document rarely survives the workforce.

And just what all that means is it doesn't mean that people aren't doing the right things. It just means that in application, a process document doesn't show all the nuance. People closest to the work make dozens of small adaptations to work every single week, right? To make it better. And if those adaptations very seldom make it back into any kind of document, and if you start to multiply that across hundreds or thousands of roles and many, many years, the gap between the documented process and the actual workflow widens very quickly, and leaders need a clear, continuous view of how work gets done so that they can be agile in the way they manage organizations in the AI era.

David Rice: Yeah, I can't think of anywhere I've ever worked where we didn't have basically a workaround, a repeatable workaround that we created as employees. You know, certain, the formal process doesn't match the on-the-ground reality. I think one of the risks that we have with AI is we, we try to automate the official workflow, and we don't actually understand the workaround that m- possibly...

In some cases, it accidentally made the thing function. It wasn't part of the initial plan, but if we start to remove those things, you know, h- how much of the process breaks is what was the question we have to answer.

Heidi Farris: Yeah. What we don't give enough credit to, I think, as leaders, is that people make work better every single day because they think, you know, they're using reasoning, all of these things, and there's a, you know, you, you go, you're clicking through a process document, you know, as a new hire.

That's, that's what you get. You get a process document. You're clicking through, like, all the steps you're supposed to click through, and you run into something, and you're like, "Oh, no." And of course, the person, you know, sitting next to you or over the cube or on Slack or whatever is like, "Oh yeah, you have to, you know, you have to hold the the shift button down and go to Excel and copy it from here and paste it over there, and then the whole thing will work again, right?

And we don't give enough credit to the fact that humans figure those things out, and they don't exist anywhere in document. And people call that tribal knowledge, right? But I think a lot of times when people think about tribal knowledge, they're thinking more about "I know a lot about work intelligence.

That's my tribal knowledge." And that's not the tribal knowledge that really makes an organization function. The tribal knowledge that makes an organization function is all those little nuances and workarounds and how I have to make something work every single day so the business can keep operating.

David Rice: So, you know, that leads to another question, which is what happens to workforce planning when AI changes the capacity of different functions?

And then we're seeing this at different rates, right? So, for an analyst, it's gonna be a lot different than somebody in marketing or somebody in accounting maybe. There's all these different fields that are now applying the same technology, but it doesn't always have the same effect across the board.

So what do we have to account for when we think about workforce planning?

Heidi Farris: We actually had a really great example of this internally for our own teams, which is our engineering team adopted AI coding tools and found a way to be very impactful doing it. You know, they cut our cycle times, they accelerated the commits, right?

Our engineering throughput you know, more than doubled. And, and that was a win, and we all started to celebrate, and then we quickly realized that now our engineers were able to code faster than our PM team was able to re- develop requirements for them, right? And so we just shifted the bottleneck from one function in our group to another function, and so as an organization, we didn't actually improve overall through- throughput.

We did temporarily 'cause they were able to get the requirements that were on their plate already done quicker, but then getting new requirements to keep up with how fast they were able to now develop code was, was a problem. And so this is where organizations have to look holistically across an organization and understand that all work in an organization connects in some way, and if one department get, you know, accelerates throughput or accelerates, you know, or decreases cycle time or something, their neighboring, right, their neighboring teams are gonna have to adjust to be able to keep up with that change as well.

David Rice: Yeah, that sounds like a prime example of the, the workforce planning issue is it's, it's not about how many people do we need, it's where's the constraint gonna move? 'Cause it's changing all the time, and it, the isolated productivity gains, like you mentioned there, you move the bottleneck. That, or you can create new ones and s- rather than eliminating old ones.

Heidi Farris: And look, some of that's gonna happen because you can't do everything at once But this is why there is a sequence to how you should think about things, right? When you-- This should really be a part of workforce planning to the original question, which is, you should sit back and think about which part of this should I accelerate first, and how long is that gonna buy me until I'm gonna run into a bottleneck somewhere else?

And being able to have that continuous view of work over a whole organization allows you to understand where those ebbs and flows are gonna happen in an organization.

David Rice: When you find a workflow that's consuming more capacity than maybe we think that it should, right, how do you determine whether the answer is AI or process redesign or, you know, some other technology or just simply better enablement?

