Dec. 10, 2025

The $10B Insight Every Lender Needs

Apple Podcasts podcast player iconSpotify podcast player icon
Apple Podcasts podcast player iconSpotify podcast player icon

Managing $10 billion in loans teaches you a lot about what truly drives repayment and performance. In this episode of Leaders in Lending, host Lynn Sautter Beal sits down with Jesse Obbink, vice president and general manager of servicing at Upstart, to break down the operational, analytical and human insights that matter most once a loan is on the books.

Jesse shares why servicing is “underwriting with an operational twist,” how data and machine learning help shape borrower outcomes and why empathy often outperforms traditional collections tactics. They also discuss what scale unlocks, how to manage dozens of vendor relationships effectively and how LLMs are transforming frontline agent workflows.

Whether you lead a credit union, bank or fintech, this conversation reframes servicing as a strategic lever for reducing losses, improving repayment and strengthening long-term performance.

WEBVTT

1
00:00:00.080 --> 00:00:03.080
It allows us to move from just predicting who will

2
00:00:03.120 --> 00:00:06.960
default to actually shaping the default. We're managing more than

3
00:00:07.000 --> 00:00:09.759
ten billion dollars in loans today, and while that might

4
00:00:09.800 --> 00:00:13.080
seem paltry compared to maybe some of the big five banks,

5
00:00:13.240 --> 00:00:16.359
of course it's a really, really large portfolio. If you

6
00:00:16.359 --> 00:00:19.239
think about all of those barbers who are delinquent, and

7
00:00:19.640 --> 00:00:22.719
you take the ones who are definitely going to pay

8
00:00:22.760 --> 00:00:25.320
you back whether or not you talk to them, well

9
00:00:25.480 --> 00:00:28.399
then you definitely shouldn't talk to them. Lms are, of

10
00:00:28.480 --> 00:00:33.240
course transformative. They're an unbelievable technology layer on our work,

11
00:00:33.520 --> 00:00:36.560
and I sort of imagine that they will eventually be

12
00:00:36.719 --> 00:00:38.679
diffused into almost all.

13
00:00:38.560 --> 00:00:39.359
Of our activities.

14
00:00:43.799 --> 00:00:46.840
Welcome back to Leaders Lending, where we explore what's next

15
00:00:46.880 --> 00:00:50.039
in the worlds of lending and financial innovation. I'm your host,

16
00:00:50.079 --> 00:00:52.560
Lynd Saderbil, and today we're talking about a part of

17
00:00:52.640 --> 00:00:56.479
lending that doesn't often get the spotlight. Servicing. It's where technology,

18
00:00:56.600 --> 00:01:00.240
operations and customer empathy all meet, and at ups art

19
00:01:00.320 --> 00:01:03.880
it's becoming a powerful driver of performance and borrowers success.

20
00:01:04.159 --> 00:01:07.319
Joining me today is Jesse Opink, who's the general manager

21
00:01:07.359 --> 00:01:10.400
of Servicing at Upstart. Jesse, do you want to introduce

22
00:01:10.480 --> 00:01:11.959
yourself for our listeners.

23
00:01:12.359 --> 00:01:16.840
Ellen, thanks for having me excited to be here. I

24
00:01:16.920 --> 00:01:21.560
joined Upstart a little bit over seven years ago, and.

25
00:01:22.079 --> 00:01:23.439
I came up on the product side.

26
00:01:23.439 --> 00:01:26.200
I was a product manager number two are our product

27
00:01:26.239 --> 00:01:31.120
management function was relatively late to develop, you know, and

28
00:01:31.519 --> 00:01:33.879
I've gotten a chance to work across our personal owns function,

29
00:01:33.959 --> 00:01:36.400
a little bit in growth, product, a little bit with

30
00:01:36.519 --> 00:01:40.480
machine learning, and and now for for the last two years,

31
00:01:40.640 --> 00:01:44.159
I've been the general manager of servicing, you know, which

32
00:01:44.200 --> 00:01:48.719
I think really combines my passion for great customer experiences,

33
00:01:49.319 --> 00:01:52.120
you know, with with my desire to run scaled in

34
00:01:52.879 --> 00:01:54.640
and highly effective businesses.

35
00:01:55.079 --> 00:01:55.519
Awesome.

