So What? Top 5 Use Cases of Job Descriptions for Business Intelligence

Katie Robbert – 00:31

Hey, everyone. Happy Thursday. Welcome to So What? The Marketing Analytics and Insights Live Show. I am Katie, joined by Chris and John. Howdy, fellas.

John Wall – 00:39

Hello.

Katie Robbert – 00:41

We’re at the awkward angle today. Somehow I’m always a little bit caught off guard when I do the introduction because in my brain I’m thinking, “Wow, it’s already Thursday again. How did that happen?” But here we are, and today we’re talking about the top five use cases of job descriptions. Yes, there is more to a job description than just, “What does this job entail?” which, quite honestly, is never accurate. So, let’s find other ways to strategically use a job description.

One of the things that I’ve really enjoyed about using a large language model is it’s helped me do that deeper analysis that I’ve always been struggling to do on my own because I’m a single human. I look at something and I see a lot of different things in it, but I can never quite put my finger on it.

Katie Robbert – 01:31

You can look at a job description and see there’s more here. It’s telling me something else, and I’m not quite sure what other insights I can glean other than, “This is the open position.” That’s kind of what we want to cover today. We’re using job descriptions as a stand-in, but think about any kind of material content that you have access to.

A lot of times, as humans, we can only see it as a singular use. This is the thing; this is what it does. But the nice thing about a large language model is it’s not constrained to only see it used that one way unless you tell it that’s how it works.

Katie Robbert – 02:13

Today, Chris is going to walk through how we get there, using the example of job descriptions. What are some other ways we can use that information?

Christopher Penn – 02:25

Yes. Let’s start off by talking about the importance of a job description. Specifically, why job descriptions? In general, you don’t hire for fun. Hiring sucks on both the job applicant side and the hiring manager side. It sucks; it’s a long, exhausting process. There are interviews, and companies, especially these days, do not open headcount for frivolous stuff.

You only open headcount for one of two reasons. Either you are trying to backfill a role that you have a mission-critical need for that you can’t farm out to an agency or AI, or you have a strategic imperative that you need to build competence or a center of excellence for within your organization, and you do not have those people or agencies to do that. You need to get net new people on.

Christopher Penn – 03:17

Those are the two things that a job description implicitly tells us. It tells us, “Hey, this is a thing.” Either we are doing the thing and we are about to lose somebody—or lost somebody—who’s doing the thing, or we want to do the thing and we can’t do it with the people we got.

Katie Robbert – 03:35

That’s not new information. That’s how job descriptions have always worked since the invention of the job description, though I don’t think we have that data. I remember applying for jobs, and job descriptions for the most part are kind of vague. They only kind of tell the end user, the applicant, what you’re meant to be doing.

If I think back to my job description at Shift and what it said versus what I was doing, it was wildly different. That is one of those caveats that I want to keep in mind—there’s an assumption that people kind of know, based on a title, what a job is meant to do.

Katie Robbert – 04:20

A job description is really just a list of high-level tasks and then the dreaded “and other duties assigned,” which could mean anything from getting John coffee every day to trying to clean up Chris’s technical debt, even though you’ve never touched a server in your life.

Christopher Penn – 04:40

Yep. Like making coffee and having to run it to your board of directors, which happened at our old agency. That was a delight.

Katie Robbert – 04:49

Yeah, that was. Anyway, let’s look at—

Christopher Penn – 04:55

—a job description as an example. We’re gonna pick on McKinsey, the largest consulting firm on the planet, because they’re the largest consulting firm on the planet. Every company that’s hiring has some place to put career stuff. Most companies, particularly enterprises, use something like Jobvite, Greenhouse, LinkedIn, or Workday.

There are all these systems connected to their HR platforms that publish stuff on the web. If we go to McKinsey here, we go to careers, and we go to search jobs, that brings us to their job portal. Their job portal, surprisingly, is not a third-party thing. It’s a single-page application. It’s actually kind of homegrown, which is interesting in its own right. There are over 500 jobs available at McKinsey.

Christopher Penn – 05:49

Now, this is out of a 50,000-person firm. It’s actually not that many, but we see them across all the different business lines. This goes on for quite some time, so let’s take a look at this one. This is the job description of a McKinsey analyst in Mexico City. It says, “Your impact. Here’s what you’re going to do.”

You’re going to execute the growth strategy, whatever that means. You’re going to build B2B and B2C paid media programs across multiple social advertising platforms as well as owned channels. You’re going to generate significant insights—

Katie Robbert – 06:22

—as opposed to insignificant ones.

