So What? Marketing Analytics and Insights Live
airs every Thursday at 1 pm EST.
You can watch on YouTube Live. Be sure to subscribe and follow so you never miss an episode!
In this episode, Katie Robbert, Christopher Penn, and John Wall talk through how artificial intelligence is changing client work and traditional billing models.
Building a profitable agency of the future means shifting your focus from tracked hours to the real economic value you create for clients. Smart business owners will swap outdated hourly retainers for pricing models that protect profits while proving clear returns. This shift gives your agency of the future the power to mix modern tech with human judgment so client work stays essential. Early adoption of smart tools will keep your team growing while competitors struggle to keep up.
Watch the video here:
So What? Defining the Agency of the Future
Can’t see anything? Watch it on YouTube here.
In this episode you’ll learn:
- Why the billable hour is a death knell for agencies of all kinds
- What the agency of the future looks like
- How the TRIPS Framework by Trust Insights can help you navigate the transition
Transcript:
What follows is an AI-generated transcript. The transcript may contain errors and is not a substitute for listening to the episode.
Katie Robbert – 00:35
Well, hey, everyone. Happy Thursday. Welcome to So What? The Marketing, Analytics and Insights Live Show. I’m Katie. This is Chris. This is John. All three of us are together.
I was just looking at the calendar, and we have now hit that point of the year where at least one of us will be missing from all shows moving towards the end of 2026, mostly for events and speaking.
John Wall – 01:02
So this is it.
Katie Robbert – 01:04
Enjoy it while it lasts, folks. Just kidding. We’ll be back in 2027 when it gets quiet.
Today, we’re talking about defining the agency of the future. This week on the podcast, I talked about how AI is changing the notion of billable hours. This is not a new conversation, but it keeps coming up. The more work that AI is taking from humans, the more agencies who use the billable hours structure are struggling to figure out how to bill for 20 hours when it took five. It’s very unethical to just make things up.
Now, the flip side of that is that a billable hour is already made up anyway. This predates AI, and I think all of us on this livestream have been on both sides of the billable hour.
Katie Robbert – 02:01
One side is being responsible for it for your team and making sure we’re filling the retainers. The other side is: what the heck did I do for a whole hour? Well, I sat through a bunch of meetings and nothing got done, but I guess I have to bill for it and fill up the whole hour.
That was the setup of the podcast. What I shared on the podcast was that it’s really an opportunity to take the time back. If AI does the work in five minutes and you have 55 minutes left, how deep can you get into the insights? What other data can you be looking at? What other meaningful, productive, and actionable things can you be thinking about?
Chris, you listened to that and said, “Wow, that’s super optimistic.”
Katie Robbert – 02:46
Before we get into your viewpoint, for what it’s worth, Chris was traveling, so it was just me. We didn’t get the other side of the conversation.
John, I want to know: what has been your experience with billable hours, and what are your thoughts on what’s going to happen to them?
John Wall – 03:05
For me, it’s always been one of those things where billable hours is kind of just enterprise. It’s a big company; maybe you’re buying legal services or super complicated software. The key has always been that you have the infrastructure to keep track of that. If somebody comes to you with 500 hours for last week, you have somebody who’s going to go back and say, “Okay, yeah, this smells right.”
This works for everybody in the mid-market, small businesses, or startups—they just need to get stuff done and they may not even have any of the expertise to analyze those bills or know if they’re getting ripped off. So they should be paying by the scope or the project. That’s kind of the general thing.
John Wall – 03:47
So for most of the world that I’ve stayed in, billable hours is just asking to get ripped off. That’s pretty much the punchline.
Katie Robbert – 03:58
Chris, thoughts?
Christopher Penn – 04:00
Brains or boots? When you hire an agency, you’re hiring for one of two things: you’re hiring for brains that you don’t have—people with skills outside what you used to have—or boots on the ground.
John Wall – 04:12
Right.
Christopher Penn – 04:12
You just have so much work to do. You need more people in the kitchen doing stuff because there’s just not enough of you. This is a familiar situation to the topic of the podcast this week.
AI provides both. AI provides a ton of boots on the ground, particularly today’s AI—and we’re going to talk about this in a bit—and it provides a ton of brains, too, particularly in skills and things that you don’t have.