36
00:01:55.640 --> 00:01:58.159
Well, I think the one thing I think, you know,

37
00:01:58.200 --> 00:01:59.959
we we do a lot of unique things here at Upstart,

38
00:02:00.280 --> 00:02:02.920
and we do I think a lot of them very well.

39
00:02:03.359 --> 00:02:05.719
And this is an area that's been a big area

40
00:02:05.760 --> 00:02:08.919
of focus for the past couple of years as you've

41
00:02:09.319 --> 00:02:13.639
become the general manager and have been really leading the servicing, product,

42
00:02:13.719 --> 00:02:18.319
engineering and operations teams. One of the things you've said before,

43
00:02:18.319 --> 00:02:19.919
and I think you may have shared this at the

44
00:02:19.919 --> 00:02:24.400
client summit with many of our clients and potential clients,

45
00:02:25.199 --> 00:02:28.840
is that servicing is really just underwriting with an operational twist.

46
00:02:29.599 --> 00:02:30.520
What did you mean by that?

47
00:02:31.479 --> 00:02:34.759
Yeah, you know, as you as you mentioned, I have

48
00:02:35.000 --> 00:02:40.919
been at Upstart for I guess seven years last Saturday,

49
00:02:41.520 --> 00:02:44.560
so just just past the seven year mark. And of

50
00:02:44.599 --> 00:02:47.879
course I had spent the first five years working, you know,

51
00:02:48.000 --> 00:02:52.400
very deeply in upstarts core business that is predicting who

52
00:02:52.560 --> 00:02:56.719
will default and then pricing them correctly, our our underwriting business.

53
00:02:57.280 --> 00:02:59.599
And so when I joined servicing a little over two

54
00:02:59.639 --> 00:03:02.319
years ago, I really tried to think quite a lot

55
00:03:02.319 --> 00:03:05.039
about what it meant for us to take our core

56
00:03:05.120 --> 00:03:09.759
competency and apply those skills to this relatively new space.

57
00:03:10.599 --> 00:03:11.960
And I guess the way that I would say is

58
00:03:12.000 --> 00:03:15.919
that in underwriting, your job is to take maybe all

59
00:03:16.120 --> 00:03:19.319
of these applicants. You're taking applicants, you're sort of lining

60
00:03:19.360 --> 00:03:21.479
them up in riskiness, you're spacing them not correctly, and

61
00:03:22.120 --> 00:03:27.479
then you're determining generally what risk based price is associated

62
00:03:27.520 --> 00:03:29.960
with the level of risk that you see for each applicant.

63
00:03:30.360 --> 00:03:33.000
And I think in many ways you really want to

64
00:03:33.039 --> 00:03:35.840
do the same kind of thing with servicing. You want

65
00:03:35.879 --> 00:03:38.199
to maybe take all of these loans, you want to

66
00:03:38.199 --> 00:03:42.319
line them up from least risky to most risky. But

67
00:03:42.400 --> 00:03:46.520
in underwriting you have this sort of unfair advantage that

68
00:03:46.680 --> 00:03:51.319
is that you can maybe take any person who's risk

69
00:03:51.560 --> 00:03:55.400
is too high, and if you don't like how risky

70
00:03:55.439 --> 00:03:57.919
they are, well you just don't offer them alone. And

71
00:03:57.960 --> 00:04:01.759
in servicing we actually have a little bit more operational complexity.

72
00:04:01.800 --> 00:04:05.560
We are forced to maintain our relationships with these customers,

73
00:04:05.639 --> 00:04:08.639
and so instead of being able to maybe just say, oh,

74
00:04:08.680 --> 00:04:10.960
this person is too risky and we're not going to

75
00:04:11.000 --> 00:04:13.360
work with them at all, we instead have to maybe

76
00:04:13.439 --> 00:04:16.800
look only at the range of operational levels that we

77
00:04:16.920 --> 00:04:19.959
have to change their riskiness today, to move from how

78
00:04:20.040 --> 00:04:23.399
risky they are today to how risky they might be tomorrow.

79
00:04:23.720 --> 00:04:27.000
And I think of this as a really really powerful

80
00:04:27.360 --> 00:04:31.920
part of our overall opportunity at Upstart, because it allows

81
00:04:31.959 --> 00:04:35.439
us to move from just predicting whole default to actually

82
00:04:36.000 --> 00:04:37.959
shaping little defaults.

