WEBVTT NOTE This file was generated by Descript 00:00:00.000 --> 00:00:03.279 Judith: Welcome to Berry's In the Interim podcast, where we explore the 00:00:03.279 --> 00:00:07.630 cutting edge of innovative clinical trial design for the pharmaceutical and 00:00:07.630 --> 00:00:10.069 medical industries, and so much more. 00:00:10.449 --> 00:00:11.130 Let's dive in. 00:00:12.200 --> 00:00:12.800 Scott: All right. 00:00:12.800 --> 00:00:15.660 So welcome everybody back to In The Interim. 00:00:15.710 --> 00:00:22.839 I'm your host, Scott Berry, and I have a, a, a guest today, a statistician, 00:00:23.240 --> 00:00:27.430 and, um, it's not uncommon on this show, we have other Berry's. 00:00:27.439 --> 00:00:28.500 So I, I 00:00:28.603 --> 00:00:28.933 Tim Berry: Yeah 00:00:29.109 --> 00:00:31.849 Scott: another Berry you've, you've probably not been introduced, 00:00:31.869 --> 00:00:34.799 but another statistician who is not at Berry Consultants. 00:00:36.039 --> 00:00:38.320 Uh, most of the Berry's you meet are at Berry Consultants. 00:00:38.330 --> 00:00:43.670 So my brother, Tim Berry, is a ho- is a guest today on In The Interim. 00:00:43.670 --> 00:00:44.440 Welcome, Tim. 00:00:44.953 --> 00:00:45.383 Tim Berry: Yes. 00:00:45.393 --> 00:00:46.133 Thank you, Scott. 00:00:46.213 --> 00:00:49.673 Uh, it's the first time I've been called a statistician in a while, so, uh, 00:00:50.134 --> 00:00:50.434 Scott: Yeah 00:00:50.483 --> 00:00:50.993 Tim Berry: exciting. 00:00:50.993 --> 00:00:54.103 And I happen to be actually right down the lake from Scott 00:00:54.113 --> 00:00:56.463 here, so, um, in a good spot here 00:00:56.662 --> 00:00:57.332 Scott: Yep, yep. 00:00:57.462 --> 00:01:01.032 Uh, you can almost see Tim's house through the window, uh, there. 00:01:01.422 --> 00:01:05.802 Um, okay, so let's, let's y- so you said it's a long time since 00:01:05.802 --> 00:01:07.201 you've been called a statistician. 00:01:07.682 --> 00:01:12.772 Um, let- let's talk about the career, his, a- and what more people 00:01:12.772 --> 00:01:15.372 call him nowadays, uh, in that. 00:01:15.572 --> 00:01:19.612 So your, your career as a statistician started, we were both at the University 00:01:19.612 --> 00:01:24.261 of Minnesota, and you got your master's degree at the University of Minnesota, 00:01:24.291 --> 00:01:29.481 and you you took a class from your father. 00:01:30.241 --> 00:01:30.771 that right? 00:01:30.949 --> 00:01:32.219 Tim Berry: Yes, that's, that's right. 00:01:32.739 --> 00:01:33.219 That's right. 00:01:33.219 --> 00:01:34.439 It was, uh, a weird… 00:01:34.979 --> 00:01:35.739 It was weird. 00:01:35.749 --> 00:01:42.279 Uh, I s- I, I saw my dad, Don, in action and, uh, I, you know, in, 00:01:42.289 --> 00:01:46.969 in, um, some of the sports he would coach me in, he would always be a lot 00:01:46.969 --> 00:01:49.849 harder on me in those sporting events. 00:01:49.879 --> 00:01:53.449 But, uh, actually in this class, he took it r- nice and easy on me, 00:01:53.869 --> 00:01:55.509 and so I was surprised for that. 00:01:55.569 --> 00:01:57.459 But yes, that was a great experience for me 00:01:58.102 --> 00:02:01.362 Scott: Uh, so I've, so I've worked, I've worked with Dad for, for many, uh, 00:02:01.362 --> 00:02:03.072 years, but I never took a classroom. 00:02:03.072 --> 00:02:07.462 I think both you and, and, uh, our sister Jen, uh, actually had him as 00:02:07.462 --> 00:02:12.021 a professor, which is, which is, uh, an interesting experience for all. 00:02:12.022 --> 00:02:19.021 Okay, so, uh, and, and your, your, um, your, your thesis project 00:02:19.452 --> 00:02:25.751 for your master's, I believe was a Bla- a Bradley Terry model 00:02:26.082 --> 00:02:28.612 predicting baseball games, even 00:02:28.735 --> 00:02:29.425 Tim Berry: Yes 00:02:31.481 --> 00:02:31.951 It was. 00:02:32.271 --> 00:02:37.361 It was, it was, uh, one of my passions, still is my passion, is, um, sports 00:02:37.361 --> 00:02:41.421 and using the numbers and data and analytics in, in the sports world. 00:02:41.921 --> 00:02:46.951 And, uh, it was the, the model was, uh, predicting whether one team 00:02:46.951 --> 00:02:49.921 would beat another in, uh, baseball. 00:02:50.421 --> 00:02:52.831 And, um, at the time, there wasn't a lot of data. 00:02:53.441 --> 00:02:54.351 There was, uh… 00:02:54.361 --> 00:02:57.271 We-- I remember we had to ship out and we had to go to a live sports 00:02:57.271 --> 00:02:59.601 bur- bureau to get some of that data. 00:02:59.601 --> 00:03:03.201 Now that data is readily available across the, uh, the web. 00:03:03.571 --> 00:03:08.721 Um, but anyway, used some unique predictors and, uh, uh, built a 00:03:08.721 --> 00:03:13.301 Bradley-Terry model to predict the odds that one team would beat another team. 00:03:13.751 --> 00:03:20.421 And then I actually applied it in real life and to see how good I was and, uh, 00:03:20.451 --> 00:03:25.251 that, and that's a theme throughout my career and my passion in using statistics 00:03:25.251 --> 00:03:27.331 is to see if I'm right, see the answer. 00:03:27.861 --> 00:03:34.471 Um, and, uh, did a lot of kinda back testing, actually pretended 00:03:34.471 --> 00:03:39.281 to bet on games and, uh, uh, um, uh, well, good news and bad news. 00:03:39.531 --> 00:03:46.361 The good news is the model was, uh, better than the bookies at predicting 00:03:46.371 --> 00:03:48.101 who would win a baseball game. 00:03:49.011 --> 00:03:53.801 The bad news is there's something in sports and betting called the vig or 00:03:53.801 --> 00:03:58.571 the, uh, you know, the, the gap between, uh, the, the, the, uh, you know, just 00:03:58.571 --> 00:04:04.291 because I was better than the bookie, they put a, a 7%, 5% cushion on top 00:04:04.291 --> 00:04:06.531 of that, and I think I was 2% better. 00:04:06.871 --> 00:04:10.011 So I had to scrap that model and realize I wouldn't be able to 00:04:10.181 --> 00:04:12.366 make money in that, uh, um, thing. 00:04:12.366 --> 00:04:13.141 But it was good. 00:04:13.521 --> 00:04:14.131 I enjoyed that one 00:04:14.920 --> 00:04:20.680 Scott: I, a- and, and, uh, v- very much a, uh, precursor to, to your career. 00:04:20.990 --> 00:04:24.870 So, a- and that was about n- uh, 1990, I, I believe, 00:04:25.119 --> 00:04:25.529 Tim Berry: Mm-hmm. 00:04:25.790 --> 00:04:27.020 Scott: where, uh… 00:04:27.020 --> 00:04:30.609 And, and by the way, Tim and I took classes together at the University of 00:04:30.609 --> 00:04:36.679 Minnesota, and I went off to graduate school, and you got a job at that point. 00:04:36.875 --> 00:04:37.475 Tim Berry: Got it, John 00:04:37.710 --> 00:04:41.559 Scott: master's in statistics from Minnesota, what's the first job? 00:04:42.969 --> 00:04:47.409 Tim Berry: So the first job is w-one is I was, uh, I was debating and, uh, 00:04:47.529 --> 00:04:52.839 with, with, with Scott and my dad, um, whether I would stay within academia 00:04:53.509 --> 00:04:55.172 and, uh, whether I was, you know… 00:04:55.172 --> 00:04:59.759 And, and I barely skated by with my master's degree, got by my 00:04:59.779 --> 00:05:05.289 thesis and, and I was very anxious to get out in the business world, 00:05:05.789 --> 00:05:08.809 to get out of academia and start. 00:05:08.879 --> 00:05:10.829 Uh, I'd been learning my whole life. 00:05:11.109 --> 00:05:15.249 I was, I was tired of learning, and I decided it's time to get out there 00:05:15.249 --> 00:05:17.339 and, and apply some of these wares. 00:05:17.339 --> 00:05:19.109 So took me a while to, to… 00:05:19.159 --> 00:05:20.649 I had a lot of opportunities. 00:05:20.659 --> 00:05:25.459 There was, um, entry-level opportunities within the pharmaceutical, uh, world. 00:05:25.459 --> 00:05:28.379 I almost took a job, um, there. 00:05:28.599 --> 00:05:34.009 I had three or four opportunities and, uh, actually coaching from, uh, my dad 00:05:34.009 --> 00:05:40.509 was, "Look it, as a master's degree person, uh, engaging in pharmaceutical 00:05:40.509 --> 00:05:44.039 companies, you might not get the opportunity than if you had a PhD." 00:05:44.629 --> 00:05:50.199 And so that was kind of a, uh, um, kind of helped me to make a decision. 