Using Data to Run a Better Agency

Sei-Wook (00:06.889) On today's episode, we're talking about how agency leaders can use data to make better business decisions and improve profitability without overwhelming the agency with reporting. We'll focus on practical, experience-driven ways data has helped surface real patterns around clients, engagements, people, and growth. Peter Kang (00:25.944) All right, so today we want to dive into the use of data in an agency. First, maybe we should talk about why — and we've been guilty of this as well — why does data feel overwhelming for many agency leaders? And why do agency leaders resist diving into the data? Sei-Wook (00:50.318) Yeah, depending on the agency leader, you can collect a lot of data, but the numbers and spreadsheets can sometimes feel overwhelming with the amount of inputs you're dealing with. And in the day to day, you have a lot going on at the agency. If things are going well, it can almost feel like a nice-to-have — it takes a backseat to everything else going on in the business. Peter Kang (01:19.298) Yeah, and one thing — you mentioned in the intro that a lot of times just because people collect data, they feel like they need to generate all kinds of reports. And this too we've been guilty of: you generate — whether it's weekly reports — you take the data and you think, okay, cool, we have some kind of reporting going, now we're somehow smarter. But it's actually not the case. And that leads to this performative analytics theater where you're like, now because we have data we're smarter all of a sudden. But really, this is all about pattern recognition. It's about how do you leverage the data to inform better decision making? Sei-Wook (02:05.804) Yeah, we've definitely fallen into the trap where it's a lot of work to put together these reports. The exercise of putting together the numbers feels like the end result, but that's just the start — you're supposed to take all that information and inform how you can run your business better. Peter Kang (02:23.565) Yeah. Maybe what would be a simple mental model for agencies to think about data? Sei-Wook (02:31.181) Yeah, taking a step back, a lot of agencies already have the raw data that's needed. Oftentimes people have access to financial data in something like QuickBooks, or a lot of agencies are doing time tracking, so they have data around how much time each team member is spending on a weekly basis on each project — which project, what kind of tasks. A lot of the raw information is there. But to what we were talking about before, how do you take all of those inputs and organize them in a way that you can actually review and reflect and make some sense of all the numbers? Peter Kang (03:12.566) Yes. So to that point, today we're going to dive into five practical lenses through which agencies can leverage data and inform better decision making. Let's dive into the first one. You take this one. Sei-Wook (03:27.149) Sure. The first one we're calling engagement level analysis. This is taking a look at the project engagement level — so if you have projects, retainers, campaigns, however you organize the work that you're doing. Taking a look at profitability metrics. Even just understanding what did you sell a project in for and then what were your costs to deliver that. If you have those inputs, you can take a step back and look at whether there were some trends — maybe the larger the project, the lower the profitability, or vice versa. Are there certain types of retainers where what you're selling in consistently goes over budget and you're always underwater? Or is there a type of project — let's say website development versus design — are there certain areas of the work that are more profitable or less profitable? Or is there a combination, where when you combine the two they're less profitable, but if you do them separately they're more profitable? Any insights like that you can glean from taking a step back and looking at everything holistically. Peter Kang (04:44.758) Yeah, basically analyzing these engagements — whether they're projects, retainers, or whatever — can give you insights into: are you offering the right mix of services? Which services are likely to be more profitable? A lot of this can be easily gleaned from really good time tracking data. But what happens to agencies that aren't into time tracking? How can they go about getting the data necessary for some of this analysis? Sei-Wook (05:13.452) Yeah, and we've seen this because not all of our agencies do time tracking in the same way. But there is the exercise of staffing. When you staff people on a project, you can look back and glean a signal on where people's time is going. You might say this person is working 50% on this engagement or 100% on this engagement, and then you can extrapolate at a high level what your cost was over a period of time. Generally speaking, it'll get you most of the way there — maybe 80% of the way there — and that direction gives you the right information that you need. Peter Kang (05:49.693) Yeah, it's very easy to understand: if someone's burning 100% for 20 weeks on a very low budget project, they're going to be over budget. So you can definitely glean stuff there. Just to round this out, I think a lot of this analysis at the engagement level informs a few different things. Pricing is definitely one of them, and the way you scope things is another. What are some other things it can inform in terms of decision making? Sei-Wook (06:21.233) Yeah, like I mentioned, if you can figure out some trend in services — what services should you double down on, what should you step back from? Maybe it's not so black and white, but maybe there's a reworking necessary on how you're doing certain types of engagements or certain service lines. And in some cases, maybe you need to just stop doing