83
00:04:38.160 --> 00:04:39.079
That's a little bit what I mean.

84
00:04:39.600 --> 00:04:42.360
Yeah, Now it's super I think super interesting is really

85
00:04:42.480 --> 00:04:47.399
kind of applying the the data science or behavioral predictions

86
00:04:47.839 --> 00:04:51.720
after origination, I'm kind of dynamically handling that relationship as

87
00:04:51.720 --> 00:04:55.079
you move forward with the customer. I like the idea

88
00:04:55.120 --> 00:04:57.600
of what you said of You know, if you identify

89
00:04:57.600 --> 00:05:00.959
a higher risk pre origination, you can than their APR.

90
00:05:01.079 --> 00:05:04.639
If you identify that after you have to still figure

91
00:05:04.639 --> 00:05:06.800
out a way to work with that customer and to

92
00:05:06.839 --> 00:05:08.839
maximize the returns that you're getting from them.

93
00:05:09.399 --> 00:05:13.160
That's something that we've been actually really focused on in Servicing.

94
00:05:13.480 --> 00:05:16.879
That is like, oh, we you know, not only want

95
00:05:16.879 --> 00:05:19.920
to be good at the risk prediction part of our work,

96
00:05:20.600 --> 00:05:23.439
but we also want to have a really wide range

97
00:05:23.560 --> 00:05:27.480
of operational levers. And so one way tactically that this

98
00:05:27.519 --> 00:05:29.720
has played out for us is that we really want

99
00:05:29.759 --> 00:05:36.720
to have collectors with a range of experience and persuasive capabilities.

100
00:05:37.079 --> 00:05:39.680
You know, you might imagine that maybe on the extreme,

101
00:05:40.000 --> 00:05:44.800
if you are collecting on very very small dollar amounts,

102
00:05:44.560 --> 00:05:47.879
it's it's probably actually not worth it for you to

103
00:05:48.160 --> 00:05:51.519
have someone really high end making that call. You might

104
00:05:51.519 --> 00:05:54.759
actually even want to stick with just digital only collections.

105
00:05:54.759 --> 00:05:56.800
But on the other hand, if you have someone who

106
00:05:57.079 --> 00:05:59.839
maybe owes you tens of thousands of dollars, maybe has

107
00:05:59.839 --> 00:06:03.160
a a secured asset that they really care about, you know,

108
00:06:03.199 --> 00:06:05.639
and and maybe need some coaching to get to a

109
00:06:05.680 --> 00:06:09.519
good a good outcome, well, that starts to look a

110
00:06:09.600 --> 00:06:12.360
whole lot more like artisanal sales. It looks like a

111
00:06:12.399 --> 00:06:16.560
really persuasive and engaged conversation, you know. And and so uh,

112
00:06:16.600 --> 00:06:18.959
I guess our work in the last couple of years

113
00:06:19.000 --> 00:06:22.439
has been both trying to build out our risk predictive capabilities,

114
00:06:22.600 --> 00:06:25.240
the ability to decide which loans are risky and which

115
00:06:25.279 --> 00:06:29.120
are not, but also a much broader range of operational

116
00:06:29.199 --> 00:06:34.000
capabilities from collections, hardships, settlements, uh, you know, and uh

117
00:06:34.040 --> 00:06:35.560
and recovery options.

118
00:06:36.680 --> 00:06:38.920
No, that's I think that's super interesting. And you know,

119
00:06:39.120 --> 00:06:41.160
and certainly the business that you're running has a really

120
00:06:41.199 --> 00:06:44.040
large scale, uh that it is a large business that

121
00:06:44.079 --> 00:06:47.720
we are servicing and across multiple products now versus just

122
00:06:47.800 --> 00:06:52.480
the the original personal loan uh fixed rate installment loan

123
00:06:52.519 --> 00:06:55.720
product that Upstart is known for. So maybe talk to

124
00:06:55.759 --> 00:06:58.199
me a little bit about like why is the servicing

125
00:06:58.360 --> 00:07:02.879
operation different, particularly for the size of the organization.