Christopher Penn – 06:24

Yes. Data analysis. You will generate insights, verify them with the appropriate organization—so, marketing—and share your recommendations with key decision makers. You’ll tell stories with the data and enable the growth marketing mix. I feel like this is a drinking game.

Katie Robbert – 06:37

John, is this getting you excited? Are you going to go apply now?

John Wall – 06:40

Yeah, I know. Could it be any more vague, really? I’m looking for some synergy in here and we’ll be—

Katie Robbert – 06:52

At the end of the day, what you’re doing is drafting ad copy and loading it into all of the different ad servers, and then you are responsible for the metrics of those things. So, you’re building PowerPoint presentations endlessly.

Christopher Penn – 07:11

It says, “Our office culture is casual, fun, and social with an emphasis on educational innovation.”

Katie Robbert – 07:19

I need a little red flag to start holding up during meetings when people say certain things. That’s one of them.

John Wall – 07:26

Yeah. All global consultancies are known for their fun workplaces. That’s definitely—

Christopher Penn – 07:30

Yeah.

Katie Robbert – 07:31

I can’t think of a single enterprise company that is considered casual.

John Wall – 07:36

Yeah, no.

Christopher Penn – 07:39

That’s the job description. What’s interesting is they are pretty thin on what they’re looking for. A bachelor’s degree in these fields, knowledge of primary paid media, willingness to travel, digital marketing is a plus, and being bilingual are the big things there. That’s it; that’s the application.

John Wall – 08:04

Yeah.

Christopher Penn – 08:05

The first thing that—go ahead, Katie.

Katie Robbert – 08:07

What I was going to say, and this is a digression, but the lack of expertise to me says that this is an entry-level role. It fits the conversation of, “What am I supposed to do? I’m just out of college.” There are still entry-level roles to be had if you want them.

They’re clearly looking for people who are pretty much fresh out of college because for the two-plus years of experience, quite honestly, a lot of times you can skate by with internships or portfolio projects—things you did in college. You really just have to demonstrate that you know what the thing is. That’s a digression; I’ll try not to do that too much this episode, but no promises.

Christopher Penn – 08:54

The first thing we can do is decode. We can and should deconstruct a job description into specific tasks. One of the things that AI, as Katie pointed out at the beginning of the show, is excellent at is reading between the lines to see what’s in here because it’s language.

One of the things that we have—and I believe it is in the Trust Insights Academy—is a job-to-skill tool that decomposes a job description into the different things this job does that could be automated by AI. We have things like pulling and compiling campaign performance reports; that is a big one. Bilingual translation reports are a shoo-in for AI. Building and matching B2B and B2C campaigns—there’s some AI in there for sure, but there’s also some integrations.

Christopher Penn – 09:48

Building and maintaining KPI dashboards is very easy for AI. But these top ones here—pulling and creating campaign performance reports and bilingual translations—those are slam dunks for AI. Immediately, just from a single job description, we can decompose it into tasks. This will help a job seeker understand, “What am I actually going to be doing?” because we all agree that vague description was kind of squishy.

Katie Robbert – 10:19

Yeah. It tells you not only what you will be doing, but how quickly you need to outpace AI, or AI is going to be doing it for you.

Christopher Penn – 10:30

Anything color-coded green on here—

John Wall – 10:34

Yeah.

Christopher Penn – 10:34

—you’re probably either going to be using AI to do that, or you will no longer be doing that at—

Katie Robbert – 10:39

—the office, which doesn’t leave a whole lot.

John Wall – 10:43

Yeah.

Christopher Penn – 10:43

Down here are client site workshops and travel. That’s a human thing.

Katie Robbert – 10:48

Sure.

Christopher Penn – 10:49

Collaborating with data scientists and engineers to compile recommendations, coordinating cross-functional digital marketing partners, verifying insights with marketing organizations, and presenting recommendations to decision makers. So, you’re going to be presenting slide decks.

Katie Robbert – 11:01

Yeah. You’re going to be scheduling a lot of meetings.

Christopher Penn – 11:10

When we look at the actual use cases, we can see what the strategy, tactics, and execution are for this to say, “Okay, here’s how you would have AI use its capabilities to actually accomplish this task.” Our first two use cases for job descriptions are: A, distilling the job down into the specific tasks you’re going to be doing; and B, doing a threat analysis, as you said, Katie—how likely is it that a machine is going to do this?