We’ve talked many times on the show about tools like Suno, for example, that creates above-average AI-generated music. Is it going to win a Grammy? Nope. But is it better than your cat just clawing on things? Yes. Or your two-year-old with a spoon and some cans?
Christopher Penn – 04:57
AI will do a credible, capable job at most of the tasks. Now, when I listened to the podcast this week—in fact, Katie said that I said that’s a really optimistic perspective—I said, “Okay, I would love to talk about this on the livestream.” But first, as John and I talked about earlier today in an episode of Marketing Over Coffee, it’s nice if you have an opinion, but if you don’t bring any data, it’s not really worth anything.
So I said, “Okay, what can we go and provide for the brains, the boots, and the bills?” In terms of the brains, one of my favorite sites out there is Artificial Analysis. This is a company that independently benchmarks AI models.
Christopher Penn – 05:47
They take various prompts or agentic workflows, try to make AI tools do things, and measure how well they do those things. One of the things they came up with was a capabilities index. They have their mass index, but they also have a capabilities index with six fields: Finance and Accounting, Healthcare and Medical, Legal, Economics, Engineering, and Strategy and Ops.
Strategy and Ops is general business and stuff like that. One of the things people like to talk about with AI is, “Oh, well, it can write press releases, but I’m the account director on this account; I’m the real strategist. I know the strategy that no AI can replace.”
Christopher Penn – 06:27
In the Strategy and Ops index, we can see all the different models and what they do, and they tell you what the benchmarks are that make it up. So I said this is a great proxy for the brains that AI can bring. If we could figure out over time how smart AI models are getting at stuff like Strategy and Ops, we can calibrate against that. Does that make sense so far?
Second, I used PR agencies, first because you and I worked at one for five years, so we know that world pretty well, but we also know some of the folks in that world and can speak to how they work. One of the agencies in that world we know of is called Gould Partners. This is an M&A firm.
Christopher Penn – 07:10
They helped broker our old agency’s sale to their new corporate masters. Rick Gould, who’s a very nice guy I’ve talked to a bunch of times, has these benchmarking reports that they publish every year. The benchmark report for this year said, “Here’s what agencies are billing at per hour by role. Here’s what their P&Ls look like.”
Again, this is PR firms specifically, so this is not applicable to everything, but at a fundamental level, these are what the bill rates in 2025 were for PR agencies. So the CEO, someone like you, Katie, on average bills out an average of $482 an hour, and the list goes down from there. So this is the boots. How much do the boots cost?
Katie Robbert – 07:58
Makes sense.
Christopher Penn – 08:00
The third thing we wanted to get was how much of a role AI could play. For this, we used the TRIPS Framework by Trust Insights. As a representative sample, I went to Allison, which is a global PR firm, and grabbed all their job descriptions for all the different levels.
By the way, I would just like to point out, if we look at this Account Coordinator, look at these responsibilities: participate in team and client calls and take notes and action items; develop the first draft of written content—they’re hiring for this now; monitor and flag client industry media coverage; conduct entry-level media relations with instructions from an AE; research speaking opportunities, events, and awards; produce monthly reports; be good at media tracking; conduct new business research.
Christopher Penn – 08:51
There isn’t a single responsibility here that a machine can’t do—not a single one. This position itself does not need to exist, other than one person to push buttons and say, “Agent, go. Agent, go.”
Katie Robbert – 09:05
That’s the disheartening thing for a lot of people: this is the reality. I agree. I look at this and there’s nothing on here that a pretty well-prompted large language model couldn’t do for you.
I think that was my whole point on the podcast: I understand there’s a lot that the large language model can do, so what does the human have the opportunity to bring to the table? That’s not listed here, and that’s a huge missed opportunity from this agency for this level.
Christopher Penn – 09:42
There isn’t anything for a human to do. Literally, this entire position does not need to exist because this is the lowest-level position in the organization. Other than to promote and uphold core values, there’s nothing here that a machine can’t do better, faster, and a heck of a lot cheaper.