126
00:07:03.680 --> 00:07:07.600
Yeah, I definitely do think there are some parts of

127
00:07:07.639 --> 00:07:11.199
servicing that are really well plumbed, you know, and well

128
00:07:11.240 --> 00:07:15.000
known in the industry, you know, and then some parts

129
00:07:15.040 --> 00:07:18.879
that really give us some opportunity. And maybe I think

130
00:07:18.920 --> 00:07:22.839
of us generally as we scale as having the opportunity

131
00:07:22.879 --> 00:07:27.000
to do more and more things with novelty. That is, like,

132
00:07:27.639 --> 00:07:30.360
you know, we're managing more than ten billion dollars in

133
00:07:30.680 --> 00:07:33.759
loans today. And while that might seem paltry compared to

134
00:07:33.879 --> 00:07:36.959
maybe some of the big five banks, you know, of course,

135
00:07:37.000 --> 00:07:40.360
it's a it's a really really large portfolio, you know

136
00:07:40.360 --> 00:07:44.000
when you think about it at an absolute level, and

137
00:07:44.319 --> 00:07:47.040
that sort of gives us the opportunity to really make

138
00:07:47.199 --> 00:07:51.600
deep investments in our technology infrastructure. You know, we have

139
00:07:51.920 --> 00:07:56.160
a large engineering team. We have not that large, but

140
00:07:56.759 --> 00:08:00.920
very high talent and growing machine learning team, you know,

141
00:08:00.920 --> 00:08:04.199
and we have a set of vendor partnerships you know,

142
00:08:04.319 --> 00:08:09.279
in the in the low low dozens, and these are

143
00:08:09.680 --> 00:08:14.839
these are kinds of infrastructure and partnerships that you know,

144
00:08:14.920 --> 00:08:17.399
most lenders just can't justify on their own. If you

145
00:08:17.439 --> 00:08:20.120
think about, you know, maybe trying to set up this

146
00:08:20.319 --> 00:08:23.839
kind of a technology organization. You know, if you were

147
00:08:24.040 --> 00:08:27.759
managing a portfolio of maybe uh, you know, one tenth

148
00:08:27.879 --> 00:08:30.800
or one fifth our size, then trying to amortize those

149
00:08:30.800 --> 00:08:33.879
costs wouldn't make sense. And you know, we're really quite

150
00:08:33.879 --> 00:08:37.759
excited about the opportunity to build out that technology in

151
00:08:37.799 --> 00:08:41.600
a way that deliberately plans for scale and then use

152
00:08:41.720 --> 00:08:45.720
the sort of operational leverage of our scale to reinvest

153
00:08:45.919 --> 00:08:47.480
in more technology build out.

154
00:08:47.840 --> 00:08:48.039
Yeah.

155
00:08:48.120 --> 00:08:49.960
Now, I think that's a really interesting point, Like you

156
00:08:50.000 --> 00:08:52.120
think about the things that upstart is known for, and

157
00:08:52.159 --> 00:08:55.360
we do really well our core competencies, and then applying

158
00:08:55.399 --> 00:08:57.879
it to this area and really being able to do

159
00:08:58.120 --> 00:09:01.679
kind of the basic things that content acts, compliance, collections,

160
00:09:01.720 --> 00:09:04.480
all the specialty items you mentioned, but then really having

161
00:09:04.559 --> 00:09:09.200
that enterprise level infrastructure run like supporting it and learning

162
00:09:09.240 --> 00:09:12.519
from all of those repayment events that your team sees

163
00:09:12.559 --> 00:09:13.000
every day.

164
00:09:13.480 --> 00:09:15.360
To bring it back to that first point that I

165
00:09:15.399 --> 00:09:17.480
was sharing a little bit earlier, I think this is

166
00:09:17.519 --> 00:09:20.519
a perfect machine learning problem. That is, it's very high

167
00:09:20.559 --> 00:09:24.720
analytical complexity, you know. That is, Oh, what we care

168
00:09:24.799 --> 00:09:28.759
about in servicing is not really whether someone pays us

169
00:09:28.840 --> 00:09:32.840
just tomorrow, but whether they pay us over the next three,

170
00:09:33.000 --> 00:09:36.519
five or seven years depending on the loan term, you know,

171
00:09:36.679 --> 00:09:41.480
and deciding whether we should be really pursuing collections hard,

172
00:09:41.600 --> 00:09:43.879
whether we should be approaching someone with an offer of