00:05:50.499 --> 00:05:55.909 And then, and I, and I ended up taking a job at AT&T Bell Laboratories, and this 00:05:55.909 --> 00:06:01.789 was back in the day that, um, AT&T was a monopoly in the long-distance world. 00:06:01.799 --> 00:06:03.609 Cellular was just coming on. 00:06:04.399 --> 00:06:09.879 And, uh, the job there was to how do I help use the data that we had within 00:06:10.019 --> 00:06:13.369 AT&T to drive business decisions? 00:06:14.443 --> 00:06:18.363 Better business decisions and, and more informed business decisions. 00:06:18.363 --> 00:06:19.433 And it was… 00:06:19.433 --> 00:06:22.683 I remember the first-- one of the first projects I was working on 00:06:22.693 --> 00:06:31.323 was the likelihood that, um, a AT&T consumer would, would, uh, disconnect 00:06:31.323 --> 00:06:34.073 from AT&T and go to MCI, right? 00:06:34.073 --> 00:06:38.763 And, and back in those days, there was, um, a battle for long-distance 00:06:38.773 --> 00:06:43.693 subscribers, uh, to the point where they were offering people money to come over. 00:06:43.733 --> 00:06:48.053 And, um, so AT&T had a big retention, um, effort. 00:06:48.203 --> 00:06:52.783 And so we built a big logistic regression model that predicted the 00:06:52.833 --> 00:06:58.943 probability that someone would, um, um, someone being the AT&T consumer, 00:06:58.993 --> 00:07:01.003 would disconnect and go to MCI. 00:07:01.483 --> 00:07:05.463 And, uh, actually, uh, I remember we built it on big mainframes. 00:07:05.743 --> 00:07:09.103 I remember the first time I submitted the model, 'cause it ran 00:07:09.103 --> 00:07:11.133 on millions and millions of records. 00:07:11.413 --> 00:07:13.173 It took two and a half weeks to run. 00:07:13.923 --> 00:07:20.433 Now that same model takes less than the le- sub-second timing to run. 00:07:20.733 --> 00:07:21.183 Um, 00:07:21.188 --> 00:07:21.298 Scott: Yeah, yeah. 00:07:21.523 --> 00:07:24.693 Tim Berry: and, uh, and interesting enough, by the way, that then when, 00:07:24.743 --> 00:07:28.623 when we, we ran the model, we did multiple iter-iterations of it. 00:07:28.973 --> 00:07:30.733 We decided it was pretty good. 00:07:31.213 --> 00:07:34.883 Um, it was better than the, the, the treatment that they were applying. 00:07:35.743 --> 00:07:39.833 And, um, we, we scored everybody in the US population. 00:07:40.193 --> 00:07:43.563 We put them into deciles, and then the question was: what 00:07:43.563 --> 00:07:45.013 do you do with those deciles? 00:07:45.623 --> 00:07:49.143 Meaning if you knew, here's ten percent of the people that are highly 00:07:49.143 --> 00:07:54.803 likelihood of disconnecting and go to MCI, um, where do you target and 00:07:54.803 --> 00:07:59.693 how do you, how do you, um, how do you keep them from, uh, uh, leaving? 00:07:59.773 --> 00:08:02.143 So there was multiple iterations there. 00:08:02.673 --> 00:08:03.103 Um, 00:08:03.157 --> 00:08:03.307 Scott: But 00:08:03.403 --> 00:08:03.773 Tim Berry: the… 00:08:03.907 --> 00:08:04.647 Scott: experiment? 00:08:04.657 --> 00:08:05.777 So are you, are you 00:08:05.863 --> 00:08:06.293 Tim Berry: Yeah 00:08:07.097 --> 00:08:10.237 Scott: that you give an incentive to somebody in the 10th percentile 00:08:10.237 --> 00:08:11.888 and you, you don't somebody else? 00:08:11.888 --> 00:08:15.397 And like your baseball example, you get to figure out are you right? 00:08:15.799 --> 00:08:16.189 Tim Berry: Yeah. 00:08:16.239 --> 00:08:17.609 Yeah, you get to figure out where the right… 00:08:17.609 --> 00:08:20.079 One, one of the things we missed there though, interesting enough, 00:08:20.079 --> 00:08:25.659 is we took the top decile, the mo- people most likely to leave, and 00:08:25.659 --> 00:08:27.859 we did a full experimental design. 00:08:28.189 --> 00:08:33.249 We did multiple treatments, uh, and then we watched to see the 00:08:33.269 --> 00:08:38.379 impact those treatments had on whether someone would leave or not. 00:08:39.159 --> 00:08:39.819 One of the things… 00:08:39.859 --> 00:08:44.679 And, and, you know, there was, I, I don't remember, um, uh, um, off the top of my 00:08:44.679 --> 00:08:49.879 head, but there was probably five or six different treatments and, um, the, the… 00:08:49.889 --> 00:08:54.129 There were-- there was only one that worked, and that was a, 00:08:54.569 --> 00:08:57.199 that was a, uh, a win-back check. 00:08:57.229 --> 00:08:59.039 You had to offer the person money. 00:08:59.379 --> 00:08:59.959 "I know… 00:09:00.099 --> 00:09:02.729 Thank you for being a valued AT&T customer. 00:09:03.159 --> 00:09:05.789 Here's $100 for staying." 00:09:06.479 --> 00:09:06.759 Right? 00:09:06.819 --> 00:09:09.839 And that, that then moved the needle to keep them from leaving. 00:09:10.159 --> 00:09:12.979 But the one thing we did miss is we didn't do enough testing 00:09:12.979 --> 00:09:14.489 into deciles two and three. 00:09:15.239 --> 00:09:21.219 And what we found later on is that it's easier to save someone and the 00:09:21.219 --> 00:09:27.089 incremental impact that you can have on saving someone and staying at AT&T 00:09:27.089 --> 00:09:33.229 was better in decile three than decile one, 'cause decile one, at the time, we 00:09:33.239 --> 00:09:35.709 would call them promo surfers, and you… 00:09:35.809 --> 00:09:37.909 And, and they were gonna leave no matter what you did. 00:09:38.539 --> 00:09:39.899 Um, so anyway, that was, uh, that was a… 00:09:39.899 --> 00:09:40.029 We, we missed that initially, and then we came back and got it. 00:09:40.029 --> 00:09:40.349 So it was 00:09:44.370 --> 00:09:44.900 Scott: So, so 00:09:44.939 --> 00:09:45.239 Tim Berry: interesting 00:09:45.750 --> 00:09:46.170 Scott: Fascinating. 00:09:46.180 --> 00:09:51.710 So you're, so you're using big data to, uh, uh, be a statistician. 00:09:51.950 --> 00:09:55.260 Uh, you wouldn't have been surprised, I assume, to be called a statistician 00:09:55.509 --> 00:09:55.929 Tim Berry: No 00:09:56.619 --> 00:09:57.580 Scott: uh, in that. 00:09:57.640 --> 00:10:04.760 So you're at, you're at, uh, for about five years, and then where do you go? 00:10:05.785 --> 00:10:06.065 Tim Berry: Yeah. 00:10:06.065 --> 00:10:13.735 So the, um, uh, one is I had a, w- I had a killer time at AT&T and I learned a ton. 00:10:14.055 --> 00:10:17.475 So I was, you know, I talked about I'd learned my whole life I wanted to get 00:10:17.475 --> 00:10:22.625 out there and in the business world and, and, and ironically, uh, that's where I 00:10:22.625 --> 00:10:26.355 learned, uh, the most I've learned, and it was around business and applying things. 00:10:26.355 --> 00:10:29.165 And so, but at AT&T at the time, you really had to go into 00:10:29.165 --> 00:10:31.725 sales to move up the ladder. 00:10:32.075 --> 00:10:37.175 And one of my ambitions was to continue to progress my career and, 00:10:37.175 --> 00:10:41.985 uh, apply some of these analytical skills that I had to other areas. 00:10:42.295 --> 00:10:46.765 And so I, I looked within AT&T but just couldn't find the right opportunity. 00:10:47.045 --> 00:10:51.035 So I ended up taking a job at a marketing agency, a company called, 00:10:51.325 --> 00:10:53.055 uh, Rapp Collins, who's still around. 00:10:53.445 --> 00:10:58.095 They're part of, uh, Omnicom, which is a large marketing advertising 00:10:58.095 --> 00:11:03.335 agency, and they wanted me to come, um, to, to join that group 00:11:03.335 --> 00:11:04.905 and build their analytics group. 00:11:06.145 --> 00:11:12.435 And, uh, um, so I joined, uh, Rapp, went to, uh, Dallas and was really, 00:11:12.645 --> 00:11:16.165 I was really the second person in that group, and I was the, the, um… 00:11:16.425 --> 00:11:19.985 at, at that time, either the senior director or the vice 00:11:19.985 --> 00:11:21.655 president of analytical services. 00:11:21.655 --> 00:11:22.605 I can't remember which. 