Katie Robbert – 11:42

I think that’s an important tool, and I did put up the link. You can find this particular skill in our academy, which you can get at academy.trustinsights.ai. If you’re a job seeker, it’s a really good use case right now because you might get hired for a job and think, “This is great, I don’t have to keep looking,” but three months from now, they say, “We hired this outside organization to build in AI efficiencies, and that’s 90% of the work you’re doing now.”

You want to go in with your eyes open to know the likelihood of that. It gives you an opportunity to really get skilled up on how to build those workflows in AI so that you’re the one in control.

Christopher Penn – 12:27

Exactly. Here’s another thing that you can learn from job descriptions: you can learn the technology stack of the organization itself. A lot of the time, for example, you’ll see it requires knowledge of Word, Excel, and PowerPoint. If it’s saying Word, Excel, and PowerPoint, we automatically know it’s at least partially a Microsoft shop.

Katie Robbert – 12:52

Which means that you’re likely tied into—

Christopher Penn – 12:54

—Copilot. If you see Google Sheets, Google Docs, and so on, it’s Google Workspace. For example, if they’re saying paid social media, you’re probably talking LinkedIn Ads, Meta Ads, or AdWords. Interestingly, they didn’t explicitly say Google Ads, but it depends; if they’re saying social media ads specifically, that’s not Google Ads, though it could be YouTube Ads.

Katie Robbert – 13:25

Hopefully. That’s a really important key as well because it gives an interviewee an opportunity to be prepared for those inferred questions you’ll likely be asked during the interview. You will face questions like, “What’s your comfort level with LinkedIn Ads? What’s your comfort level with Meta Ads?” even though those platforms were not explicitly called out in the job description.

If I were a job seeker looking for this particular job, I would find this useful because then I could say, “Okay, Meta, LinkedIn, Google,” and prepare accordingly.

Katie Robbert – 14:08

Are there certifications that, while I’m looking for a job, I can go get to at least say, “Hey, I’ve opened the platform before and I know the basics of how to navigate it”? You can learn a lot when you get in there, but if you’ve never opened the platform before, you’re going right to the bottom of the list.

Christopher Penn – 14:29

Now here’s where we start talking about not just one job ad. If we look at this list here, there are 572 of these things. If you wanted to—and I do—you can take a tool like Claude Code and say, “Build me the Python code necessary to extract all 572 job descriptions.” You go back and forth with Claude, and it will say, “Great, let me help you design a database to store them all.” Then you can start gluing together the different aspects.

I did this morning. I had Claude take one of our existing scraping tools and modify it specifically for McKinsey’s website because McKinsey has its own sort of custom-rolled job board. It’s not Jobvite; it’s not the other one, so my existing templates didn’t work.

Christopher Penn – 15:29

When you do that and start asking, “What does the martech at McKinsey look like based on all the job ads?” what you end up with is a five-layer sandwich of marketing and commercial applications, their agentic AI product layer, their data science layer, and their cloud and DevOps layer. What technologies are mentioned in their job ads? Enterprise platforms and their BI.

At McKinsey, based on their job ads, they definitely use generative AI. They for sure have their own RAG databases of some kind. They’re using LangChain and vector databases, so they’re building their own martech to a degree. They’ve got a lot of Python and SQL, some Airflow, and some PyTorch, but it’s very much a Python shop.

Christopher Penn – 16:22

They have a lot of AWS, Azure, and GCP, which is unusual until you realize that’s not for them. That’s because their big enterprise clients all use those platforms.

Katie Robbert – 16:34

Right?

Christopher Penn – 16:36

You have McKinsey ID, the McKinsey internal platform, Lilli, their internal AI, Okta, and SAP HANA. That almost certainly means that is their ERP and their CRM. It’s SAP and ServiceNow for their internal ticketing, and then Tableau and Power BI. Power BI combined with these other things, plus Azure, tells me they’re almost certainly a Microsoft shop.

Katie Robbert – 17:01

It reminds me of when I started working with you a decade ago, Chris. There was a Chrome extension that you would use called BuiltWith. For people who are thinking, “Wow, you really did a deep dive on McKinsey,” this is not a new technique that companies use to understand their competitors.

McKinsey is way out of our reach; they’ll probably never see any of this. But this is all based on publicly available information from their website. What we’re showing is just a different way to do the same kind of analysis that marketers and sales folks have been doing for decades.