173
00:09:44.759 --> 00:09:47.679
a modified payment schedule for short term relief or actually

174
00:09:47.720 --> 00:09:52.559
offering a much longer term and more permissive settlement. Well, well,

175
00:09:52.600 --> 00:09:55.559
this is this is something that has an objectively correct

176
00:09:55.679 --> 00:10:00.440
answer but is extremely extremely hard to predict. So the

177
00:10:00.480 --> 00:10:05.039
combination of high analytical complexity, massive personalization, and as we

178
00:10:05.080 --> 00:10:07.879
were just talking about this sort of ten years of

179
00:10:07.960 --> 00:10:11.080
data from Upstart's history and servicing, I think this is

180
00:10:11.240 --> 00:10:14.799
this is something that really gives us a great position

181
00:10:14.919 --> 00:10:18.440
to build out the kind of scaled program that it

182
00:10:18.440 --> 00:10:21.320
would just not be economical for smaller portfolios to deliver.

183
00:10:21.879 --> 00:10:22.159
Yeah.

184
00:10:22.240 --> 00:10:24.320
Absolutely, I think the scale is important. And you know,

185
00:10:24.360 --> 00:10:26.720
it's a lot of the reason that that partners work

186
00:10:26.759 --> 00:10:30.960
with us on the on the underwriting and customer acquisition side, right, Like,

187
00:10:31.039 --> 00:10:35.360
we have the very large marketing engine to to identify

188
00:10:35.399 --> 00:10:38.480
and acquire the customers that they will want to convert

189
00:10:38.519 --> 00:10:41.759
into members or or into customers and cross sell them,

190
00:10:42.240 --> 00:10:44.720
and we can apply that same scale in this area

191
00:10:45.000 --> 00:10:49.360
and really deliver a delightful experience while maximizing their returns.

192
00:10:50.440 --> 00:10:52.440
One thing that you said that that kind of caught

193
00:10:52.440 --> 00:10:54.799
my attention is that you've got in a third party's

194
00:10:55.440 --> 00:10:57.840
and vendors maybe in like the dozens that you're working with,

195
00:10:58.759 --> 00:11:01.840
and so you know that is certainly an operational challenge,

196
00:11:01.879 --> 00:11:05.080
and how do you think about all of that vendor

197
00:11:05.159 --> 00:11:08.480
management and like quality and oversight and more of kind

198
00:11:08.519 --> 00:11:11.879
of the core operational function of working with those sorts

199
00:11:11.879 --> 00:11:13.519
of BPOs and vendors.

200
00:11:14.080 --> 00:11:15.960
I think this is such a good question because there's

201
00:11:16.080 --> 00:11:20.879
really a strong tension in servicing and that is maybe

202
00:11:21.000 --> 00:11:24.279
number one that I want to invest our energy in

203
00:11:24.399 --> 00:11:27.720
doing things that we can do differently and better than

204
00:11:28.000 --> 00:11:32.759
other kinds of organizations. You know, and boy call centers,

205
00:11:32.799 --> 00:11:35.519
they've been around for a long time, They've been optimized

206
00:11:35.799 --> 00:11:38.639
in really really deep ways. And at the same time,

207
00:11:39.120 --> 00:11:41.519
you know, we really want to set a very very

208
00:11:41.600 --> 00:11:44.120
high quality bar and we want to make sure we're

209
00:11:44.159 --> 00:11:48.000
delivering better than simply industry standard. And the way that

210
00:11:48.039 --> 00:11:50.840
we think about that in relationship to vendors is really

211
00:11:50.879 --> 00:11:54.159
that it starts with pretty clear accountability. We have this

212
00:11:54.360 --> 00:11:57.600
dual ownership model where both our internal teams and our

213
00:11:57.639 --> 00:12:02.000
partners really share responsibility for these outcomes. And the way

214
00:12:02.000 --> 00:12:05.440
that we sort of materialize that in the world. While

215
00:12:05.440 --> 00:12:07.840
we run random audits, of course, we try to spend

216
00:12:07.879 --> 00:12:08.919
a lot of time with our vendors.

217
00:12:08.960 --> 00:12:10.159
We're doing a lot of site visits.