00:11:23.485 --> 00:11:23.815 Um 00:11:24.790 --> 00:11:27.580 Scott: But, but still, you, you're still a statistician at that 00:11:27.585 --> 00:11:28.295 Tim Berry: a statistician. 00:11:28.300 --> 00:11:28.630 Scott: Okay. 00:11:28.700 --> 00:11:28.940 Yep, yep 00:11:28.975 --> 00:11:32.855 Tim Berry: And, and I, I, I, I, uh, I didn't know if you wanted to, uh, 00:11:32.895 --> 00:11:34.235 talk about my time there 'cause… 00:11:34.585 --> 00:11:36.805 But anyway, my time there was pretty short. 00:11:37.475 --> 00:11:38.935 It was a y- a year and a half. 00:11:39.585 --> 00:11:41.725 And, um, what… 00:11:41.985 --> 00:11:43.255 And I didn't love the job. 00:11:43.765 --> 00:11:48.855 And I didn't love the job because at that time, I, I was treated 00:11:48.865 --> 00:11:55.145 more like window dressing, which was, "Hey, analytics is…" 00:11:55.525 --> 00:11:56.735 Uh, it, it's a lot… 00:11:57.942 --> 00:11:58.112 Scott: So, 00:11:58.295 --> 00:11:58.925 Tim Berry: I, I was… 00:11:59.082 --> 00:11:59.572 Scott: So you're, 00:11:59.635 --> 00:12:00.005 Tim Berry: yeah 00:12:00.562 --> 00:12:06.012 Scott: AT&T, you're, you're helping AT&T by doing data analytics for internal 00:12:06.105 --> 00:12:06.445 Tim Berry: Yeah. 00:12:06.745 --> 00:12:09.025 Having a real impact, real impact it feels like 00:12:09.682 --> 00:12:12.002 Scott: Okay, so they, they, they're living, they're, they're running 00:12:12.002 --> 00:12:13.141 your models, they're trusting. 00:12:13.312 --> 00:12:14.382 Now you go to Rapp Collins. 00:12:14.401 --> 00:12:17.951 Are you selling this data analytics to clients, or are 00:12:18.041 --> 00:12:18.521 Tim Berry: Yes 00:12:18.641 --> 00:12:21.061 Scott: it internal for Rapp Collins stuff? 00:12:21.249 --> 00:12:22.249 Tim Berry: Oh, no, no, sorry. 00:12:22.481 --> 00:12:23.922 Scott: this, you're selling this to 00:12:24.049 --> 00:12:25.379 Tim Berry: They have clients. 00:12:25.599 --> 00:12:30.549 Those clients are-- those clients, what, what RAP, what RAP would do is 00:12:30.559 --> 00:12:33.669 they would, uh, run marketing programs. 00:12:33.709 --> 00:12:38.089 A lot of how they made their money was buying media, right? 00:12:38.139 --> 00:12:44.489 And so they would go work for clients like Hyatt, um, Hotels and, and they 00:12:44.489 --> 00:12:50.399 would help, um, do better marketing across Hyatt's, uh, landscape, and 00:12:50.399 --> 00:12:52.029 they would be, uh, paid to do that. 00:12:52.649 --> 00:12:57.129 And, um, so I don't know if you wanted to add something to that, but I know you and 00:12:57.129 --> 00:13:00.339 I did some work at, uh, Hyatt at, at RAP. 00:13:00.649 --> 00:13:04.529 And, uh, but, uh, but I should probably, uh, finish the point around why I 00:13:04.529 --> 00:13:08.669 didn't spend a lo- uh, last long there and, and the whole window dressing is 00:13:08.669 --> 00:13:12.449 just because analytics was not really core to what they were trying to do. 00:13:13.059 --> 00:13:15.129 They weren't really committed to it. 00:13:15.599 --> 00:13:19.129 Um, a client would say, "Hey, do you guys do analytical services?" 00:13:19.749 --> 00:13:21.919 And they'd trot me out there and they'd… 00:13:22.179 --> 00:13:25.039 You know, I'd, I'd, I'd give them a little song and dance, and then 00:13:25.039 --> 00:13:26.299 I'd get put back into my room. 00:13:26.889 --> 00:13:30.109 Now, that's a little bit of an exaggeration to some extent, because 00:13:30.109 --> 00:13:31.489 I did do some real work there. 00:13:31.859 --> 00:13:36.409 But I knew that it wasn't really, um, ultimately what I was looking for there. 00:13:37.129 --> 00:13:37.399 Um 00:13:37.714 --> 00:13:40.044 Scott: I remember we did a really cool change point 00:13:40.345 --> 00:13:40.795 Tim Berry: Yeah. 00:13:40.874 --> 00:13:44.894 Scott: predict if somebody hadn't stayed within a certain time, are they, are, are 00:13:44.894 --> 00:13:46.524 they going to a different hotel chain? 00:13:47.289 --> 00:13:47.889 Tim Berry: That's right. 00:13:47.904 --> 00:13:48.564 Scott: yeah, yeah 00:13:48.779 --> 00:13:52.639 Tim Berry: another version of a retention model with a different frequent stair. 00:13:52.889 --> 00:13:53.609 Frequent stair 00:13:53.806 --> 00:13:54.356 Scott: Yep, yep. 00:13:54.516 --> 00:13:54.926 Okay. 00:13:55.246 --> 00:13:58.836 So you're not having the impact you'd like to have, so you're there a couple years. 00:13:59.046 --> 00:14:02.006 So there, then you leave and you go somewhere else. 00:14:02.055 --> 00:14:03.166 And yep 00:14:04.175 --> 00:14:04.195 Tim Berry: Yeah. 00:14:04.195 --> 00:14:07.845 I, I, I leave and I go back, I, I go to a place in Maryland. 00:14:08.505 --> 00:14:11.795 Um, at the time it was called Merkle Computer Systems. 00:14:12.715 --> 00:14:19.485 It was a, oh boy, 80, 90 people person company, so really pretty 00:14:19.485 --> 00:14:21.425 small, about 10 million in revenue. 00:14:22.155 --> 00:14:27.955 And they, they were great at, at, um, at the time was called merge-purge, 00:14:27.955 --> 00:14:33.395 which merge-purge today is called data engineering and bringing data together. 00:14:33.645 --> 00:14:36.085 But they were experts at combining data. 00:14:36.665 --> 00:14:40.655 Uh, merge-purge being that they were, they were really good at taking name 00:14:40.655 --> 00:14:43.585 and addresses, combining them together. 00:14:43.585 --> 00:14:50.385 So if you have Scott Barry living at 123 Main Street, and Scott David Barry living 00:14:50.405 --> 00:14:55.445 at 12 Main Street, realizing that that's the same person, bringing that together 00:14:55.445 --> 00:14:59.735 so that when you market to that person, you're not sending two messages out. 00:14:59.985 --> 00:15:04.695 Sometimes you even see that today where you get emails, the same emails come 00:15:04.695 --> 00:15:08.705 to your address 'cause they haven't done proper deduping, so to speak. 00:15:08.785 --> 00:15:11.395 So that's what w- that's what Merkle was. 00:15:12.115 --> 00:15:17.875 And, uh, they, they, uh, the CEO of that company really saw an opportunity 00:15:18.275 --> 00:15:24.555 in bringing analytics to that data environment for their clients and, uh, 00:15:24.565 --> 00:15:30.125 were looking for someone to lead and build their analytical services, um, capability. 00:15:30.905 --> 00:15:31.365 And so, 00:15:31.796 --> 00:15:35.276 Scott: So, so, so still you're, you're now at a service company that is 00:15:35.276 --> 00:15:38.056 selling data analytics to clients, 00:15:38.457 --> 00:15:38.687 Tim Berry: Yep 00:15:38.806 --> 00:15:41.706 Scott: you're leading that, that group, uh, at, at 00:15:41.717 --> 00:15:42.937 Tim Berry: A pretty brand new. 00:15:43.177 --> 00:15:44.837 Our clients weren't expecting that. 00:15:44.867 --> 00:15:47.727 Uh, the organization didn't know much about analytics. 00:15:48.237 --> 00:15:53.287 Um, and so it was a, it was an exciting opportunity, but it also meant, um, 00:15:53.547 --> 00:15:55.217 you know, I did a lot of education. 00:15:55.447 --> 00:15:56.347 There's a funny story. 00:15:56.347 --> 00:15:56.667 I was… 00:15:57.197 --> 00:16:02.227 I get to, I get to Merkle and, um, uh, I, I know you guys are gonna be 00:16:02.227 --> 00:16:04.417 surprised at, at this, uh, story. 00:16:04.597 --> 00:16:07.937 But I get to Merkle and it's about two months in and I get the 00:16:07.937 --> 00:16:12.127 opportunity to address the company on what analytical services is. 00:16:13.127 --> 00:16:14.507 And so there's a… 00:16:14.547 --> 00:16:15.647 And again, it's not a big company. 00:16:15.647 --> 00:16:16.907 There's about 100 people there. 00:16:17.347 --> 00:16:19.967 And, uh, I knew one of the ladies in the, uh… 00:16:20.067 --> 00:16:21.147 She worked with me. 00:16:21.427 --> 00:16:24.357 She was one of the salespeople, and she had a, she had a, uh, 00:16:24.377 --> 00:16:25.667 sister that also worked there. 00:16:26.437 --> 00:16:30.187 And while I'm up there talking, uh, the sister's kinda… 00:16:30.737 --> 00:16:34.377 She's, she didn't know much about, um, um, analytics. 