Katie Robbert – 18:19

With BuiltWith, you used to be able to scan someone’s website and get roughly this kind of information about what’s in their tech stack. If I remember correctly, Chris, it was limited to things like what pixels or integrations were within the website, so it wasn’t a full picture. This is just another version of that, but it’s based on what they’re listing in terms of their roles, responsibilities, and needs. We’re inferring what systems they likely have.

Christopher Penn – 18:19

If you wanted to add that in and do it yourself without paying for the premium version of BuiltWith, it’s actually relatively easy to do. If you go to the McKinsey website—maybe the homepage, or ideally a page where there’s a conversion form of some kind—go to View, go to Developer, and turn on Developer Tools. Make sure it’s set to the network tab and refresh it. You can see all of the martech that just fired.

If I scroll down here and go to the form, I can download this file. This is a great big, honking data file that tells me what happened when I loaded the McKinsey website. This thing is 15 megabytes, and it’s all in JSON format.

Christopher Penn – 19:14

You and I ain’t going to read this. However, we can hand this to Claude Code, Claude CoWork, or the AI tool of your choice and say, “Using jq, which is a JSON querying tool, take this audit file apart and tell me what marketing technology was in operation when I was on this form on McKinsey’s website.”

Katie Robbert – 19:37

I think that’s useful just to know. Again, this is all publicly available; every website has the option to take a look and view it that way. When you’re looking at using job descriptions to infer that information, this is a good way to double-check that the information you’re inferring is roughly accurate, since we know job descriptions are vague. You’re never going to get the full 100% picture.

I think about people like you, John, who do biz dev. You might look at that tech stack and think, “Wow, Trust Insights has expertise in 80% of those things. Maybe they need help cleaning up their data or integrating things.”

Katie Robbert – 20:22

It might be worth a pitch because we can credibly say, “Hey, here’s our 10 case studies that speak to all of these different technologies that we can help you with.” I can see that being a way to use the data extracted from the job descriptions—you hand it off to your sales team and go after the companies that match our expertise.

John Wall – 20:45

Yeah, to be able to get a profile of what’s in there just sets up a bunch of discussions. We know they’re going to have this set of problems because they use [insert your hated CRM system here], and there’s a whole bunch of things. BuiltWith is still doing their thing, too; I get emails from them saying they can find these people for you.

The thing that’s jaw-dropping to me, though, is 15 megabytes. What the heck happened to Google mobile data speed? That is just a berserk amount of data to be going through a form like that. Again, you’re a consulting house and you’re building your own form, so who cares?

John Wall – 21:28

The other one was that I did see on the list there that SEO does not show up as part of their stack or their analysis. They’re treating it like SEO doesn’t matter anymore, or they’re giving that one up.

Christopher Penn – 21:39

It says they’re not hiring for it, so it’s likely they already have those capabilities or have partially outsourced them. This tells us either where they’ve got a shortfall in their staffing or where they desperately need to expand their capabilities.

Just by looking at this alone and knowing who their clients are—the biggest companies and organizations on the planet—they’re trying to grow their forward-deployed engineering team to be able to drop anybody into a client and say, “Okay, we’ve got this engineer or analyst who can come in and work within your Google Cloud to connect it to our system.” Their system is called Quantum Black.

Christopher Penn – 22:26

They want to know, “How can we get your Google Cloud, your BigQuery database, or your data warehouse connected to Quantum Black so that we can run analytics on it?”

Katie Robbert – 22:37

Makes sense. I see those as a huge opportunity when looking at it from the lens of someone like John. They can keep hiring, or they can hire an agency to hit the ground running and handle it. There’s a lot of opportunity there for someone like us to step in and say, “Hey, we do that, and we’re not going to cost you the overhead of health benefits and all that extra stuff.”

Christopher Penn – 23:06

It’s funny you mention that because that’s the ultimate use of these job descriptions. If you were to download them all—which we did—convert them, and put them into a system like Claude Code, you can say, “I want to do a very deep analysis of this. I want to understand what this is in the context of a company like Trust Insights.”

So, I built this prompt saying, “Here are the 572 jobs that I pulled in. You’re a competitive intelligence director. Here’s some information on how that file is formatted and everything you need to know about Trust Insights,” which I got off the about page of our website because it’s the most current version. My core thesis is that companies hire based on need, and the major needs are either backfill or expand.

Christopher Penn – 24:07

To think this through, we instruct it to use nothing less than the 5P Framework by Trust Insights.

Katie Robbert – 24:07

I am so surprised by this revelation of information. If you want to learn more, it’s trustinsights.ai/5p. The 5Ps are Purpose, People, Process, Platform, and Performance.