Katie Robbert – 09:99
No, I understand that. But typically—and this is predating AI—we didn’t put critical thinking in job descriptions. We typically put in the very tactical button-pushing things, like “must be proficient in Microsoft Word and Excel” or “must produce eight reports a week,” or whatever it is. We never get into the responsibilities that are more of those critical thinking skills. We say “must be good at problem solving,” but what does that mean?
This particular agency aside, it’s a huge missed opportunity for anyone posting a job description to not really outline the expectation of you being able to use your brain.
Christopher Penn – 10:44
Yep. So if we marry these data sources—the TRIPS analysis that I did from Allison’s roles, plus Rick Gould’s great research about the PR industry, plus Artificial Analysis—what we end up with is a lovely visualization that looks like this. Here is essentially late 2025 and what AI was capable of.
As I drag this slider, you’ll see it go from assistive—meaning you would chat with the AI—to agentic, where there are workflows, to fully autonomous, where the machine just does things end-to-end. We have the percentage of the billable hour based on TRIPS that goes away as AI capabilities increase.
PR agencies—and again, this is limited to PR for this particular visualization—have a target net margin of approximately 15%—
Christopher Penn – 11:43
No, 17% to 21% is where they try to be, especially on the higher side, because at that point, a holding company considers them worth coming in and buying. I remember back in the day when we worked at our old agency, the goal was to, by hook or by crook, get to a 21% net margin so that the owner could get the maximum dollars during the sale, which they did, mostly on the back of our team.
Katie Robbert – 12:07
Not salty, but here we are today.
Christopher Penn – 12:12
So let’s watch what happens over time. This is late 2025, Q4, and we start to see a divergence as we get into early 2026. The three lines are hourly billing (what tasks fall under an hourly billing agreement), retainer agreement tasks, and value-based billing—what you talked about on the podcast, Katie, about the different ways to bill things.
As we get to the spring of 2026, look at what happens here: retainer-based billing goes to the zero-profit line. To your exact point, Katie, when a client comes to an agency and says, “What are you billing me for? I’m spending all this money per month, what am I getting?” We’ve heard that many times over the years.
Christopher Penn – 13:05
“What exactly am I paying you for?” And even hourly billing.
Now, this is as we move into agentic AI. Agentic AI is where these advanced, multi-step workflows have a human sign off. This is like Claude Code or Claude Cowork. This corresponds well to where things were in the middle of this year, back in May or June, when all of 2026, Katie, you have spent all-in on the Claude ecosystem and done some amazing stuff with it.
Katie Robbert – 13:43
Quick plug: if you’re going to be at MAICON in Cleveland the second week of October, I am doing a Claude for Business workshop. I will also plug that if you want your own Claude workshop for your team, you can reach out to that guy down there in the blue shirt.
John Wall – 14:00
Hello, hello.
Katie Robbert – 14:04
To that point, you’re absolutely right. I’m trying to be mindful of not replacing the thinking that I do. Really, what it is: what was the analysis that I wasn’t getting to because of my workload and being one person that I can now do? I still have to think about it; I still have to have the inputs.
I can have Claude build me a presentation deck for an event, but they’re not paying us to have me use Claude to build a presentation deck. They’re paying for my expertise that goes into the slides and into the talk, and that’s not something I feel comfortable having a large language model replace. So my expertise and my brain are still very much happening.
Katie Robbert – 15:00
The amount of work I’m able to produce has increased exponentially—more than 10x—because a lot of the work that I do includes data analysis that I just didn’t have the time or the skill set for.
Christopher Penn – 15:17
Yep. And we can see at this point mid-year, the roles that are getting affected are basically from that management role down: 17% of billable hours gone, 20% of billable hours gone, 23% of billable hours gone.
As we start to move later into this year, you start to see those roles turn yellow. You’re now talking 30% to 40% of their billable hours vanishing. As we get to today—autonomous AI end-to-end—we’re talking 71% of billable hours gone. Even at the top, at the President/CEO level, 12% of billable hours are gone.
The only thing that gets kept is value-based billing, where you are paying for something that no one else can deliver.
Christopher Penn – 16:05
If you are paying for boots, that’s gone. Even if you’re paying for brains, a good chunk of that is gone.