00:16:34.377 --> 00:16:36.507 And so finally she says, "I don't get it. 00:16:36.527 --> 00:16:37.337 Who is that?" 00:16:37.947 --> 00:16:40.947 And, and the lady that I knew, she says, "Oh, that's Tim. 00:16:41.247 --> 00:16:43.187 He's our, he's our modeler." 00:16:43.877 --> 00:16:48.297 She thinks a minute, looks back and forth, says, "Boy, he's not that good-looking." 00:16:50.117 --> 00:16:51.907 And so, uh, that, that… 00:16:52.287 --> 00:16:56.507 I, I spent a ton of time edgiting- educating people on 00:16:56.717 --> 00:16:58.677 what analytical services were. 00:16:58.897 --> 00:17:03.437 They had a lot of nonprofit organizations, and we started to 00:17:03.437 --> 00:17:08.157 build predictive models for those nonprofits, which is who's likely to be 00:17:08.157 --> 00:17:11.347 a donor and how much they might give. 00:17:11.677 --> 00:17:13.507 And there was tons of pushback. 00:17:13.597 --> 00:17:18.817 This is early '90s, mid '90s, on you can't model emotions. 00:17:19.647 --> 00:17:24.257 So even though you think you can predict who's gonna make a decision, we as 00:17:24.257 --> 00:17:29.607 marketers, um, um, um, use emotions to… 00:17:29.697 --> 00:17:32.837 And, and they still do it in the fundraising world, to 00:17:32.837 --> 00:17:34.167 use emotions to get donors. 00:17:34.337 --> 00:17:35.617 How can you predict that? 00:17:35.687 --> 00:17:37.727 How can you model that? 00:17:37.727 --> 00:17:38.017 So, um- 00:17:38.048 --> 00:17:41.308 Scott: it almost sounds a little bit like Moneyball, where it's the fight 00:17:41.637 --> 00:17:41.907 Tim Berry: Sure 00:17:41.928 --> 00:17:44.058 Scott: the analytics people and the scout who 00:17:44.121 --> 00:17:44.741 Tim Berry: In the gut, 00:17:44.888 --> 00:17:45.008 Scott: at 00:17:45.011 --> 00:17:45.511 Tim Berry: for sure 00:17:45.957 --> 00:17:48.528 Scott: you know, uh, his muscle tone and tell you whether he's 00:17:48.528 --> 00:17:50.848 gonna be a good baseball player as opposed to the numbers." 00:17:51.728 --> 00:17:56.077 you're, you- a company, and a huge part at, at this time was getting data. 00:17:56.367 --> 00:18:01.747 It, it was who had data, selling data, and so you're, the company has this 00:18:01.748 --> 00:18:05.647 amazing data, and it curates it, and it, it, you know, so it has good data. 00:18:05.888 --> 00:18:09.388 But now you're selling the service of using the data to make 00:18:09.398 --> 00:18:11.448 better decisions to companies. 00:18:11.718 --> 00:18:13.148 Now does this catch on? 00:18:13.528 --> 00:18:16.558 Does, does, does data analytics catch on for Merkle? 00:18:17.273 --> 00:18:17.713 Tim Berry: Yes. 00:18:17.993 --> 00:18:19.053 Yes, it does. 00:18:19.093 --> 00:18:20.023 It, it, it did. 00:18:20.093 --> 00:18:21.473 It, it took a while. 00:18:21.883 --> 00:18:22.483 It took… 00:18:22.683 --> 00:18:26.663 You know, I, I, I talk about building a business, and we were really 00:18:26.663 --> 00:18:29.303 building a business within, uh, Merkle. 00:18:29.743 --> 00:18:32.873 Um, and, uh, it, it's like pushing a flywheel. 00:18:33.413 --> 00:18:37.263 You start to push it, and it's heavy, and it's not going anywhere. 00:18:37.263 --> 00:18:39.173 And you push it, still it's not going anywhere. 00:18:39.173 --> 00:18:40.283 And you push it. 00:18:40.283 --> 00:18:43.983 It finally starts to turn a little bit and starts to turn a little bit. 00:18:44.333 --> 00:18:48.083 And then once, after or after you continually push it, then it's, 00:18:48.153 --> 00:18:49.633 uh, it starts going on its own. 00:18:50.163 --> 00:18:56.413 And it-- I remember when we hired, it took me about a year to hire two people, and 00:18:56.413 --> 00:19:03.473 then after three years, we had almost, uh, 80 people in that group, right? 00:19:03.503 --> 00:19:04.213 Just because it… 00:19:04.513 --> 00:19:07.973 They had data and analytics and statisticians and hiring people with 00:19:08.203 --> 00:19:12.573 master's degree, some with PhDs, um, some with bachelor's degrees. 00:19:12.593 --> 00:19:16.613 But, but about 80% of who we were looking for were people with master's 00:19:16.613 --> 00:19:19.713 degrees and, uh, PhDs degree, PhDs. 00:19:20.363 --> 00:19:21.223 And then that continued. 00:19:21.453 --> 00:19:22.713 That was just three years into it. 00:19:23.233 --> 00:19:28.223 Um, and, um, so that continued to grow and scale within Merkle and 00:19:28.223 --> 00:19:33.013 became a huge part of its value proposition, um, in the marketing arena. 00:19:33.693 --> 00:19:35.313 Um, so 00:19:36.414 --> 00:19:38.344 Scott: Okay, so you're still a statistician, but now 00:19:38.344 --> 00:19:39.404 you're leading this group. 00:19:39.464 --> 00:19:39.714 You, 00:19:39.883 --> 00:19:40.373 Tim Berry: leading the group 00:19:40.594 --> 00:19:42.044 Scott: founded, you're leading the group. 00:19:42.414 --> 00:19:49.184 Um, and then, um, this grows to a size that your position changes, uh, 00:19:49.455 --> 00:19:49.625 Tim Berry: Yeah 00:19:49.754 --> 00:19:53.043 Scott: from statistician to president of MERCKLE. 00:19:53.534 --> 00:19:54.524 Co-president? 00:19:54.583 --> 00:19:54.883 Yep. 00:19:55.353 --> 00:19:55.593 Yep. 00:19:55.641 --> 00:19:57.461 Tim Berry: There was a couple steps in between that, but 00:19:57.461 --> 00:19:58.471 yes, that's what happened. 00:19:58.801 --> 00:20:00.281 And, um, important… 00:20:00.281 --> 00:20:02.041 It would be important to, to note. 00:20:02.041 --> 00:20:05.631 So we-- I was out, you know, I spent a lot of time doing, 00:20:05.861 --> 00:20:08.341 uh, customer pitches, right? 00:20:08.381 --> 00:20:11.501 And, uh, I think that's one of the, you know, the, the, uh… 00:20:11.532 --> 00:20:12.592 Scott: data analytics. 00:20:12.602 --> 00:20:13.232 So, uh, 00:20:13.391 --> 00:20:17.611 Tim Berry: Selling data analytics to our clients, trying to convince them that 00:20:17.621 --> 00:20:23.561 they should spend money on analytics to help them do better things, right? 00:20:23.731 --> 00:20:29.841 And so, um, as part of that, I, I, I, um, connected with a, um, uh, 00:20:29.891 --> 00:20:33.981 a partner, uh, Patrick Hennessy, who was a salesperson at the time. 00:20:34.331 --> 00:20:35.811 I was the analytics guy. 00:20:36.111 --> 00:20:39.411 He would-- He and I would go out and do customer pitches together. 00:20:39.891 --> 00:20:44.601 And we just realized that a lot of the skills that I brought to the table 00:20:44.911 --> 00:20:48.131 helped him, and a lot of the skills he brought to the table helped me. 00:20:48.581 --> 00:20:53.441 And so we started to combine, um, w- uh, you know, we combined our 00:20:53.441 --> 00:20:56.951 pitches, and then all of a sudden we decide, let's build a group together. 00:20:57.431 --> 00:21:00.151 So he joined me in helping build the analytics group. 00:21:00.941 --> 00:21:05.911 And then, um, uh, realizing that we were, you know, we were more effective 00:21:06.151 --> 00:21:08.511 as what we like to call two in a box. 00:21:08.991 --> 00:21:13.231 And so then we just continued to progress as Merkle grew, and Merkle 00:21:13.231 --> 00:21:17.371 ended up scaling to almost a billion dollars in revenue by the time I left. 00:21:17.371 --> 00:21:18.801 We came at, at, uh… 00:21:18.861 --> 00:21:20.331 When it was less than 10 million. 00:21:20.841 --> 00:21:24.821 And then he and I stayed working and leading groups together. 00:21:25.151 --> 00:21:26.751 And to Scott's point, we… 00:21:27.091 --> 00:21:30.981 By the time I had ended up leaving, uh, Merkle after about 20 years, I 00:21:30.981 --> 00:21:32.841 was co-president with Patrick Hennessy 00:21:33.784 --> 00:21:37.884 Scott: How big did the data analytics group come by, by the end of this? 00:21:37.974 --> 00:21:42.954 Um, uh, what fraction of the billion dollars is data analytics? 00:21:43.065 --> 00:21:43.395 Tim Berry: Yeah. 00:21:43.445 --> 00:21:46.715 It was a, it was about, there was a couple of things in there. 