Heidi Farris: Yeah, I think the question that leaders have to answer first is what's actually causing the drag. You know, if the work is repetitive and well-defined but high volume, right, that's, that's a automation or AI candidate, right? But if the sequence itself is wrong, right, too many approvals, unnecessary handoffs, duplicate steps, that's a process redesign problem, right, that you have to go in and fix.

And if the process and the tools are fine, but the people don't know the best way to do the work or they lack the skill to do the work, then that's an enablement problem, right? And so the fix has to match the actual cause of why the work isn't as efficient or as good as you'd like it to be. And most organizations default to, again, just buy more tech, you know, without first checking on which of those three things is it actually so I can apply the right solution to it.

David Rice: Yeah, it, it seems like maybe there's a little bit of an incentive problem here. C- we've allocated AI budgets, and executives have made these AI commitments, told everybody to be AI first, right? So then identifying something as an AI use case becomes a little bit more attractive than identifying it as a boring management or process problem.

I think once you decide to go AI, like you say AI is strategically important to us, I wonder if there's a danger that every inefficiency suddenly starts looking like an AI opportunity when really it's, you know, recur-- like you mentioned some of them there, but like a recurring manual task just that might need automation, but like an inconsistent task that just needs clearer standards.

That's right ... or, you know, a confusing task needs training. We've gotta think about it in these different buckets.

Heidi Farris: I think the way we, we came out of the gate strong with AI, and AI's still going strong, and it's, it is a sea change. I'm not AI is very valuable. It's gonna change the way we all work.

I, th- there is no part of me that believes anything different than that at all, right? And I think that initially for change management, it's very hard unless you kinda go all in on something. So I completely understand why organizations had a big AI push, because to change the habits and the way people work, and to get them to think out of the box, you have to kind of do something, you know, kind of over the top sometimes to get the change rolling in an organization.

But now I think the change is well rolling, because now you're seeing, you know, people like, "Oh, we fired too quickly. We need to hire them back," or, "We told to- we told everyone, you know, token, token maxing gets you to the top of a leaderboard, and you get bonuses for that," and now they're putting limits on spend, right?

And so now there has to be a more thoughtful approach long term to how AI is used in these organizations, and I think a framework that kind of understands roles, tasks, human, human-assisted, AI only. Like, where does AI belong, and how are you gonna get the long-term benefits out of AI now that you've convinced everyone AI is a way to go, and you have people using it?

Now what are you gonna do?

David Rice: I mean, it's it's almost like a game of is it Sims, where you build the city, you know? It's like your first city was like a, you know, a very clean map, but I think now it's gotta be, you've gotta build it out into a very detailed much more intricate and sophisticated city, right?

Your understanding of what w- what's going on in work. You mentioned it there. We're, we're well into it, right? We're rolling along with it now, and it's changing workflows across functions. You know, HR traditionally owned jobs, talent systems, workforce planning. Operations is very much processes, IT, technology, finance, the money.

We-- the whole structure, right? But as it starts changing workflows across those functions, I'm curious who has to be accountable for just redesigning the work itself? Because if everybody's doing it very differently, I, I don't know that it syncs up the way that it used to.

Heidi Farris: You know, redesigning the work itself needs an owner with-- that has two things most AI initiatives lack, I think, today, right?

One is visibility across the whole workflow, as we've been talking about, not just one function's piece so you understand where those ebbs and flows are and where work connects. And then the actual authority to change how work is structured, right? Because without the second part, you're sort of hamstrung.

You can point at all the places in an organization, but, you know, without the authority to actually go in and say, "Hey, we're gonna really rethink how work is structured for this function or this role or this department," that requires some autonomy and authority to be able to go in and do that. Because functional leaders, right, are very close to the work, and sometimes as functional leaders that are really close to the work, you get tied to a certain way, and it makes you very nervous to think about changing how things, you know, things are being done.

And so it's, it's not solely an AI or IT team, 'cause they don't understand that nuance that those functional leaders have, and it can't be functional leaders alone, because they don't understand all the capability that A- AI can bring to an organization. And so really it needs to be someone often a COO-type role Right?