Katie Robbert – 16:13
We’ve seen this with the idea of the digital twin. Not that I would encourage this because it’s actually not very ethical, but there’s enough publicly available information from certain experts in fields where you could put together a decent approximation digital twin in your large language model for someone based on what they write about, what they speak about, and their opinions on things.
You’re not going to get everything, but you’ll get a decent amount. You could squint your eyes and say, “Oh, now I know how Chris Penn thinks. I have a digital twin of Chris Penn. Why do I need to hire Chris Penn?”
Katie Robbert – 16:55
Now again, it’s not going to be as good as the real human version because there’s so much nuance missing from what you publicly put out versus what you keep behind the closed doors of your own IP. But that’s where I see some of this falling apart as well: people are going to say, “Why do I need to hire a fractional CMO when I can just build one?”
Christopher Penn – 17:21
Exactly. Now we get into the contrast in Gould’s world. Rick published an opinion piece on how AI is reshaping PR firm valuations, and Rick’s conclusion there was that strategy is fine—no one’s going to take away strategy. Yet we know from Artificial Analysis that strategy is not fine.
In our tool here, when you toggle back and forth, you can see that in that disagreement, strategy—which forms a lot of these role parts—eats into more billable hours. AI starts taking away more billable hours.
When we look at the specific task list of all the tasks by role as I drag the slider across over time, we see the things that get held vanishing, things that are going, and ultimately things that are just gone. These are all the tasks that are completely gone.
Christopher Penn – 18:18
Log media items and issues into the tracking system, write call and action notes, review utilization, billable hour reports—all of that, gone. This is available on the Trust Insights website. If you’re not a member of our Analytics for Marketers Slack group, go to TrustInsights.ai/analyticsformarketers, and we’ll put the URL in there after this livestream.
The other thing you can do, using the TRIPS Framework, is put in a task and evaluate it yourself: “How much time does this task take me? It takes me a lot of time. How important is the task? Not particularly important. Do I have sufficient examples? Yep, I sure do. How repetitive is the task? Very repetitive. How painful is this? It sucks.”
Christopher Penn – 18:56
You can see that by doing that—and I have the slider all the way to the right—that task has been eaten by AI somewhere. It is incumbent upon you to figure out, if this tool says AI can do it, maybe I should see if I can have AI help me with it.
Katie Robbert – 19:18
This opens up the conversation for a lot of companies. There’s always a lot of discourse about the use of AI. What I’m seeing on social media is leadership wants everything in the red—they want AI taking all of these things and they want to pay less money to humans.
The humans are pushing back, saying, “No, I’m not going to let AI take this thing. I am better at this than AI. Let me prove it to you. I can build a slide deck faster and more accurately than AI can because AI doesn’t know what I know.”
Katie Robbert – 20:04
This is a good representation of what’s possible, but I don’t feel like it’s reality for a lot of companies. Whether it’s because there’s pushback from employees or because they just don’t have the capabilities to give AI all of these things, this isn’t actually what’s happening.
What we’re seeing is more of a hybrid: AI is doing some things, humans are still doing things, but there’s no process on either side, so billable hours are still a nightmare.
Christopher Penn – 20:38
Billable hours are definitely still a nightmare. If we step back and think about it—I’m going to sound like AI, I swear—the work was never the thing.
Katie Robbert – 20:49
Oh, you do?
Christopher Penn – 20:50
I do. What an agency provides is a neck to choke and a person to blame when things go badly. Agencies are an accountability shield for a lot of people in corporate, so it doesn’t really matter in a lot of ways whether the agency does anything better than a half-assed job, as long as they’re not going to get you fired and you can blame them when things go badly.
AI cannot be that accountability shield. If you are paying an agency to cover your assets, you will still pay an agency for that. You will not stop paying billable hours if you need someone else to blame.
Katie Robbert – 21:43
That’s not an AI problem. Almost all the issues with billable hours predate AI. I think it’s in here somewhere—where is it? Down here: “New tech doesn’t solve old problems.” I saw that earlier and thought, “I gotta use it.”
It’s true. We talk about how AI is not solving problems; it’s amplifying things that are wrong within your organization. If there’s a lack of process, agreement, or measurable outcomes—and in this case with billable hours, if there is no clear standard of how we do billable hours or what is billable—then you don’t have a good proxy to say, “Well, now AI is doing this, so that’s the one-for-one swap.”