00:21:47.125 --> 00:21:50.645 Um, so we, as you mentioned, we had data that would help… 00:21:50.665 --> 00:21:53.825 We, we would sell some data to our clients that would help 00:21:53.825 --> 00:21:55.925 them get more predictive data. 00:21:56.465 --> 00:21:59.915 You know, like we had age and income and occupation and things like that. 00:22:00.095 --> 00:22:01.525 So that was included in that. 00:22:01.525 --> 00:22:05.695 But it was about $350 million of the billion dollars in revenue. 00:22:06.515 --> 00:22:08.185 Um, so about 35% of that. 00:22:08.995 --> 00:22:09.155 Uh 00:22:09.820 --> 00:22:10.420 Scott: Uh, amazing. 00:22:10.430 --> 00:22:12.700 Now, i- at this time, and this is… 00:22:12.730 --> 00:22:17.230 I, I think you're at Merkle till 2018-ish, uh, uh, 00:22:17.369 --> 00:22:17.679 Tim Berry: Yep. 00:22:17.720 --> 00:22:19.250 Scott: I think you go to… 00:22:19.329 --> 00:22:19.439 Tim Berry: So 00:22:19.580 --> 00:22:22.750 Scott: you spend time in London, you lead Europe Merkle and all that. 00:22:22.750 --> 00:22:27.999 But is there any AI at Merkle before 2018? 00:22:30.081 --> 00:22:30.881 Tim Berry: Wow. 00:22:31.141 --> 00:22:38.881 Well, um, in, in, in a word, not as we know AI today, but the way we define 00:22:38.931 --> 00:22:42.381 AI today, there was tons of it, right? 00:22:42.581 --> 00:22:46.721 And, and, uh, but it was, it wasn't defined as AI. 00:22:46.731 --> 00:22:47.901 So the… 00:22:47.911 --> 00:22:51.621 And again, so it boils down to the definition of how you wanna define AI. 00:22:51.891 --> 00:22:56.711 But there was really an evolution in my life cycle as a business person from when 00:22:56.711 --> 00:23:01.941 I first started to even as I sit here today, which was there was statisticians, 00:23:03.071 --> 00:23:09.011 then the-- then, then they became analysts or analytical folks, right? 00:23:09.181 --> 00:23:12.981 And, and back in my day when I was a statistician, if you called me an 00:23:12.991 --> 00:23:15.351 analyst, I'd be upset with you, right? 00:23:15.581 --> 00:23:18.351 Um, and then, but it just became a common term. 00:23:18.401 --> 00:23:22.851 So there was an evolution from being a statistician to then being in analytics 00:23:23.211 --> 00:23:25.241 to then being a data scientist. 00:23:26.171 --> 00:23:31.031 And then as we sit here today, data scientists are now AI people, right? 00:23:31.081 --> 00:23:36.181 AI engineers, AI scientists, um, because it's the way of the world. 00:23:36.611 --> 00:23:41.431 But a lot of the things that are going on that we call AI today is 00:23:41.701 --> 00:23:46.831 fundamental data engineering, data science, sta-statistical work. 00:23:47.541 --> 00:23:47.841 Uh, so 00:23:48.858 --> 00:23:49.408 Scott: Okay, interest. 00:23:49.538 --> 00:23:56.118 So, and now Merkle, Merkle gets bought and you, you stay on there, but then 00:23:56.118 --> 00:23:58.868 you leave Merkle, uh, at this point. 00:23:58.897 --> 00:24:01.208 Um, and, and what next? 00:24:01.657 --> 00:24:02.037 Tim Berry: Yep. 00:24:02.547 --> 00:24:05.297 Yeah, so, so I'll bring Patrick in again on that. 00:24:05.867 --> 00:24:12.537 And so he and I were in the boardroom of, at, at Merkle about 2010. 00:24:13.287 --> 00:24:13.687 With the… 00:24:13.717 --> 00:24:15.637 Now Merkle got sold in 2018. 00:24:16.057 --> 00:24:16.637 We had… 00:24:16.687 --> 00:24:19.397 And if you look at what Merkle as a company, there was a lot of 00:24:19.397 --> 00:24:21.187 different evolutions of Merkle. 00:24:21.737 --> 00:24:25.787 We, and, and it's an incred- incredibly important lesson that I learned in, 00:24:26.147 --> 00:24:30.317 in, and at a company called Blend, in building and growing Blend, is 00:24:30.317 --> 00:24:32.057 making sure you evolve and change. 00:24:32.617 --> 00:24:38.367 And so we were going through one of those changes, and we, we, as I mentioned, we 00:24:38.367 --> 00:24:44.417 were a data company that brought analytics in, that had technology, and we were… 00:24:44.567 --> 00:24:47.647 Our secret sauce was data analytics and technology. 00:24:48.347 --> 00:24:51.317 And most of our clients were in the marketing world. 00:24:52.247 --> 00:24:58.087 And we had a decision to make that was, "Hey, um, where do we go next? 00:24:58.257 --> 00:25:04.167 Do we wanna then stay engaged in the marketing world and evolve our 00:25:04.317 --> 00:25:09.727 services around what, what we were calling a addressable marketing agency? 00:25:09.997 --> 00:25:14.397 Use that superpower of data analytics and technology, and apply 00:25:14.397 --> 00:25:17.807 it in the agency world and apply it in marketing for our clients." 00:25:18.177 --> 00:25:23.627 We were debating whether we should do that or we should, we should use data 00:25:23.627 --> 00:25:28.577 analytics and technology and apply it outside of marketing into all kinds 00:25:28.577 --> 00:25:35.537 of other areas, from healthcare to HR, to, to manufacturing, to supply chain. 00:25:35.967 --> 00:25:40.787 And Patrick and I were voting and arguing that, "Let's do that. 00:25:41.377 --> 00:25:45.587 Let's bring that superpower to other domains of the business." 00:25:45.927 --> 00:25:50.727 And there was a he- heated debate and back over, over a couple weeks actually, and in 00:25:50.727 --> 00:25:55.267 the end, our CEO, David Williams, got to make that decision, and he decided, "Nope, 00:25:55.297 --> 00:25:57.197 let's stay within the marketing space." 00:25:58.267 --> 00:26:00.987 And that was always a point that we said, "Boy…" 00:26:01.197 --> 00:26:04.867 And it, by the way, that was a good decision for Merkle, and in the end, 00:26:04.887 --> 00:26:06.827 everybody benefited from that decision. 00:26:06.827 --> 00:26:08.717 I'm glad that decision was made. 00:26:09.347 --> 00:26:14.597 Uh, but at the time, Patrick and I were saying, "When our run at Merkle is over, 00:26:14.807 --> 00:26:16.607 we wanna create that type of company." 00:26:17.387 --> 00:26:21.087 Um, and so that's, as, as our run at, uh… 00:26:21.097 --> 00:26:26.037 When, when Merkle got sold, um, we, we both said, "This is the time. 00:26:26.057 --> 00:26:30.297 Step out and s- and start what we now, what we now call Blend." 00:26:31.836 --> 00:26:32.216 Scott: Okay. 00:26:32.256 --> 00:26:34.786 So you, you keep with the two in the box. 00:26:35.203 --> 00:26:36.363 Tim Berry: Keep with the two in a box 00:26:36.856 --> 00:26:42.086 Scott: you and Patrick go out and you acquire a group 00:26:42.841 --> 00:26:43.101 Tim Berry: Yeah 00:26:43.396 --> 00:26:47.866 Scott: to build into this much broader business outside of marketing. 00:26:47.875 --> 00:26:51.886 Maybe you'll do marketing, but largely a much broader business, uh, aspect of it. 00:26:52.215 --> 00:26:55.706 Uh, and, and Consultants To Go, I think, is a, is 00:26:55.737 --> 00:26:55.747 Tim Berry: Hmm 00:26:55.815 --> 00:27:01.466 Scott: company C2G, uh, acquired, and this becomes Blend, uh, at this point. 00:27:01.466 --> 00:27:05.655 So now you … By the way, you went from sta- statistician, analyst, 00:27:05.655 --> 00:27:08.165 data scientist, co-president. 00:27:08.216 --> 00:27:09.625 Now you're co-founder. 00:27:10.275 --> 00:27:10.815 Tim Berry: Koffar. 00:27:10.815 --> 00:27:11.476 Scott: co-founder. 00:27:11.755 --> 00:27:14.515 Uh, and, and this is Blend360. 00:27:14.775 --> 00:27:17.075 And so what is Blend360? 00:27:17.657 --> 00:27:17.967 Tim Berry: Yep. 00:27:18.607 --> 00:27:19.057 Yes. 00:27:19.067 --> 00:27:24.327 So we, um, we, we had a decision make is, is we wanted to start this data 00:27:24.327 --> 00:27:28.167 analytics technology company and apply it to different areas of the business, so we 00:27:28.167 --> 00:27:29.627 had to decide how were we gonna do that. 00:27:30.427 --> 00:27:34.317 And one way to do it is you plant the flag, say, "I'm open 00:27:34.337 --> 00:27:35.697 for business," and you find… 00:27:35.937 --> 00:27:37.997 you know, you start finding clients, right? 00:27:37.997 --> 00:27:39.697 Another way to do it is buy a company. 