that type of work in your agency. Peter Kang (06:49.897) Yeah, definitely. Okay, the second lens through which data can be helpful is at the client level. This is taking a step up from the engagement and looking at the relationship with the client. At the surface level, the revenue and maybe the profit contribution of that client is important. But there's definitely more to this. What are some other data points that agency leaders could review? Sei-Wook (07:25.129) Yeah, with a long enough time horizon you can see client tenure — how long have they been with you? And with that, what's the lifetime value of that client? Not just looking at one project or one engagement, but over several years, how much have they contributed? And then from that broader look, on an annual basis, are they profitable? Can you see maybe that when you first started with them three years ago you were very profitable, but over the years you've had some margin compression and now you're working at a loss or at a smaller margin with this client? Looking on a longer time horizon to see if there are any good trends there. Peter Kang (08:12.864) Yeah, the annual billing — I do think if you have client relationships that last more than a few years, it's always interesting to take a look and glean insights, because sometimes depending on your service offering there's a shape to your engagements. On something like website design and development, you typically have this upfront big project — a redesign, migration, or whatever — and then it tapers into ongoing maintenance and support and maybe optimization type work. But then you might have other relationships where you have multiple spurts of campaigns, where some years you might have a really big billing peak and others it's more chill, and between seasons too. It's very interesting to analyze because then you get a sense of, all right, these are the types of clients that we have, and you can dig even deeper into that. Sei-Wook (09:15.69) Yeah, and actually on that point, it's interesting to think about — it doesn't have to be so linear. Let's say your business model is you might do an initial project where your margin is lower, but you'll make up a lot of the margin later on. That could work. It's just about understanding how the shape of the margin is and being intentional about retention in that model and how important that is to the business. Peter Kang (09:46.013) Yeah, and on retention, client satisfaction is super important. So there are signals there. If you're using different types of client survey feedback mechanisms — NPS, CSAT, or just forms with questions on how you're performing — those are data points to understand: are we doing a good job of servicing these clients, and what are areas we can improve in? And more importantly, what are the trends — where is our level of service as an agency? Sei-Wook (10:20.305) Yeah. And even thinking about it, there are so many different data points when you look at a client relationship. You'll get feedback from the team on how it is to work with this client, with this company. Is there any friction that always happens? Is there a lot of pushback at every turn? And from a finance standpoint, their payment history — sometimes if they're always holding their invoices and not paying on time, it's always a challenge to chase down payments. So these are aspects of a relationship where it could be highly profitable, but if there's a lot of friction there it may not be worth the struggle. Peter Kang (11:04.371) Yeah. And if you layer on top of all this data the client type — the basic stuff about the client, like what industry or sector are they in, how big are the clients, maybe even what maturity level is their business — are you working with a founder at a startup or are you dealing with a scaled organization with multiple stakeholders, maybe a director of marketing or a VP of Ecom or something like that? And then what kind of work are you doing with them? Is it a strategic relationship where you're coming in to do a lot of roadmapping and high-level consulting, or is it more execution level — hey, we're here to fix a very specific problem and get it done fast? If you layer on that categorization across the client base, all of a sudden you can start to segment the clients more easily. Sei-Wook (12:11.079) Yeah, definitely. There are so many data points around client engagement — we talked about the financial aspect, the relationship, et cetera. And it's about focusing on what's most impactful. How do you get 80% of the signal to determine what to focus on? Client profitability is an important one to look at: of all your clients, which ones drive the most profit and how is that profit shaped? And from a team perspective, do certain clients take a lot of time or energy from the team to deliver that revenue? It's a balance — there could be downstream impact where a high-profit but high-stress client may not be the best. Peter Kang (13:09.245) Yeah, and what you're talking about reminds me of something common when it comes to data — the Pareto principle, the 80-20 rule, where 20% of something is probably responsible for 80% of the results. So 20% of clients may be responsible for 80% of the headaches, 20% might be responsible for 80% of profits, and 20% might be responsible for 80% of the most exciting work you could showcase. It just shows up everywhere. That's why segmentation is very important — you can identify and surface this dynamic across your analysis and it makes this work that much more insightful. Sei-Wook (13:57.894) Yeah, the goal of all this data for clients — what's the end result? Really refining who you're targeting, what your ICP is, what type of client in terms of industry, size, category, and who the stakeholder is — all of that helps you define who you're working with. And from that, what are the