Katie Robbert – 22:32
It’s still a mess. On the podcast, I used legal as the industry that sets the stage for billable hours because that’s how they operate. There is no value-based option that you can buy. Generally, if you’re engaging a lawyer, you are engaging them hourly for their expertise and their ability to know stuff that you don’t.
A lot of law firms are now using AI to do some of that research—work that a paralegal or junior associate might have done in terms of preparing everything—so that the senior lead on the account can say, “Yes, this is how we’re going to proceed,” and then go argue it in court.
Katie Robbert – 23:23
You’re not necessarily going to get AI to replace a human to go to court and argue. People love to argue—why would they want to replace that with a machine? But you’re going to get a machine to replace a lot of the research that goes into building that argument.
Christopher Penn – 23:38
One of the things that the folks at METR (metr.org) said earlier this year was that your average agency is really an inverse pyramid.
The standard pyramid model has a few senior people and a whole bunch of junior people, and you bill by the hour because you’re billing for brains and boots. Boots are the easiest to bill for—a blended rate of $279 an hour. The more people work and the more hours they spend, the more revenue you make.
The inverse pyramid in value-based billing says you have more people at the top and fewer or no people at the bottom because machines do that stuff. That judgment and expertise is what you’re paying for.
Christopher Penn – 24:24
If you think about it, in a lot of ways, that’s how Trust Insights is shaped. We have a bunch of senior people, one junior person, and a bunch of machines. We’re never going to have that layer of 100 associates all churning away copy-pasting things—that’s never going to happen. It’s going to be machines.
That’s what the agency of the future likely looks like. Besides being an accountability shield, humans are particularly good at ideation. (The law also supports human-created works as having copyright, which is a tangential thing). Machines are good at “new to you,” but not net new.
Christopher Penn – 25:15
If you are sitting there asking how you can come up with creative ideas for your next campaign, with AI, you’re going to get stuff that has been done in some fashion before. In most situations, going back to your point, Katie, people are not going to have the sophistication or systems to force a machine to truly ideate something net new.
There is still value in having a human ideate. Then the machine can help you refine ideas, find blind spots, and things like that. That supports the inverse pyramid structure of a few big brains and a bunch of machines.
Katie Robbert – 25:54
What do you think, John?
John Wall – 25:57
Yeah, it does make sense. It’s just how much of that hour model was just turning the crank—doing stuff that agents can make completely go away. If you’ve been digging ditches with a shovel, I got bad news for you: this bulldozer is gonna ruin your entire career.
Katie Robbert – 26:23
That brings up the question that every agency is struggling with: what do we do instead? If we were previously employing four account coordinators to do research and drafting things, and that was part of the billable work, and now one single large language model is doing that, how do we account for all of that revenue loss?
John Wall – 26:52
Well…
Katie Robbert – 26:55
I’m not saying we have to solve it here, but that’s why we talk about it—that’s really what the question is. It’s not “Can AI do it?”, it’s “What do I do about all the revenue I’m losing because I can no longer bill for this work?”
Do I just not use AI so that I can keep the billable people? That’s one solution, but then you have the other side of that conversation: the expectation around speed and volume. Companies don’t really care whether or not you’re using AI anymore; they just want it, and they know it’s possible. You can’t bill them for work you’re not doing, yet you have to produce more. How do you reconcile that? It’s the whole “good, fast, cheap” triangle.
Katie Robbert – 27:43
I feel like that has now become a dodecagon or something with all the different factors.
Christopher Penn – 27:51
The fundamental premise of value-based billing is that your deliverables are priced by their economic impact. If you are good at measurement, analytics, and understanding what your work means to a customer, then you can price it accordingly.
If I give you this thing and you use it to 2x your profits—and you go from being a million-dollar company to a $2 million company—then I could say, “I’m gonna charge 5% of that as the bill for this thing because you paid in 5% to get 100% more revenue.” You have a 20x ROI, you’re happy, you’re gonna get promoted, and you look great. The challenge for a lot of firms is they don’t know what the economic impact of their work is.