00:27:40.017 --> 00:27:42.307 Gives you a foundation to build off of. 00:27:42.547 --> 00:27:47.427 You got people who can do billing, you got some clients, you've got… 00:27:47.447 --> 00:27:51.787 And so, um, our thinking was, let's see if we can find a company that we could 00:27:51.797 --> 00:27:56.827 buy that could give us the platform to build the company that we now call 00:27:56.837 --> 00:27:58.407 Blend or the company that we were after. 00:27:58.717 --> 00:27:59.187 And we did. 00:27:59.187 --> 00:28:02.557 We came across this company called, uh, I, I haven't heard that name, 00:28:02.707 --> 00:28:06.427 Consultants-To-Go, in a while, because we- as soon as we, as soon as I heard 00:28:06.427 --> 00:28:10.667 that term, I said, "I'm not buying a company that's Consultants-To-Go. 00:28:10.917 --> 00:28:13.157 I get a hamburger, I get a consultant." 00:28:13.794 --> 00:28:14.054 Scott: Yeah. 00:28:14.167 --> 00:28:17.037 Tim Berry: and so we typ- we, we immediately changed it. 00:28:17.237 --> 00:28:19.287 We changed it's, uh, a strong point. 00:28:19.637 --> 00:28:24.517 We changed it to C2G, and people would say, "What does C2G stand for?" 00:28:24.927 --> 00:28:26.977 And, uh, we would say, "Committed to growth." 00:28:27.797 --> 00:28:32.147 And so, uh, but it was e- but if I go back to the Consultants-To-Go 00:28:32.147 --> 00:28:36.337 point, it was about a, it was about a $8 million company. 00:28:36.757 --> 00:28:37.317 It was… 00:28:37.357 --> 00:28:38.577 It had consultants. 00:28:38.707 --> 00:28:44.847 It w- it had a unique value proposition of it was largely 80 to 90% of the 00:28:44.867 --> 00:28:49.457 consultants were women, and they were s- it was started by two women coming 00:28:49.457 --> 00:28:55.867 out of AT&T and Amex that, um, um, the, the, a lot of the women had stepped out 00:28:55.867 --> 00:29:00.467 of the workforce, wanted to get back in, engaged in the business, wanted 00:29:00.467 --> 00:29:06.267 flexibility, wanted to work on their terms, and, and, and had great skill sets. 00:29:06.467 --> 00:29:10.157 So they got started to get a Rolodex of these people, and they would then go in 00:29:10.157 --> 00:29:13.127 and offer them to their, um, clients. 00:29:13.307 --> 00:29:17.017 And they, they had some good clients in Citibank, um, AT&T, Amex. 00:29:18.547 --> 00:29:24.277 And, um, and so that gave us a foundation to build off of and, um, um, 00:29:24.317 --> 00:29:26.147 and then we started Blend using that. 00:29:26.657 --> 00:29:30.587 And, um, we, we were looking to buy some analytic strength 00:29:31.176 --> 00:29:31.485 Scott: Hmm 00:29:31.681 --> 00:29:34.961 Tim Berry: Um, and we just couldn't find anything that we were excited for. 00:29:34.961 --> 00:29:39.341 One of our key value propositions still today is we wanna just be 00:29:39.341 --> 00:29:43.661 known for exceptional people, and a lot of time we just didn't find the 00:29:43.661 --> 00:29:47.681 quality that we wanted, so we just decided to build that on our own. 00:29:48.351 --> 00:29:53.751 And, um, and, and so we started the Blend endeavor in 20-- late 2017, early 2018, 00:29:55.921 --> 00:29:58.521 and, uh, we scaled, um, um… 00:29:58.741 --> 00:30:02.401 Really it's, it's a broad value proposition, but we would just go to 00:30:02.401 --> 00:30:06.171 our clients and say, "You have data here that can help you be smarter. 00:30:06.661 --> 00:30:11.861 Let us come in there, let us understand what your business problems are, and 00:30:11.861 --> 00:30:15.591 then let us apply solutions there that are gonna impact your results." 00:30:16.191 --> 00:30:16.951 And you know what? 00:30:17.191 --> 00:30:21.021 Sometimes they would say, "I don't know," and say, "Just give us a shot. 00:30:21.141 --> 00:30:22.101 Don't even pay us. 00:30:22.571 --> 00:30:26.951 If we don't-- If, if we're not able to impact your results, you should fire us." 00:30:27.511 --> 00:30:31.821 And so we would get in there with a pretty aggressive, um, approach 00:30:31.821 --> 00:30:35.281 there and then, uh, just be able to make sure that we could demonstrate 00:30:35.281 --> 00:30:36.501 our ability to drive results. 00:30:36.951 --> 00:30:40.141 And that, that, that mentality exists today, um, 00:30:40.886 --> 00:30:44.596 Scott: I, I mean, it's fascinating from the time you're at Rapp Collins 00:30:44.606 --> 00:30:47.126 where, uh, it's an afterthought. 00:30:47.516 --> 00:30:51.156 You're, you're a, you're a, a, "Oh, yes, we have data analytics. 00:30:51.156 --> 00:30:52.136 Somebody go find them." 00:30:52.416 --> 00:30:57.066 Um, to, you know, data analytics at the time, now it's, it's, uh, 00:30:57.066 --> 00:30:58.716 you know, almost like AI is now. 00:30:58.716 --> 00:31:01.356 E- everybody's talking about data analytics, how do we use this data? 00:31:01.566 --> 00:31:05.526 So an easier sell, but you're still trying to convince companies 00:31:05.526 --> 00:31:07.816 that the data they have is gold 00:31:08.281 --> 00:31:08.711 Tim Berry: Yeah 00:31:08.726 --> 00:31:10.865 Scott: and, and it can make you better and smarter. 00:31:10.866 --> 00:31:15.566 So, so now you're largely building data analytics at Blend. 00:31:16.566 --> 00:31:21.295 and I know you went out to various, uh, schools and brought in whole classes 00:31:21.296 --> 00:31:24.445 of, of master's students, for example. 00:31:24.445 --> 00:31:28.125 And, uh, so you're getting-- You're, you're, you're, you're hiring large 00:31:28.125 --> 00:31:33.535 numbers of statistician data analysts and, and this is growing rapidly at Blend 00:31:33.535 --> 00:31:35.085 now, this, this part of the business. 00:31:35.673 --> 00:31:35.923 Tim Berry: Yep. 00:31:36.403 --> 00:31:36.673 Yep. 00:31:36.863 --> 00:31:40.223 And yeah, 'cause when you think of Blend, you know, if we're… 00:31:40.253 --> 00:31:43.203 Right now we're about $200 million in revenue. 00:31:43.203 --> 00:31:45.133 We have about 1,500 employees. 00:31:45.673 --> 00:31:48.203 And, um, um, the, uh, c- the… 00:31:48.263 --> 00:31:52.093 If you think about Blend, we really only have two big areas. 00:31:52.323 --> 00:31:54.163 We have our people, and we have clients. 00:31:55.033 --> 00:31:57.963 And if we don't have people, our clients have nothing to buy. 00:31:58.213 --> 00:32:01.373 So we're very much a services-based company, right? 00:32:01.443 --> 00:32:07.443 And, um, and so we, we, we, we said very quickly, "We gotta make sure that 00:32:07.443 --> 00:32:09.513 we hire people that people wanna buy. 00:32:09.983 --> 00:32:13.403 We gotta, we gotta be, um, producing talent." 00:32:14.353 --> 00:32:17.853 And, uh, a b- you know, we would go out and find very experienced 00:32:17.893 --> 00:32:19.613 people that knew how to do that. 00:32:20.143 --> 00:32:24.583 Um, but we also said to ourselves, "We gotta start bringing in great talent, 00:32:25.193 --> 00:32:26.743 training them up to some extent." 00:32:26.763 --> 00:32:32.983 When I say training them up, it's a huge part of what makes a statistician, I'll 00:32:32.983 --> 00:32:38.843 go back to that term, um, data engineer, analyst successful in our world, is 00:32:38.913 --> 00:32:41.213 understanding the business side of things. 00:32:41.713 --> 00:32:46.313 And so we would hire people right out of multiple schools. 00:32:46.313 --> 00:32:47.933 We'd put them in cohorts. 00:32:48.103 --> 00:32:49.223 Cohorts from six… 00:32:49.243 --> 00:32:49.933 We still do this. 00:32:49.933 --> 00:32:51.643 So I said we'd, we, we would. 00:32:52.213 --> 00:32:58.583 Um, we put them in cohorts of six to 10 people, and then for three months we, 00:32:58.583 --> 00:33:01.373 um, expose them to the business problems. 00:33:01.703 --> 00:33:06.263 They go through, they go through some training around, uh, the different tools 00:33:06.383 --> 00:33:11.173 and technologies that we use that, uh, they might not have used in, uh, in 00:33:11.183 --> 00:33:13.233 the universities or colleges, right? 