Purpose: What the heck are you doing? People: Who’s involved, both internally and externally—your customers and your end users. Process: How are you doing it? These are your instructions, your standard operating procedures, and your processes. Platform: What tools? Performance: Did you do the thing you set out to do?

Christopher Penn – 24:37

Exactly. In this prompt, Purpose answers one question: Based on McKinsey’s hiring, what is their corporate strategy for the next 12 to 18 months, and what should our boutique firm do about it? Should we counter it? Should we coattail it? Or should we identify the white space that they’re not talking about because they’re not hiring for it right now? That’s our Purpose.

The People are our firm’s leadership, AKA Katie, who’s making investment bets, and we’re going to look at McKinsey’s people. You, Claude, are the competitive intelligence director speaking to our leadership, so we define the People.

The Process is as follows: prep the corpus, because you have to split it up since it’s a big file. Map each job description into a structured record. Give it weight—junior role versus senior role.

Christopher Penn – 25:35

A senior role should have more weight in this kind of competitive analysis because you’re going to hire a lot of boots on the ground, for sure, but somebody’s got to be in charge. The higher up the role that you’re hiring for, the more interesting it is because those are the really expensive people; you don’t hire them for fun. Group together similar roles, so if you have an analyst in Mexico City and an analyst in Boise, they’re still analysts.

Triangulate by using your web search tools to check things like news articles to see if the patterns are informed by recent news or changes in the news, and then synthesize a report. You have to state your evidence for this. The Platform, of course, is the data chunking and our own custom code for this.

Christopher Penn – 26:24

The outputs are a Markdown file, in case we want to use it again for AI somewhere else, and an HTML file. Performance: you are done when you have gone through all these steps, can explain everything that you did, and can ask up to five questions. Here is the example; here are the deliverables.

When we run this, what do we get? We get a great big, honking, huge playbook. Here is what McKinsey’s up to. I thought this was interesting: based on their job descriptions, they’re aiming to become an AI deployment company, meaning they’re going to come and build the stuff and do the stuff. It walks through and says, “Here’s what we see.” They are hiring forward-deployed engineers to do AI at their client sites.

Christopher Penn – 27:24

They’re not giving binders. They’re saying, “Our clients just want us to come in and build the thing, and they want it based on McKinsey’s tech so it’s harder to rip out later.” It goes through all these different things. I thought this was really interesting—they’ve repositioned from the firm that tells you what to do about AI to the firm that builds the AI and runs your digital transformation internally. This is going from talking about doing it to actually doing it, which Katie, you’ve been talking about a lot lately.

Katie Robbert – 27:55

Yeah, no, it tracks. Clearly, they stole it from me. I’m kidding, McKinsey, I know you didn’t; you really have way more clout than I do. It’s a common theme that we’re seeing across all of the AI thought leaders who are speaking credibly about it. We’ve moved past “What is AI?” to “How do I get it to run, and run tomorrow?”

This is exactly what they’re saying. They’ve repositioned from the firm that tells you what to do about it—which is the education—to the firm that builds it and runs it, which is the actual doing of things. That’s unsurprising to me. I would actually be shocked if they were still saying, “No, we’re not shifting gears. We’re still just going to tell you what to do,” because that doesn’t matter anymore.

Katie Robbert – 28:41

We’re way past that. We’re three-plus years into “What is AI?” and now the conversation is almost solely focused on, “How the heck do I get this thing, how do I get it to stay, and how do I get it to work?” What it says to me is that if you have any skills working with code and software development—any of those foundational skills—now is the time to get in with a firm like this if you’re looking for a job.

John Wall – 29:14

Yep.

Christopher Penn – 29:15

162 of the 572 job descriptions are for forward-deployed engineers.

Katie Robbert – 29:19

That’s a lot.

Christopher Penn – 29:20

Yeah. They have concluded that advice alone no longer commands premium fees in the AI era, so they’re moving down the value chain to become like Palantir. They’re saying, “We’re gonna build the thing for you.”

John Wall – 29:31

They have over a billion dollars in Quantum Black, too. That’s just mind-blowing. They are moving all their chips onto that; that’s it.

Christopher Penn – 29:40

Yep. Shift two: they are changing their internal structure. They are ditching their junior roles, junior research, and back office in favor of engineers. The huge corpus of junior analysts is being replaced because of AI tooling on their internal platform, Lilli, which handles 500,000 prompts a month according to web research. All those junior PowerPoint jockey roles are going away, and they’re replacing them with engineering roles, which makes sense.