behaviors you can have internally to retain those clients and have long-term engagements with the people you want, so you can increase that percentage — not just 20% delivering 80% of the profits, but trying to grow that. And for the team, it's thinking about how you improve the experience for those clients. What does the delivery look like, taking feedback from NPS and any qualitative client feedback, and how do you improve your delivery to ultimately get that positive mix of clients everyone is working with. Peter Kang (15:05.712) Yeah, that's good. I'll do this next one as well. The third lens is the employee level analysis. This is about looking more specifically at your team members — what's the data around each employee and what can you gather from there? This is not to point out anyone for blame, but it is to understand performance — whether it's high performers or folks that are struggling to perform — and having data around that is going to sharpen your decision making. We've had a lot of experience thinking about the different data points here. We started with our gut, but we've become more sophisticated over time, hopefully. What are some inputs that folks should consider? Sei-Wook (15:59.815) Yeah, one is the most formal of performance reviews, where part of that we've done peer reviews — people you work with provide feedback on their experience working with that person, as well as a self-reflection and manager reflection on performance. That gives you the most structured way to look at a team member's performance. And then even looking at the engagements they were staffed on and seeing were they profitable or not, did the clients have a good experience working with that person. Just gathering more information about how they're working on the team. Peter Kang (16:44.528) Yeah, and time tracking is something we're always considering as well. Let's say we take these and whatever other inputs we might get. A couple of patterns you might want to look at: how consistent are they on these engagements? On the negative side, are they continually missing deadlines, are there quality issues? On the flip side, somebody who's always hitting deadlines, moving things forward — and over time, this is where you do get a sense from your gut, but the data helps to confirm or not: if you staff certain people on certain projects, it's going to get done and done well. Those are positive signals, and if you have data to back that up, that's even better. And for folks where, for whatever reason, deadlines are always getting missed, projects are always running over budget, there's always some kind of issue — the data really helps you understand those things. Sei-Wook (17:56.047) Yeah, really good. And it's definitely a balance because from a quality standpoint people could really like working with somebody, so peer reviews are always like, I really like this person, we're friends and we enjoy the experience. But you've got to balance that with the numbers. That can be true, but there could be delivery challenges — because of that or some other reason, projects always go over budget when a certain combination of people work together. So it's not just a popularity contest. Peter Kang (18:35.175) Yeah, and it is a fine balance and you do have to be careful about it — there are tough decisions to be made. But sports does provide a good analogy to all this. When you think about sports, there are a lot of statistics around performance, and these days with different sports there's advanced analytics, so you can see: when someone's on the court or on the field, what's the plus-minus impact? When they're playing for your team, is the team expected to score more, perform better? Or when they're on the bench, is it better or worse? And that player might be very popular with the team, but ultimately as a general manager or owner, you have to make the tough call of: do we want to win, or do we want people to be happy and friendly with their team members? It's very tough to make that call, but data does help make a stronger case. Sei-Wook (19:43.394) Yeah, ultimately for employee level analysis, it's things like coaching — how do you help this person improve in these areas? You get all the inputs, the feedback, and if there are a lot of positives and there's a way to help this person grow or address very specific challenges, or even maybe it's not the right role for them — maybe they have strengths in different areas and there's an opportunity for them to take on a different role and responsibility within the agency. That's an opportunity using all this data. Peter Kang (20:22.88) Yeah, and like we said, it's not always to spot the ones that are struggling — it's also to identify the stars and high potential employees. In many respects, a lot of your attention should go to those folks around professional development. If you spend all your time on underperforming folks, you're actually going to lose your stars. You should try to flip it around, because once again the 80-20 rule strikes again — 20% of your team most likely is driving 80% of the results. It's just the fact of these power laws that happen everywhere. It's very important to use data to identify that as well. Sei-Wook (21:00.49) Yeah, that's totally true. And my point about the role change especially applies here when there's a role your high performers can grow into — that's something to look out for. I'll move on to the next lens, which is new business analysis. When you look at all of your new business stats, you're looking at win rate, you might be tracking the number of leads that came in. But there's a lot of depth you can go into here. Taking a step back and looking at years of data around what types of prospects came in — whether that's by industry or role, or even taking a look at their budgets. Are you seeing any trends by company size