00:33:13.233 --> 00:33:14.153 So there's some of that. 00:33:14.393 --> 00:33:17.573 There's some exposure to some of the algorithms we have and some of 00:33:17.573 --> 00:33:19.293 the platforms that we've created. 00:33:19.673 --> 00:33:23.473 But a, a big chunk of it is the, the business problems we're solving. 00:33:23.483 --> 00:33:28.693 Some of the things I went on earlier in the, the podcast around here's some of the 00:33:28.703 --> 00:33:34.653 use cases and w- and, and have them work through that as a team so they can get 00:33:34.723 --> 00:33:38.773 better understanding of the business side of things and how analytics can help that. 00:33:39.073 --> 00:33:40.103 So we, we have that. 00:33:40.133 --> 00:33:41.583 We call it our All-Star program. 00:33:41.993 --> 00:33:46.753 They go through that three months, then we deploy them, um, in three months, 00:33:46.863 --> 00:33:51.443 after three months into our clients, and they've got a three-month trial period. 00:33:51.443 --> 00:33:54.923 And after six months, if they m- grad- graduate, they make 00:33:54.923 --> 00:33:56.223 it as full Blend employees. 00:33:56.963 --> 00:33:59.853 That was, that's been a really great program for us. 00:33:59.853 --> 00:34:05.863 It's scaled ac- It's, it's now in LatAm, it's in India, and it's in EMEA as well 00:34:07.308 --> 00:34:10.568 Scott: And, but this is largely in still the data analytics, which still 00:34:10.841 --> 00:34:11.321 Tim Berry: Yeah 00:34:11.368 --> 00:34:14.168 Scott: what fraction of Blend today would you… 00:34:14.208 --> 00:34:18.068 I, and I know it may be hard, but i- is roughly data analytics? 00:34:18.809 --> 00:34:19.049 Tim Berry: Yeah. 00:34:19.339 --> 00:34:24.759 So we, i-i-i-it, it, it's better to say what are the capabilities of the people? 00:34:25.189 --> 00:34:32.169 And so pre-AI, pre-AI, about 80%, and it's about… 00:34:32.309 --> 00:34:37.099 It was if I use 100%, 40% of our people were data scientists. 00:34:38.789 --> 00:34:40.779 40% of our people were data engineers. 00:34:41.819 --> 00:34:42.139 All right? 00:34:42.309 --> 00:34:47.059 So, you know, in, in your guys' world, the statisticians 00:34:47.069 --> 00:34:48.499 more than the data scientists. 00:34:48.769 --> 00:34:52.679 Um, and that's made up blend, and then the other was, was, was client 00:34:52.679 --> 00:34:57.999 people or, um, marketing people or domain expertise people and so forth. 00:34:57.999 --> 00:35:01.399 But a large portion was our, the data and analytics coming together. 00:35:01.769 --> 00:35:05.829 Um, as you sit here today, that's now transformed into, you know, 00:35:05.899 --> 00:35:11.219 AI engineers and context prompt, prompt engineers, context engineers. 00:35:11.579 --> 00:35:12.879 We still have data engineers. 00:35:12.879 --> 00:35:17.099 We still have some data scientists, but that world's evolving, um, 00:35:17.139 --> 00:35:19.209 into new titles and new areas. 00:35:19.399 --> 00:35:20.399 Um, but, 00:35:20.482 --> 00:35:22.952 Scott: so, and you're at, yeah, I mean, it's this interesting thing. 00:35:22.952 --> 00:35:26.502 I asked the question about AI at Merkle because largely, well, you had data 00:35:26.502 --> 00:35:30.721 engineer- engineers, and they were doing things that you might think of that, that 00:35:30.721 --> 00:35:34.381 this whole jumping into the AI bandwagon. 00:35:34.631 --> 00:35:39.191 Now, now Blend is, is right where this explosion sort of happens. 00:35:39.432 --> 00:35:43.022 Are you having similar discussions that you had at Merkle about should 00:35:43.022 --> 00:35:44.271 we do this, should we do that? 00:35:44.272 --> 00:35:46.542 About should we be embracing AI? 00:35:46.542 --> 00:35:48.161 Should we be selling AI? 00:35:48.401 --> 00:35:51.361 Are we an agentic AI creation company? 00:35:51.361 --> 00:35:55.271 What are … I mean, is that process in Blend of that 00:35:55.839 --> 00:35:56.159 Tim Berry: Yeah. 00:35:56.529 --> 00:35:56.719 Yeah. 00:35:56.719 --> 00:36:01.799 One is the marketplace and, and, and I use that term a little bit loosely. 00:36:01.849 --> 00:36:04.239 Is it the AI, AI marketplace or not? 00:36:04.589 --> 00:36:07.729 But the, the world we live into is changing every day. 00:36:07.969 --> 00:36:09.019 It's fast-paced. 00:36:09.499 --> 00:36:14.789 And one of the lessons we learned at, uh, Merkle was to make sure that we stay-- 00:36:15.149 --> 00:36:20.269 we continue to change, and we continue to adopt to what people need, what the 00:36:20.269 --> 00:36:23.559 trends are, where this business is going. 00:36:23.819 --> 00:36:26.939 And AI is a huge word, huge word these days. 00:36:26.939 --> 00:36:30.719 And, and, uh, we have transformed ourselves into 00:36:30.749 --> 00:36:33.239 much more of an AI-native shop. 00:36:34.079 --> 00:36:38.519 But with that said, there's still fundamental roots in the data 00:36:38.519 --> 00:36:40.139 science and data engineering. 00:36:40.559 --> 00:36:45.759 And most of our work we do in the AI world is around agentic AI. 00:36:45.829 --> 00:36:52.719 It's developing agents for our clients to help them drive more productive, 00:36:52.749 --> 00:36:55.109 more efficient business results. 00:36:55.229 --> 00:36:57.409 And so I, I heard one of the- 00:36:57.825 --> 00:37:01.145 Scott: what's an example of that without confidential information 00:37:01.433 --> 00:37:01.853 Tim Berry: Yeah. 00:37:02.003 --> 00:37:02.443 Yes. 00:37:02.445 --> 00:37:02.825 Scott: build? 00:37:03.453 --> 00:37:07.963 Tim Berry: A lot of it is around, um, kinda right now it's, it's around 00:37:07.993 --> 00:37:10.313 back office workflows type stuff. 00:37:10.673 --> 00:37:16.643 So for example, we do work with a, a, a large financial advising firm, and 00:37:16.643 --> 00:37:21.333 they advise their clients on where the, the investments they should make, 00:37:21.333 --> 00:37:23.213 how do they deal with their portfolio. 00:37:23.763 --> 00:37:29.593 And we're building agentic AI agents to replace what some of the back 00:37:29.593 --> 00:37:34.623 office, um, researchers would do around the data associated with this 00:37:34.653 --> 00:37:40.383 to make better recommendations to the financial advisor themselves that then 00:37:40.383 --> 00:37:41.833 they can go to their clients with. 00:37:42.133 --> 00:37:47.043 So just replacing the whole process around, um, doing that. 00:37:47.043 --> 00:37:52.283 It's more efficient, it's more effective, um, and it, and it requires 00:37:52.793 --> 00:37:55.833 less, although there's still some of it, less change management. 00:37:56.713 --> 00:38:00.063 Because one of the things you'll hear in the industry, and I'm sure 00:38:00.063 --> 00:38:04.173 a lot of you guys have heard it, is how agents are gonna replace people. 00:38:04.743 --> 00:38:09.873 And, and we're seeing some of that, um, but not nearly the 00:38:09.883 --> 00:38:12.003 extent that the noise is going on. 00:38:12.493 --> 00:38:17.573 And, um, and what's, what's having-- There's, there's-- I could, I could 00:38:17.573 --> 00:38:20.893 probably have multiple podcasts on this and that line, but one 00:38:20.893 --> 00:38:24.953 of the big things that's hurting that is change management, right? 00:38:24.953 --> 00:38:29.023 Which is, okay, I got this agent that's more efficient 00:38:29.813 --> 00:38:32.273 and, and cheaper than a person. 00:38:32.743 --> 00:38:34.103 How do I just replace that? 00:38:34.743 --> 00:38:38.413 It's very hard, and people aren't used to that, and that's gonna take some time. 00:38:39.103 --> 00:38:42.963 And, um, and again, we're gonna see evolutions of all kinds of different 00:38:42.963 --> 00:38:45.763 things here and, and, uh, um… 00:38:46.133 --> 00:38:49.453 But a lot of, lot of cool things going on around agentic right now. 00:38:49.823 --> 00:38:53.413 But it's not moving, I guess a lot-- I've said it, I'll say it again. 00:38:53.413 --> 00:38:57.453 It's not moving at the pace that the press is talking about 00:38:58.165 --> 00:38:59.606 Scott: Uh, but forecast a little bit. 00:38:59.635 --> 00:39:01.106 What, what is your view on… 00:39:01.106 --> 00:39:04.946 I, and I know 10 years is kind of too, too challenging, but you're 00:39:04.966 --> 00:39:07.145 predicting baseball games in the future. 