Christopher Penn – 28:40
This has always been true in public relations. New tech doesn’t solve old problems. What economic value does a public relations firm drive? If you’re paying Allison, Edelman, or Weber Shandwick $50,000 or $100,000 a month, can you say, “For every dollar we paid in, we got $25 out”?
If you can say that, then whatever they charge you, as long as that math holds, you’re going to be happy paying it. If you hire Trust Insights, pay us a dollar, and we give you work where you get $2 of value out, you’ll put a dollar in the machine as fast as you can.
Christopher Penn – 29:21
Because then more dollars come back to you. If you can’t prove that, you’re going out of business—that’s the harsh truth. If you can’t prove the economic value of what you’re doing, no one’s going to hire you because you’re dead weight.
Katie Robbert – 29:40
How does an agency start to think about that? We don’t have to give them a formula, but for an agency still using a billable hour model while AI expectations are on the table, how do they start to think about that economic impact? What should they be looking at to be able to say to their client, “Here’s why we can justify you paying us $50,000 a month”—which still makes me cringe a little bit, because that’s a lot of money.
Christopher Penn – 30:18
Our team earned that back in the day.
Katie Robbert – 30:20
I know. One client, I know, which was amazing. But let’s say I’m the CEO of a company—let’s just play pretend for a second.
John Wall – 30:37
Shaq is a basketball player. All right, here we go.
Katie Robbert – 30:43
I don’t have the energy for big imagination today, so here’s where we go. Say I’m the leader of a PR agency, and suddenly I see our retainers dropping by half or by thirds—more than we’re comfortable with—because clients are saying, “We know you’re using AI, so why am I going to pay that?”
How do I start to think about how to justify that economic impact? What should I be looking at, what should I be talking to my team about, and what should I be saying to my clients?
Katie Robbert – 31:26
We’re just coming up with it right now—there’s no magic formula.
Christopher Penn – 31:31
To be clear, we actually did this back in the day at the old shop. We were called in on the carpet as an agency by one client, and the VP on that account called me in a panic and said, “You need to help me prove that I’m worth what they’re paying us and save this massive account.”
What we did was we took a couple of years’ worth of the client’s data—like Google Analytics data—back in the days when that data was reliable, and whatever else we could get our hands on.
Christopher Penn – 32:17
We looked at the economic value of a website visitor following all the way through the funnel, using the KPI maps that we’ve since taught, and then did uplift modeling. We knew when this team started working with you, what things were like before then, and what would have happened without hiring us.
What is the statistical difference between the treatment period—when you hired us—and the control period, when we didn’t work for you? What is that delta in those measures? Because we knew that a website visitor to this company was worth $17, I could measure that as an outcome based on what this team did and say, “This is what this is worth.”
Katie Robbert – 33:12
Yeah.
Christopher Penn – 33:12
Anyway, there we go.
Christopher Penn – 33:16
In that particular case, we were able to show with reliable math from the client’s own analytics, “Yes, what you are paying this team for is generating positive returns because this traffic goes here, and then it’s up to your sales and marketing team to convert that traffic into paying customers. We’re doing our job of getting people to the door. Once they come inside, you have to do your job and not punch them in the face.”
That’s how you would apply it today. If I work with a client and say, “Here is what you are trying to do—you want more of these people to the website, marketing qualified leads, whatever—and here’s when you started working with XYZ Public Relations firm.
Christopher Penn – 34:01
Here’s the delta in that measure that we agree upon, based on your own data: this is the value that XYZ delivers to you.”
The first thing I tell the CEO of XYZ is to stop hourly billing immediately and go to value-based billing. You say, “Here’s how we calculate our value: we calculate our value based on the economic value we create for your company. We will get people to the door; it is then your responsibility to get them in the door and sell them something. We will get the right people to the door.”
Katie Robbert – 34:33
I remember having this exact conversation with our good friend Gini Dietrich, who is the founder and owner of Spin Sucks. She’s had to do that justification many times over for what clients are paying for.
To your point, that’s exactly what she’s been able to demonstrate. She builds this into her engagements as a line in the monthly report showing everything brought to the client. When the question comes up, she’s able to say, “Here are all the qualified leads we brought you. It’s on your sales and marketing teams to close them, but look at what we did for you. Here’s where you started, here is your baseline before us, and here’s what we did since you engaged with us.”