or industry and the budgets they bring in? Some other ones are what services they're looking for from you — which could be a signal on how they came in or what your reputation is in the industry — and speaking of that, lead source: how did they find out about you, is it from marketing or referrals, et cetera? There are a few more if you want to dig into the biz side of this. Peter Kang (22:30.982) Yeah, lead source is important because you could dig into that a lot. One thing I'll point out is I get really peeved when some of our agencies put "website" as the lead source, because that's not a source — you've got to think a little deeper. People submitted their form on the website, but they heard about us somewhere else. Usually: did they find us on search, what were the keywords they were looking for, or if it was AI, what was the prompt they put in to get the result which then led them to the website and then to the form submission? And then a lot of it is multi-touch — who else did you hear about us from? Sometimes there are partners that might have mentioned us, or they might have seen a LinkedIn post or something. It's important to capture all those things because lead source can go really deep, and like I said, it's a multi-touch journey a lot of times. The sales cycle length is super important too, because a lot of these engagements are months in the making, if not years. Understanding when was the first time you talked to them, when something of interest came up, and being able to track that and see it — and if you have a good CRM that's recording all the meetings, calls, and emails that went back and forth, that leaves a really good breadcrumb for understanding the sales cycle. Those are important data points to analyze. And you mentioned role — the buyer committee and decision dynamics are important because different types of companies have different types of buyers. These data points help you refine your ICP because there might be shifts over time. You might have started with smaller clients where you're generally talking to the founder or a CMO, but then as you move up market you might be talking more with a director of Ecom or something like that. Understanding what those trends are and who gets looped into the decision making. And these days with AI recording apps, you're bound to get a ton of data from calls. Call transcripts are really interesting ways to analyze and glean reasons for winning or losing deals, because sometimes you'll get feedback from clients and also pick up things between the lines that might be worth revisiting. Sei-Wook (25:22.835) Yeah, the win-loss analysis is one that's interesting on both sides — number one is just asking and getting that insight. And especially for losses, they may not be as transparent about the reason why they selected another partner. But jumping on a call and taking that input and seeing if you can gather some bigger trends you're seeing for losing. Peter Kang (25:47.383) Yeah. And this type of analysis, especially on the new business side, does bleed over into the client level analysis we talked about before, because for the wins, if you want to dig in a bit more, you want to then go: what is the long-term relationship quality? That is definitely the client level analysis we talked about. What were the subsequent engagements, the way that account expanded — were they good or bad clients over time? So when you think back on your wins, you can understand: did we go after the right kinds of deals, did we win the right kinds of deals? That type of analysis is super helpful. Sei-Wook (26:34.044) Yeah, and especially if there are any signals you could have seen earlier on — of clients, good or bad — that these types of relationships or companies or stakeholders ended up being really positive or really negative years down the line. Peter Kang (26:50.666) Yeah, and if you're not already doing this within your agency, a quick way to get started — and we found this helpful — is if you just take the previous year's wins and go down the line and ask: looking back now, a year later, would you go after this deal again? Why or why not? Just having a qualitative conversation, maybe with your leadership team or just personal reflection on it. That yields a lot of insights, because you'll quickly find, oh my goodness, even during the process, down in my gut, I knew this was a bad fit because the founder had really high expectations with very little budget — all those things come screaming back. And it'll just help you strengthen your qualification focus later. Sei-Wook (27:44.126) Yeah, definitely. So all of this informs qualification, obviously a very important area where you can be refined. From a marketing and ICP standpoint, are you targeting and going after the right people? From a sales process standpoint, how can you get even better at tracking the right things? And from that, how can you forecast a little more accurately? Because ultimately the more you can try to predict the future, the better chance you have at planning for the agency. Peter Kang (28:19.208) Yeah, and just real quick on that sales process thing — a concrete example that comes to mind from some of our agencies: finding out from losses is instructive, and sometimes clients are open to sharing. One of our agencies was curious why they lost a deal, because they were finalists and the presentation seemed to have gone well. The client was very transparent: you guys were great, but during the process we never got to meet the team that was going to work on the engagement. The winning agency was really good about bringing those folks into the sales process and we got to meet them, so we were just a bit more confident there would be less of a bait-and-switch dynamic. That was a good signal, because now bringing team members into these proposals and sales conversations