218
00:12:10.440 --> 00:12:13.240
Uh, you know, we have a pretty clear performance scorecards,

219
00:12:13.720 --> 00:12:15.000
you know, and then we want to make sure that

220
00:12:15.519 --> 00:12:20.720
actually those things are backed up with very clear quality

221
00:12:20.759 --> 00:12:25.080
incentives in our contracts with those vendors. You know, we

222
00:12:25.120 --> 00:12:27.799
want to make sure that our incentives are aligned and

223
00:12:28.399 --> 00:12:31.039
when we do well, they do well, you know. And

224
00:12:31.039 --> 00:12:33.440
and generally what that means for us is like, if,

225
00:12:33.480 --> 00:12:35.799
for example, an error rate it creeps above our target,

226
00:12:36.279 --> 00:12:38.840
then we want to make sure that there are contractual penalties.

227
00:12:39.360 --> 00:12:42.200
But if conversely, quality improofs the partner is going to

228
00:12:42.240 --> 00:12:44.200
win more volume, they're going to be able to grow,

229
00:12:44.240 --> 00:12:46.600
and they're gonna be able to sort of benefit from

230
00:12:46.600 --> 00:12:50.120
that relationship with us. So you know, in this regard,

231
00:12:50.360 --> 00:12:52.559
you know, I would say that managing vendors well and

232
00:12:52.840 --> 00:12:56.120
doing doing contracts, this isn't really the flashy part of servicing.

233
00:12:56.639 --> 00:13:00.159
It's really just about building trust and enforcing consistency and

234
00:13:00.159 --> 00:13:04.120
and really identifying root causes really clearly when things go wrong.

235
00:13:04.559 --> 00:13:07.559
Yeah, and I think so less scenario where you'd maybe

236
00:13:07.559 --> 00:13:11.399
focus on novelty and more about just being precise and

237
00:13:11.440 --> 00:13:15.600
having discipline and having good communication between you and the

238
00:13:15.960 --> 00:13:17.960
vendors as well to make sure that you're all I

239
00:13:17.960 --> 00:13:20.559
think agreeing on what quality is and how you're measuring

240
00:13:20.559 --> 00:13:23.039
it is certainly critically important.

241
00:13:23.559 --> 00:13:25.360
I love that you use that word novelty.

242
00:13:26.120 --> 00:13:30.720
There's a quotation from Angela Duckworth, who I think is

243
00:13:31.080 --> 00:13:34.600
one of the foremost thinkers on grit that talks about

244
00:13:35.200 --> 00:13:38.600
how you maintain focus in an area when it becomes

245
00:13:39.200 --> 00:13:42.360
a little less fun And I just remember so clearly

246
00:13:43.000 --> 00:13:48.519
her sort of saying, Oh, your ability to substitute nuance

247
00:13:48.559 --> 00:13:51.360
for novelty, to think about doing the same thing but

248
00:13:51.440 --> 00:13:55.639
with ever increasing nuance is highly predictive of your willingness

249
00:13:55.639 --> 00:13:58.960
to stick with something, you know, even once it's no

250
00:13:59.000 --> 00:14:01.559
longer flashy, you know. And I really think this applies

251
00:14:01.600 --> 00:14:04.240
to servicings so deeply that the best servicing models they're

252
00:14:04.240 --> 00:14:07.120
not going to be they're just really relentlessly consistent.

253
00:14:07.639 --> 00:14:10.679
Yeah, that's funny. That's one of my favorite quotes, and

254
00:14:10.679 --> 00:14:12.960
I'm probably going to paraphrase it a little bit wrong.

255
00:14:13.159 --> 00:14:16.799
Is that like a differentiator between people as people who

256
00:14:16.799 --> 00:14:19.600
are willing to sprint when the finish line is unclear,

257
00:14:20.600 --> 00:14:22.519
but are willing to sprint anyways.

258
00:14:22.559 --> 00:14:23.879
I'm going to take that. I love it.