00:39:07.145 --> 00:39:10.526 You're predicting if people are gonna jump from AT&T. 00:39:10.865 --> 00:39:13.445 What, what do you think AI's gonna look like, though? 00:39:13.465 --> 00:39:16.696 Is it gonna have the huge impact that, that the press is talking 00:39:16.696 --> 00:39:20.725 about, we're gonna have layoffs, we're gonna replace people in three years? 00:39:21.305 --> 00:39:21.705 Tim Berry: Yeah. 00:39:23.025 --> 00:39:28.285 I, um, um, I-- one thing I'm pretty positive of, and you call me a bit old 00:39:28.285 --> 00:39:33.015 school, is I c- you know, I've s- there's a, there's a big paper out there, um, 00:39:33.015 --> 00:39:36.745 I, I wish I remember the, the name of the paper, but it got a lot of press, 00:39:36.745 --> 00:39:44.815 especially in the markets, around, um, the unemployment rates going to 30 40, 50%. 00:39:44.815 --> 00:39:47.835 A, um, AI coming in and replacing jobs and I just don't… 00:39:47.905 --> 00:39:50.375 I-- it's that-- I can't see that happening. 00:39:50.785 --> 00:39:58.065 I cannot see a world where-- that we allow that as human beings, that to happen. 00:39:58.255 --> 00:39:59.795 We just need to be productive. 00:39:59.795 --> 00:40:00.315 We need to… 00:40:00.585 --> 00:40:06.325 So that, that whole Armageddon side of AI, uh, of AI, can't see happening, right? 00:40:06.515 --> 00:40:09.815 Now, then I, then I say to myself, "Well, I wonder what…" 00:40:10.105 --> 00:40:14.395 And, and unfortunately, how old I am, I lived through some of 00:40:14.395 --> 00:40:16.035 this, the whole tech revolution. 00:40:16.385 --> 00:40:20.495 Remember back in the '80s and '90s, and I talked about it took us two and a 00:40:20.495 --> 00:40:26.755 half weeks to run a logistic regression on a mainframe computer, right? 00:40:26.825 --> 00:40:31.255 That's a, that's a whole revolution that went on, and technology made things 00:40:31.255 --> 00:40:36.685 so much easier to do, and people were worried about losing their job then. 00:40:37.155 --> 00:40:42.525 And, uh, and in many ways, this is just another phase of that, but on steroids. 00:40:42.695 --> 00:40:46.585 I think this is on steroids compared to that, and the whole on steroids 00:40:46.585 --> 00:40:51.865 is gonna be, uh, interesting what that really means and how fast it is. 00:40:52.205 --> 00:40:54.935 Um, and so I think AI is gonna be great. 00:40:54.975 --> 00:40:56.455 I think it's gonna save lives. 00:40:56.805 --> 00:40:59.645 I, I, I love this concept of AI for good. 00:40:59.975 --> 00:41:04.305 It's gonna do some bad things as well, just like technology does some bad things. 00:41:04.775 --> 00:41:05.965 Um, and so… 00:41:06.185 --> 00:41:10.895 But hopefully, um, it'll, it'll just be much more of an accelerator for us and 00:41:10.895 --> 00:41:13.235 make us easier, make our lives easier 00:41:13.446 --> 00:41:17.096 Scott: So, so we at Barry are trying to use it to save lives, uh, in 00:41:17.391 --> 00:41:17.991 Tim Berry: Yes 00:41:18.096 --> 00:41:19.406 Scott: drug development the whole time. 00:41:19.766 --> 00:41:26.156 Uh, I, I'm interested in maybe even somewhat, uh, uh, um, uh, thinking 00:41:26.156 --> 00:41:27.856 of your, your nephew Cooper. 00:41:28.176 --> 00:41:31.406 He's a, he's going to be a senior next year in college. 00:41:31.415 --> 00:41:32.606 He's a math major. 00:41:32.606 --> 00:41:34.326 He's a quantitative kid. 00:41:35.196 --> 00:41:38.125 People like that, I, I don't know scared is the right word, they're, 00:41:38.156 --> 00:41:40.846 they're uncertain as to what this is. 00:41:40.886 --> 00:41:43.106 Advice to somebody like that? 00:41:43.506 --> 00:41:46.346 Should they become AI engineers? 00:41:46.346 --> 00:41:49.136 Is this a skill that everybody's gonna have? 00:41:49.495 --> 00:41:52.186 Uh, within it you gotta be so- I mean, what's advice to, 00:41:52.275 --> 00:41:53.826 to, to your nephew Cooper? 00:41:54.185 --> 00:41:54.775 Tim Berry: Yeah. 00:41:54.995 --> 00:41:55.315 Yeah. 00:41:55.505 --> 00:41:55.765 Yeah. 00:41:55.765 --> 00:41:56.375 It's, it's… 00:41:56.735 --> 00:41:57.995 I, I get that question a lot. 00:41:57.995 --> 00:42:00.705 One is I just wouldn't over-engineer it. 00:42:01.175 --> 00:42:04.875 Don't, you know, like, like I can't tell you the people, some of the 00:42:04.875 --> 00:42:09.755 smartest people I've meet in the corporate world that didn't plan 00:42:09.755 --> 00:42:11.315 to do what they're doing, right? 00:42:11.345 --> 00:42:12.175 And so you, you… 00:42:12.215 --> 00:42:15.815 No matter what you do in school and as you're looking into the workplace, 00:42:16.095 --> 00:42:20.115 your background's important, but I wouldn't over-engineer that 'cause 00:42:20.115 --> 00:42:24.045 you don't have this skill or that skill, that you can't get places. 00:42:24.295 --> 00:42:28.185 So I, my, my advice to someone like Cooper would be be aware. 00:42:28.435 --> 00:42:29.085 Absolutely. 00:42:29.085 --> 00:42:33.535 The more you can be aware of these technologies, understand them, 00:42:33.885 --> 00:42:38.165 be, be, um, uh, knowledgeable in them, can only help you, right? 00:42:38.165 --> 00:42:45.215 I'm a big believer i- in, um, learning even after school, even I, I, I can't 00:42:45.215 --> 00:42:48.755 tell you how many business books I've read, 'cause that became a passion for me. 00:42:49.135 --> 00:42:50.495 But get in, get… 00:42:50.495 --> 00:42:55.125 Just get in a spot where you could start to have an impact somewhere. 00:42:55.465 --> 00:42:56.925 Take a job, right? 00:42:56.935 --> 00:42:57.925 Start to learn. 00:42:57.935 --> 00:43:00.495 Get your head up around what do I need to do? 00:43:00.495 --> 00:43:05.155 Where's the, where's the, the, the, the, the, you know, the business going? 00:43:05.155 --> 00:43:06.345 Where's the opportunity? 00:43:06.645 --> 00:43:09.145 Where's the gaps, um, in there? 00:43:09.195 --> 00:43:12.015 And so I think people tend to worry too much about I don't have this 00:43:12.015 --> 00:43:13.115 background, I don't have that. 00:43:13.355 --> 00:43:15.885 You, you've got the opportunity to get that background in a business. 00:43:16.505 --> 00:43:18.185 Just get in a spot where you can do it 00:43:19.718 --> 00:43:20.448 Scott: Fantastic. 00:43:20.448 --> 00:43:23.408 I, a-and I'll go back to you started this, you left the University 00:43:23.408 --> 00:43:26.678 of Minnesota saying, you know, "I, I've learned a bunch, I'm 00:43:26.697 --> 00:43:27.177 Tim Berry: Yeah 00:43:27.198 --> 00:43:30.418 Scott: work," and you probably learned a heck of a lot more working, 00:43:30.758 --> 00:43:33.058 uh, and, and continue to learn, uh, 00:43:33.231 --> 00:43:33.491 Tim Berry: Yep. 00:43:33.767 --> 00:43:34.017 Scott: Yeah. 00:43:34.241 --> 00:43:34.631 Tim Berry: Yep. 00:43:34.911 --> 00:43:35.161 Yep. 00:43:35.548 --> 00:43:35.798 Scott: All 00:43:35.811 --> 00:43:35.991 Tim Berry: sure 00:43:36.208 --> 00:43:37.098 Scott: fantastic. 00:43:37.107 --> 00:43:39.818 Well, appreciate you joining us here In the Interim. 00:43:40.489 --> 00:43:43.949 Tim Berry: Yes, hopefully my story could be of some help to, to people. 00:43:43.949 --> 00:43:47.399 And, um, you know, I know some of you are connected to Scott, probably all 00:43:47.449 --> 00:43:52.419 of you, and I'm always happy to have an additional conversation if you have a 00:43:52.419 --> 00:43:56.559 certain situation and would like some, you know, advice or how I dealt with it. 00:43:56.919 --> 00:43:58.389 Uh, open invitation there 00:43:58.790 --> 00:43:59.460 Scott: Fantastic. 00:43:59.460 --> 00:44:01.840 Well, thank you very much, Tim, and, uh, thank, 00:44:01.923 --> 00:44:02.583 Tim Berry: for that, Miss Scott 00:44:02.750 --> 00:44:05.680 Scott: thanks everybody for joining us, and until next time, 00:44:05.680 --> 00:44:07.590 we'll be here, uh, in the interim