Who owns outcomes end to end and treats the redesign as an actual deliverable, right, across an organization. And AI is just one way you may solve the problem, not the way.

David Rice: Yeah, I would, I would agree with you. I, I think we've had a lot of conversations over the last year about sort of the CHRO's remit needs to expand and to, from designing jobs and organizations towards designing how human and machine capacity combine to produce work.

But I think what that's led to for those who have done it, is sort of an increase in operations responsibility rather than just it's still sitting, feeling like you're, you're mostly in HR or people operations, right? It is, it has become... It's, it's all evolved the role into a place where I don't know if that's the long-term solution, to be honest, 'cause I think that the people challenge is big enough as it is.

Heidi Farris: I agree. I think what, what AI brings to HR leaders, which is very exciting in, in my mind for, for HR, is that it actually starts to assimilate people and operations together. You know, a lot of times organizations put HR over on the side and say, "Well, HR does a thing over here, and then we all operate a business over here, and HR comes in with compliance things and whatever they need to when they, you know, w- and they have their own data, and we have the data that we run the business on."

And now that data has to all kind of be the same view of work because if you're gonna operate and redesign roles, HR obviously has, is a key, if not the owner, a key stakeholder in that whole conversation, right? And because it has implications throughout, it has implications on, on comp and title and structure and all of those things.

And so I actually think that AI is very exciting for-- I-- well, while very taxing on one hand, I think it's also very exciting for HR leaders who wanna now lean into actually being part of running the day-to-day business as opposed to, you know, um Uh, you know, a compliance center that sits on the side of everything.

David Rice: You mentioned earlier, you know, like the usage metrics, and I think that there's, there's a lot of that out there. The licenses activated, the weekly active users, the use cases launched, all these sort of things that tell us like, "Oh, we are using this technology." I don't think that they necessarily tell us that you've moved beyond just adopting and you've changed operationally.

What would you say is sort of like the signs that that's ha- that somebody has done that and that the business is now running differently?

Heidi Farris: Yeah. I think you know an organization has moved past adoption when leadership, or really anyone in the organization, can articulate what work was redesigned, what the outcome was, and where that's showing up in goals, right?

And, and for leadership, that's the P&L. For the teams, it may be sub-goals under that. But actually being able to, instead of saying the one-off like, "Oh, our CS team now saves..." You know, they, they used to do QBRs and they had to do them manually and collect all the data for a customer and create the deck and everything.

Now, you know, we- we're able to do that in what used to take four hours in five minutes. That's one thing. But to be able to say, "Okay, but where did that show up in a goal or on the P&L?" Because that by itself is not the thing. The thing is, and with that extra time, they were able to increase their book of business by 50 customers each, and now I need you know, a 1 to 75 ratio of CSMs to customers as opposed to a 1 to 25 ratio of customers.

That's where it starts to actually show up in goals and things. So when an-- what I believe that those are the key markings of when an organization actually moves from just adoption to impact, which is, don't just tell me what you did with AI, but tell me then what you did with the benefit that you got from AI.

David Rice: Yeah. I, I, I'm a more and more, as I have these conversations, I'm arriving at the conclusion that like eventually it has to become visible somewhere. It can't just be in a dashboard. It's gotta be like your role design, your customer outcomes- organizational structure, the skills, whatever it is. But if it's genuinely transforming the org, I should basically be able to see evidence without looking at a metric.

Maybe that's the ultimate sign of AI maturity is that we stop talking about AI really, because we're talking about how the business operates.

Heidi Farris: Yeah, I couldn't agree with that more. I think that HR and operations are gonna partner, and AI is one of many solutions as we dig into the world of how does work get done today.

David Rice: Absolutely. Well, Heidi, that's actually all the questions I have for you today. Thank you for coming on the show and for, for giving us some of your insights. I really appreciate this.

Heidi Farris: Again, David, thanks so much for having me.

David Rice: Listeners, if you haven't done so already, head on over to peoplemanagingpeople.com/subscribe. Get signed up for all our newsletters. There's three that go out every week. You'll get all this content packaged up nicely straight to your inbox.

And until next time, people are already changing work for the better every day. Don't forget that.