Katie Robbert – 35:24
As you’re describing it, I remember doing that. That seems like something any agency should be including in their monthly reports proactively, whether or not it’s asked for: “Here’s what we’ve been able to do for you since engaging with our agency.”
I’m making a mental note for us to make sure we include that in everything we’re doing. It’s a natural question for the person paying to ask, “What am I paying for?” especially if there’s not a lot of transparency or if it’s not blatantly obvious that, since hiring Trust Insights, revenue has gone up exponentially.
Katie Robbert – 36:11
That may not be what you hired us to do, but it’s very black and white: did revenue go up or down? That’s not why we’re here in our case.
Christopher Penn – 36:21
Management consulting firms often have to do this. What tangible things do you have? “Oh, I have a box of these things that I didn’t have before, and I understand the value of these things.”
Earlier today, I was doing work for a client and pulled data from something I had seen out in the world. I grabbed the data, created a handoff to the client, and they said, “Wow, I’ve never seen this before.”
When the client stakeholder comes back and asks, “What did you do for us this month?” I can say, “I proactively made this thing for you that no one else has ever made, telling you exactly what’s going on with your AI visibility.”
Christopher Penn – 37:01
That is real data, and it’s something you can’t get anywhere else because it didn’t exist until literally 11 a.m. today. That may not necessarily have a dollar figure attached to it, but if the stakeholder asks, “What things do I have that I did not previously have?” we can credibly say, “You have never had this before. Now you have it and you can use it.”
Katie Robbert – 37:29
Now…
Christopher Penn – 37:29
If you don’t use it, that’s your fault. We can provide enablement materials to show you how to use it, but you still have to do the work.
Again, that goes back to value-based billing. I didn’t count how much time it took me, and frankly, nobody cares. What counts is: if I give you this thing, does it give you an advantage that you previously did not have?
Katie Robbert – 37:53
A lot of this is foundational: the KPI map. What KPI specifically were you able to influence? If you take the very top of that KPI map, it’s likely always going to be revenue or sales—something tied to money. Then you start to break down what influences revenue and which KPI Trust Insights was able to influence.
There’s a trickle effect: as you bring it down to the bottom of that KPI map, there are more things that start to influence each number as they roll up. If you can influence even one small part of that, there is going to be a trickle effect and you should be able to track that.
Katie Robbert – 38:38
It really comes back to: can you measure the work that you’re doing? That is not an AI thing; that is a general best practice for any agency.
Christopher Penn – 38:47
Yep. Agencies need to get a handle on this right now, today, because the pace at which AI models are improving is going to get much faster very soon.
This was a paper that came out from Shanghai Jiao Tong University last week, representing 13 different AI labs in China, called “Toward Genuine Recursive Self-Improvement.” This is a methodology Chinese AI labs are using to take massive leaps forward in agentic capabilities where tools can autonomously improve themselves at any given task.
Christopher Penn – 39:34
Not just getting smarter generally: given a task like writing a press release, the system is capable of not only building its own tooling, but also building a self-improvement system around that tooling to get better results without human intervention.
What you’re seeing now is going to show up in the pipeline in the next three to six months. As you get models like Qwen 4, ByteDance Seedance 3, or Hunyuan 4 from Tencent, you’re going to see this set of capabilities showing up in systems. Google just released their version of this paper a few days ago, so Google is working on it in Gemini.
This matters because if we go back to Artificial Analysis and these AI capabilities—
Christopher Penn – 40:27
You’re going to see these charts go up and to the right for all these different areas. Brains and boots are going to be less of a shield for your revenue than they are now. As we saw in the visualization, that’s already going away.
You have to get a handle on value-based billing and outcome-based billing—which is really what we should call it—as quickly as possible. Because if you don’t, these tools are going to eat your lunch.
Katie Robbert – 40:56
If you can go back to that academic paper for a second—as a side note, I think this is the first time I’ve personally seen an academic paper include a “too long; didn’t read” statement. When did that happen?
Christopher Penn – 41:14
Let’s see.
John Wall – 41:15
That’s the executive summary.