Katie Robbert – 30:15

We talk about the value of doing deep research and how to get started with it—that’s actually our Casino Framework. You can find that at trustinsights.ai/casino; that’s our deep research prompt. If you think about an organization like McKinsey that has the resources to really, truly automate that, I can understand why they don’t need those junior analysts compiling data anymore. That is something that AI does really well.

Christopher Penn – 30:48

Yep. The third major shift in here is that instead of saying they’re everything to everyone, they’re doubling down to create industrialized AI per vertical. They want to be like Palantir. They want to say, “We want to be the AI implementation vendor for life sciences, for defense, and the public sector.”

They’re saying this is where they can defend ourselves. They have proprietary data, decades of expertise, and the Rolodex. If they can build Quantum Black to be strong enough, they can be the tech stack for the Pentagon, as an example.

Katie Robbert – 31:26

It’s interesting. I never think about companies like McKinsey having aspirational companies, but McKinsey might want to do this kind of analysis on Palantir to understand what they’re doing. So, McKinsey, we can do this for you! It’s interesting because I don’t personally think about the big enterprise companies worrying about what somebody else is doing, but in this sense, it totally makes sense.

Christopher Penn – 31:55

Yep. They’re in a constant knife fight with Bain, BCG, and Deloitte. They’re all jockeying for the trillion-dollar companies.

Katie Robbert – 32:05

Yeah.

Christopher Penn – 32:07

Now that we know what they’re up to, we can look at how they’re going to do it. Industrializing AI—what are the specific things that they are doing? They’re building production-grade AI delivery, so they’re going to be dropping this stuff in as their own applications. It even has recommendations for us as Trust Insights.

Katie Robbert – 32:28

Right.

Christopher Penn – 32:28

We obviously can’t be McKinsey. We can’t build a trillion-dollar data system. As much as I would like to think that we could, realistically, we can’t. But what we can do is help companies with the components of the 5P Framework that McKinsey isn’t touching at all. McKinsey is going all-in on Platform.

If you look at Purpose, People, Process, Platform, and Performance, they’re saying, “We’re Platform.” This means our opportunity is to say, “Yes, you got the platform, but let us help you with the People, Process, Purpose, and Performance so that instead of an expensive binder, you don’t just end up buying an expensive server.”

Katie Robbert – 33:11

Makes sense.

Christopher Penn – 33:12

With proprietary products, organizations buying these black-box products will need an independent partner to validate the outputs. We don’t build black boxes; we need to know what’s in the box. But that’s definitely the route that they’re going with Quantum Black. This next one I thought was interesting: McKinsey is running itself like a tech company.

John Wall – 33:31

It’s—

Christopher Penn – 33:32

They are becoming a tech vendor.

Katie Robbert – 33:36

Like Palantir, which supports the pivot and the job descriptions that they have open. That is not an easy pivot. There’s a lot that goes into that because, Chris, you and I talk about this all the time: there’s a big difference between advisory consulting and software-as-a-service.

There’s a big reason why I’ve kind of held us back from being a software-as-a-service provider—it comes with a lot of headaches. Sure, there are dollar signs attached, but at the end of the day, depending on the size of your company, it’s not worth the headache for a team like ours. Whereas McKinsey, which has 50,000 employees, can probably find someone who’s willing to deal with the headache for enough money.

Christopher Penn – 34:26

Exactly. What was also interesting is that they’re pushing a lot of their jobs away from the US core. They’ve got a lot of hiring in APAC and Latin America especially. Were you gonna say something?

Katie Robbert – 34:45

Go ahead. Nope.

Christopher Penn – 34:47

This tells us that they’ve got a cost issue and that the talent in the U.S. they’re trying to go after is way too expensive. That makes sense; Anthropic, OpenAI, and Google are all competing for the exact same people. They’re thinking, “Maybe we should try these other places that have functioning education systems.”

Katie Robbert – 35:01

It makes sense, especially when you’re trying to pivot your entire organization. You’re going to have razor-thin margins.

Christopher Penn – 35:14

It’s really interesting. With the Casablanca Graduate Program, they correctly recognize that places like Nigeria have some of the best growing AI talent. Lagos, Nigeria, long the butt of jokes for spam farms, actually is one of the places where you see a lot of papers being submitted to conferences like NeurIPS. They have invested heavily in their people and their education system, and it is paying dividends. They have a tremendous AI hub in Lagos—

Katie Robbert – 35:49

—which totally makes sense because McKinsey has the resources to do that research to find out where that talent is.