is a consideration. Sei-Wook (29:23.2) Yeah, and from there it's understanding that every prospect has a different set of things they care about. Part of the process is how do you glean insights early on to say, all right, this client really cares about team process — let's get other faces involved in closing. Peter Kang (29:45.204) Yep. Okay, let's do the last lens. We're doing five different lenses, but this is not the exhaustive list — there are so many other ways data could be used, and I just want to make sure we call that out. But this last one is about team sentiment and capacity. From a data perspective, it's about using data to get a sense of that sentiment. A lot of times you get some anecdotal things, and anecdotes are a type of data, but you can get over-anchored on a couple of stories. A good example: back in the day when we were running Barrel, we had an anonymous feedback form — anybody at the agency could just write whatever they wanted and we had no idea who it was coming from. We thought this was a good way to keep a pulse on the agency. But every now and then you'd get a submission where someone went on a pretty big rant about how things were terrible and wrong. We didn't know any better and we'd overreact, thinking, wow, this is how everybody must be feeling, we have to fix things, and we'd go pretty hardcore about that. Same with a few Glassdoor reviews — we'd overindex on, wow, we're terrible, we need to do better. But over time we realized we were giving too much weight to what were basically just one of many signals we should be considering. And that's where this idea of more data becomes important if you want to make better decisions. Sei-Wook (31:43.432) Yeah, along those lines of collecting more data from employees, we've used various employee engagement surveys. The latest was a survey system called Q12 — the Gallup survey, where it's a series of 12 exact questions that you have to ask in exact wording. And over time you can glean insights from how the team sentiment is and if there are specific areas that get better or worse. It covers things like, one question is: do you have a best friend at work? It's strange to ask a question in that way, but it makes people really think about how they define "friend" and how they define "best" — and that might give you a signal about something. Or how many times have you received positive feedback in the past X number of days? It's all these different areas around team sentiment. Peter Kang (32:47.715) One of my favorite ones from that is: do you have the tools or resources to do your job well? That's really key because I remember we learned, crap, some of these people don't have access to certain software or equipment to do their jobs. It's actionable stuff. Those are all really good data points you can act on. Sei-Wook (33:08.188) Yeah, it's a combination of that where it's very numerical and you can have a score. But then the commentary is really important, because if you just have the score, you don't know what the tool is and you don't know who said it — it's hard to take any action from that. Peter Kang (33:24.336) Yeah. When we talk about data for team sentiment and even capacity — thinking about how to plan better — it is stuff like performance reviews, because you're going to get signals about: this is how I'm feeling about the types of work I'm getting assigned to, this is how I feel about working with my team members, this is how I'm feeling about my workload. You're going to get a lot of that from performance reviews, especially if you have a really great system with your managers and employees and the relationships are solid — you can get a pretty transparent view there. Engagement debriefs help too. After a project, or during a retainer at check-in points — just understanding, what do you think you could have done better, what are some ways the team could have done better? You're going to get feedback that reveals a lot about the culture. And lastly, the resourcing data — whether that's time sheets or whatever staffing plans you might have — you can see if certain folks are overloaded versus others. You'll sense it before you look at the numbers, but it's good to have that quantitative data to back it up as well. Sei-Wook (34:43.324) Yeah, all of that can really get a sense of things like burnout risk — if there are specific team members or departments that are overstaffed for a long period of time, or even on the flip side, if you're understaffed and people have a sense there isn't enough work for me or my department, is there a risk they'll start looking around for new opportunities? Using the data and the tools to help predict what's going to happen with the team. And to your earlier point, are there any gaps in tools and training? People are usually pretty open about communicating that. And then communication itself — if the team isn't clear about the vision of the company or why certain decisions were made, it could just be a gap in leadership sharing and getting everyone on the same page. Peter Kang (35:44.389) Yeah, 100%. To wrap this lens up — so much of what drives employee satisfaction, client satisfaction, and team culture in an agency setting is how well and how effectively you staff the right people on the right kinds of engagements. The more data you could have to make better decisions here is, I think, one of the more high-leverage decisions you can make. And then there's also what are your investment priorities with your team? Do you need to hire more, bring in other folks, level up certain types of roles? Beyond that, are there team bonding activities to invest in — more team get-togethers or retreats? Do you need to provide more professional development in the form of coaching, training, team-wide training of sorts? All those things can be well informed through the data. And