Christopher Penn – 41:17
Yeah, right, the executive summary.
Katie Robbert – 41:20
That’s wild to me, because I worked in academia for a long time and that was never included. The abstract was meant to be good enough, but now it includes a TL;DR.
Anyway, that’s an aside that took me by surprise, but I agree. Agencies still operating on the billable hour need to shift their thinking to outcome-based billing, because it’s only going to get harder, you’re only going to get more scrutinized, and you’re going to get more pushback.
Katie Robbert – 41:59
If you can’t say, “This is how I impacted this particular KPI on your map that leads up to your revenue,” you’re going to have a hard time justifying that $50,000-a-month retainer, and that’s going to be a hard hit when you lose it.
Christopher Penn – 42:16
I would just like to—
John Wall – 42:18
I was just going to say, this is a drag race. There are multiple firms out there, and the firm that gets there first can cut prices and basically pull the rug out from under anybody else.
Christopher Penn – 42:29
I would just like to point out we have been saying this. I have a blog post up from October 2023, “Stop Being an Hourly Knowledge Worker Business.” We have been warning you for three years now: get away from hourly billing.
John Wall – 42:48
They just don’t listen.
Christopher Penn – 42:52
Which means the things we’re warning you about now in 2026, by 2029 we’ll be saying, “We told you back in 2026 that recursive self-improvement was a thing and you didn’t listen to us.” And now Skynet runs everything and it’s just like WALL-E.
John Wall – 43:06
Where’s my drink?
Christopher Penn – 43:15
There we go. The TL;DR of this entire episode: stop hourly billing immediately, but with—
Katie Robbert – 43:24
—the caveat that you can do other things without overhauling your entire agency.
Christopher Penn – 43:30
Exactly. So, Katie, how optimistic do you feel now?
Katie Robbert – 43:37
I still feel optimistic. I feel like it’s just a matter of a little bit better planning upfront. It can’t be a livestream without the 5P Framework: Purpose, People, Process, Platform, Performance.
In this instance, as you’re bringing on an agency, it’s your job as the payer to think about why you are bringing on an agency and what they are meant to have an impact on. As the agency, your job is to proactively say, “These are the metrics that I know we are going to have a positive impact on.” Bring it back to the numbers and back to those outcomes, and you’ll have to worry less about the billable hour.
Katie Robbert – 44:22
Because to your point, Chris, if someone’s getting $25 for every dollar they spend, they’re going to spend more dollars to get that back.
Christopher Penn – 44:30
Yep, exactly. All day, every day. That is the win. But you’ve got to be good at measurement and proving your worth.
Katie Robbert – 44:41
You tried! You can’t kill my optimism, Chris.
Christopher Penn – 44:44
I know. I’ll keep trying.
Katie Robbert – 44:45
Keep trying.
Christopher Penn – 44:48
That’s going to do it for this week, folks. Thanks for tuning in. We will see 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.
|
Need help with your marketing AI and analytics? |
You might also enjoy: |
|
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday! |
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday. |
Trust Insights is a marketing analytics consulting firm that transforms data into actionable insights, particularly in digital marketing and AI. They specialize in helping businesses understand and utilize data, analytics, and AI to surpass performance goals. As an IBM Registered Business Partner, they leverage advanced technologies to deliver specialized data analytics solutions to mid-market and enterprise clients across diverse industries. Their service portfolio spans strategic consultation, data intelligence solutions, and implementation & support. Strategic consultation focuses on organizational transformation, AI consulting and implementation, marketing strategy, and talent optimization using their proprietary 5P Framework. Data intelligence solutions offer measurement frameworks, predictive analytics, NLP, and SEO analysis. Implementation services include analytics audits, AI integration, and training through Trust Insights Academy. Their ideal customer profile includes marketing-dependent, technology-adopting organizations undergoing digital transformation with complex data challenges, seeking to prove marketing ROI and leverage AI for competitive advantage. Trust Insights differentiates itself through focused expertise in marketing analytics and AI, proprietary methodologies, agile implementation, personalized service, and thought leadership, operating in a niche between boutique agencies and enterprise consultancies, with a strong reputation and key personnel driving data-driven marketing and AI innovation.
Newsletter
Podcast