Christopher Penn – 36:00

The big thing is they have said, “Strategy without execution is a hallucination,” which is our thing. We want to make sure that you don’t just get a binder, but actually do the thing.

Then we scroll past all this stuff into execution. How are they going to do this? They’re moving around forward-deployed engineers, they’ve got 15 client-facing named products, and they’re becoming a software company, which is what they do. What I thought was interesting is they have a partnership with Google on that front because they’re running into a shortage of qualified people.

Katie Robbert – 36:51

That makes sense. I’m assuming that was publicly available information, probably some kind of a press release, because for companies like McKinsey, that’s a big part of how they communicate. Especially because they are—I believe they’re publicly traded. Are they?

Christopher Penn – 37:08

I don’t know.

Katie Robbert – 37:09

They’re a big enterprise company, and press releases are a big part of how they communicate pretty much everything for full transparency into what’s happening.

Christopher Penn – 37:20

Yeah. What are they specifically optimizing for? They are looking at revenue per head; when you look at the big-picture pattern, they are trying to figure out how to make themselves more profitable. That goes completely in line with slimming down that huge base of junior analysts and making it an engineering firm. That’s how they decide what success looks like, and they are measuring cost based on where they are looking for—

Katie Robbert – 37:56

—talent. And I misspoke; McKinsey is a privately owned company. They are not publicly traded.

Christopher Penn – 38:02

Yeah, but all this information itself is public.

Katie Robbert – 38:05

Right.

Christopher Penn – 38:05

What it delivers for us afterward is to see what Trust Insights should be doing based on everything McKinsey is doing. It analyzes whether we should coattail, counter, or identify a white space. For coattail plus white space, there is vendor-neutral AI governance, measurement, and adoption—that’s what we do. For proprietary products, we can coattail and counter; we can validate that you’re not being sold something that doesn’t actually work just because it has the McKinsey logo on it.

For nearshore delivery, we counter—we’re onshore. Then for vertical depth, we are going to ignore the defense and government sectors because we’re just not equipped for that. We can focus on monetizing AI governance and literacy via the 5P Framework for the mid-market.

Christopher Penn – 38:59

This one is a huge part because if you think about a company like Palantir—Maven is the name of their targeting AI that they provide to organizations like the Department of War. It is in part responsible for some of the tragic accidents that have happened in the Iran conflict. When we talk about making sure you’re using AI properly and it’s doing what it’s supposed to do, those are the kinds of things that we want to be able to help with.

Katie Robbert – 39:45

What we’re looking at in a nutshell is a competitive analysis. This is under the assumption that McKinsey is an aspirational competitor for us right now. But now that they’re pivoting towards being more of a software-first company, I don’t know that I would necessarily consider them a competitor because that’s not the route that we’re going with Trust Insights.

As we are coming up to the last third of the livestream, we’ve covered three or four use cases for job descriptions.

Christopher Penn – 40:21

I think we’ve hit all five in varying flavors: task decomposition—looking at what you are hiring for so you, as an individual job seeker, can figure out if you have those skills; decomposing the tech stack of a company on both an individual and a larger basis; looking broadly at the overall strategy of the company; and then specifically performing a competitive analysis, which is where we are now.

Katie Robbert – 40:55

Makes sense. Again, this is all based on publicly available information. It just came from the careers page of their website and a little bit of extra research that Chris added into his prompt.

The accuracy of it really depends on the accuracy of the job descriptions themselves, which is a big asterisk, so proceed with caution. But if McKinsey were a true competitor that we wanted to do something serious with, this gives us enough directional information to work with.

Christopher Penn – 41:39

For us, this highlights the white space. The language model identified what it didn’t see in their 572 job descriptions compared to Trust Insights. A couple of things stand out to me. Obviously, our heritage is measurement and analytics—those are our roots. The mid-market is another big one, but the main thing missing from McKinsey’s descriptions is our approach. McKinsey is saying, “We just want to do all the AI for you.”

Our tactics are different. If you look at the Trust Insights Academy and how we consult with clients, we take a “teach you to fish” approach. If you look at our speaking and workshop packages, we don’t try to do all the AI for 50 or 100 companies ourselves because we don’t have the headcount.

Christopher Penn – 42:20

We would rather have you do the AI yourself. We want to help you get past, “How do I prompt?” and then bring us in to do some really fun stuff.