then, like you mentioned, how can you improve internal communication? At the end of the day, we know all too well: if there's a vacuum, people are going to project whatever the worst-case views they might have about the business onto that vacuum. It's very important as leaders to be proactive in communicating what's going on and where things are headed. Sei-Wook (37:29.979) Cool. So we talked about these five lenses, and there's a lot of data, as you can see, that agencies do collect probably already — and it may just sit there. I would say there are some caveats we should talk about around data. The first is overreacting, or just reacting to different signals based on time horizon. If you look back at data from the past month, but you realize the past month was a very busy time for the agency with a lot of new projects kicking off and a lot of uncertainty, that may skew how you're looking at the data and how people are reacting to scenarios. You've got to take a longer-term view and see if there are any trends you can identify. This is essentially recency bias — are there things that just happened that skew your vision of what the real story is? Peter Kang (38:38.735) Yeah, this is the sky's-falling dynamic, where you just had a rough few days or a week and you convince yourself the sky's fallen, like everything is going bad, things are going up in flames. In that instance you probably just need to take a few steps back and say, okay, these are some challenges we need to address but it's not the end of the world, and there's definitely a lot of positive outweighing the negative as well. Sei-Wook (39:15.864) Yeah, definitely. Trailing 12 and even multi-year views on all these signals are important to look at — try to get bigger overarching trends on what's happening across the agency. And even just thinking about tough economic situations, like tariffs were a big one we went through last year where clients put things on pause and there were a lot of macro conditions that affected engagements and led to changes in client sentiment, employee uncertainty, et cetera. Just be aware that one or two data points isn't enough to give you the information you need. Peter Kang (40:02.302) Yeah, I'd make a point about cycles. When times are good, you can overindex on that and make decisions that assume the good times you just experienced — and it could be a year, year and a half even, which feels like a long time in agency life — are going to last forever. The past 18 months have been awesome, it's going to be like this forever, and you make decisions with that lens. We fall into that trap too. It's actually better to stretch that out a bit more and understand there are going to be some dips ahead — it's inevitable for any kind of business — and how do you better prepare for that? Sei-Wook (40:50.425) And the most important thing looking back is you kind of need the data, right? The consistency and building the habit of data collection over time is really important here. You could be really busy for a period of time and say, we'll just skip that employee survey or skip that other thing. And suddenly you wait, and when you look back on it a year later you're like, wow, we really should have done that — we have a hole in our data set. That consistency over time is important. Peter Kang (41:25.155) Yeah, and it'd be remiss to not mention that since we're the AgencyHabits podcast — make it a habit. Data consistency as well as data quality: make that a habit. Sei-Wook (41:38.275) Yeah. Cool. You want to talk a little bit about how people should think about collecting data? Peter Kang (41:48.686) Yeah, so we're talking tooling, maybe some of the infrastructure. We could probably spend a lot more time specifically on this, but most simply, a lot of this can just start in a spreadsheet — exporting it from different tools and stitching it together in a spreadsheet as a start. There are definitely tools out there and SaaS products you could buy, dashboards you could build, and you could hire consultants too. There are fractional CFO and COO types that can come in and help support a lot of this stuff if you're really strapped for time or want that third-party perspective. A lot of that is out there. But going back to it, it's less about the tooling and more about: are you going to make the time and have the discipline to do the reviews? That's where a lot of this comes in. And with AI there's less of an excuse now, because you can get richer analysis teed up for you to think about. You can synthesize huge amounts of data that in the past might have been tough — think about loading up dozens of sales call transcripts and getting insights there, or analyzing various project team meetings and getting stuff there. You can get a lot more value out of the data. So there's no excuse to not make the time to think about it and analyze it. Sei-Wook (43:18.969) Yeah, and on the AI note, you really have to apply your judgment on top of it. You can't just dump the data into AI and say, hey, what do you think, what's the trend here, what's the sentiment here? Think about how you're prompting and suggesting and giving background context on all of this. It doesn't replace your judgment for how to look at and analyze, assess, and decide how you're going to take action from all of the data and analysis. Peter Kang (43:54.797) Yep. All right, we've covered a lot of ground today on the data front. Like I said, we could spend several more episodes on data itself. AI isn't the one making the decisions — you still have to do that. And agencies that make data review a habit are the ones that are going to learn more lessons quickly and avoid making the same mistakes. This is about embracing this as a habit and letting it compound over time. All right, thanks for joining us. Till next time. We'll see you. Sei-Wook (44:40.856) Thanks.