Katie Robbert – 42:32

It’s interesting. This is really important for us as a company, but for someone watching this livestream who is looking for a job, this is a great way to prepare for what’s coming and what the interview questions are likely to be. This is especially true if there’s technology vaguely alluded to but you want to get a deeper dive or really set yourself up for success by seeking out certifications.

Maybe McKinsey is an aspirational company for you to work for. This is a really good opportunity to evaluate what you need to do with your own skill set to make yourself a desirable candidate to an organization like McKinsey.

Katie Robbert – 43:17

You can imagine that McKinsey is only one of many firms like them moving in this direction.

Christopher Penn – 43:25

I completely agree with you on that. If I were a job seeker with a skill set in AI—assuming I didn’t get hired just for having the words “AI” on my LinkedIn—I would put together something just like this, tailored to my profile. I would say, “Here’s what I used AI to find out, and here’s how I can help you achieve these specific goals.”

Katie Robbert – 43:50

If you’re interested in getting a copy of that skill we mentioned, we have it for sale as a digital download in our academy at academy.trustinsights.ai.

I think this is a really interesting way to look at job descriptions because, as humans, we tend to have a narrow bias that a job description is only going to tell us what tasks are involved. A large language model isn’t restricted to those guardrails unless you impose them. It’s going to look at the text and say, “Hey, this is a whole bunch of cool data. Look at all these different things this tells me.” I think that’s fantastic.

Christopher Penn – 44:34

Oh, yeah, totally.

Christopher Penn – 44:37

When we think about strategy, we’re thinking about the why. Why are we doing anything? Nothing tells you the why quite like opening up headcount; we’re about to spend tens of thousands of dollars on a person to sit in a seat.

If we can get the data, which today’s AI tools can help you do, and you have an opinionated take on that data with your 5Ps all mapped out, you can do some pretty cool stuff. There are things above and beyond this that we’re probably going to take for ourselves to work on. But for everyone else, if you work at a company and want to know what your competitors are up to, this is how to do it.

Christopher Penn – 45:31

If you’re asking, “How in danger is my job?” this is how to do it. If you’re wondering, “Should I acquire this company? What kinds of tasks are they hiring for, and what do they think their strategic priorities are?” this is how to do it.

If you are merging companies together, you can take all the job descriptions and say, “Not only are we going to consolidate headcount, but we’re also going to look at how many tasks we can unload across the portfolio to AI so that we can make the most of the people we have.” What McKinsey is doing—trying to uplevel revenue per headcount—is exactly how you do it.

Katie Robbert – 46:13

If you want to learn how to do it, go to trustinsights.ai/contact and talk with a real human being: John Wall. He will be there to answer your email and tell you all about what you can actually do with this function of our tool. Chris, as you mentioned, we just scratched the surface of the possibilities.

With large language models, the limitation is really only your own imagination, and this is proving that point. We look at a job description and think, “This is what the tasks are.” A large language model looks at it and says, “Here are 20 different things I can do with this data. Which one do you want to start with?” I think that’s fantastic.

Christopher Penn – 46:56

Exactly. John, any final thoughts?

John Wall – 47:00

Again, give us a call if you want to dig more into this. As far as a strategic window, it’s fantastic to be able to do this; there’s no way you’re going to comb through thousands of job descriptions manually and figure out what’s going on. This is the use case that can give you a map of where the opponents are going, and there’s no substitute for that.

Christopher Penn – 47:20

For clarity, Katie gave me the guidance about which company we were going to cover on the livestream this morning at 9:30 AM. In the last four hours, we were able to get AI to process this. Granted, we had a lot of pre-existing code and our own web scrapers, but just in terms of getting the data and processing it, we were able to come up with something in less than half a day—

Katie Robbert – 47:46

—which is why you should hire us to do this for you.

John Wall – 47:49

Operator standing by.

Katie Robbert – 47:51

Yes, John Wall has his switchboard ready.

Christopher Penn – 47:58

I think we’re done.

Katie Robbert – 47:59

I think we’re done.

Christopher Penn – 48:01

Thanks, everyone. We will talk to you all on the next one. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it.

For more resources and to learn more, check out the Trust Insights podcast at trustinsights.ai/tipodcast and our weekly email newsletter at trustinsights.ai/newsletter. Got questions about what you saw in today’s episode? Join our free Analytics for Marketers Slack group at trustinsights.ai/analyticsformarketers. See you next time.