259
00:14:24.600 --> 00:14:26.279
So, you know, we've talked a lot about like the

260
00:14:26.320 --> 00:14:29.240
machine learning pieces, and then some of the core competencies

261
00:14:29.279 --> 00:14:31.240
and the vendors and different things that you've been working

262
00:14:31.240 --> 00:14:33.799
on with the team that maybe talk to me a

263
00:14:33.840 --> 00:14:36.200
little bit about like the people side. So when you

264
00:14:36.200 --> 00:14:39.919
think about like identifying that differentiated talent, like, as you said,

265
00:14:39.919 --> 00:14:42.440
you know, if you're collecting on say a fifty thousand

266
00:14:42.519 --> 00:14:45.399
dollars loan that's passed you, you may need a more

267
00:14:45.440 --> 00:14:48.080
skilled senior collector. You may be spending a little bit

268
00:14:48.159 --> 00:14:51.720
less effort on something that's maybe a smaller dollar or

269
00:14:51.720 --> 00:14:53.840
maybe less likely that you don't need is skilled of

270
00:14:53.879 --> 00:14:57.639
a person in there. So you know, maybe kind of

271
00:14:57.639 --> 00:15:00.639
a couple of things differentiating talent, like identify people who

272
00:15:00.639 --> 00:15:02.399
are the right fit for the different roles and what

273
00:15:02.440 --> 00:15:05.120
those roles are. And then as you think about all

274
00:15:05.159 --> 00:15:09.799
of the work with LLLMS and automation, how you kind

275
00:15:09.799 --> 00:15:11.960
of bring humans into the loop and where the humans

276
00:15:12.000 --> 00:15:15.639
are there, and then where the automation is best the

277
00:15:15.879 --> 00:15:17.480
kind of forward forward point.

278
00:15:18.080 --> 00:15:20.519
I think these are both really really rich topics. Maybe

279
00:15:20.519 --> 00:15:23.279
we can start with sort of thinking about human talent

280
00:15:23.360 --> 00:15:26.440
and then talk about how we're incorporating ELMS, because of

281
00:15:26.480 --> 00:15:31.000
course that is the technology shift of our age. On

282
00:15:31.039 --> 00:15:35.159
the human side, I guess I really tend to think

283
00:15:35.200 --> 00:15:39.240
about the first and most important thing as being able

284
00:15:39.279 --> 00:15:45.480
to understand what causal role particular human is playing in

285
00:15:45.600 --> 00:15:48.559
a servicing decision. And what I mean by that is like,

286
00:15:48.759 --> 00:15:51.720
if you think about all of those barbers who are delinquent,

287
00:15:52.320 --> 00:15:55.759
and you take the ones who are definitely going to

288
00:15:55.759 --> 00:15:57.799
pay you back whether or not you talk.

289
00:15:57.720 --> 00:16:00.399
To them, well, then you definitely shouldn't talk to them.

290
00:16:00.759 --> 00:16:00.919
You know.

291
00:16:01.080 --> 00:16:04.440
There's no there's no reason to spend your time and

292
00:16:04.480 --> 00:16:06.919
money trying to reach out to somebody who is sort

293
00:16:06.919 --> 00:16:09.120
of gonna pay you no matter what. In that case,

294
00:16:09.159 --> 00:16:12.480
the human would have no causal impact. And maybe on

295
00:16:12.759 --> 00:16:15.200
the way other end of the extreme, if we take

296
00:16:15.200 --> 00:16:17.279
people who are definitely not going to pay you back

297
00:16:18.000 --> 00:16:20.799
whether or not you call them, well then don't call them.

298
00:16:20.879 --> 00:16:24.639
You're wasting their time and you're wasting your own money,

299
00:16:25.159 --> 00:16:28.039
you know, and again you're having really no causal impact.

300
00:16:28.720 --> 00:16:31.519
And so the first question for us has really been

301
00:16:31.600 --> 00:16:36.159
trying to isolate the causal impact of particular individuals in

302
00:16:36.279 --> 00:16:39.320
changing people's mind I often talk about the best collectors

303
00:16:39.600 --> 00:16:43.960
as being highly empathetic, highly persuasive, you know, and there's

304
00:16:44.000 --> 00:16:47.120
a there's a big analytical question in sort of how

305
00:16:47.279 --> 00:16:51.320
we attribute a certain amount of the collected dollar value

306
00:16:51.799 --> 00:16:54.360
to the exact person who was on the phone at

307
00:16:54.399 --> 00:16:55.080
a given time.

308
00:16:55.559 --> 00:16:57.879
This is one really really interesting question.

309
00:16:58.720 --> 00:17:02.360
But even once you have solved that pretty rich analytical problem,

310
00:17:02.440 --> 00:17:05.400
and we have not solved it, we're sort of making advances,

311
00:17:05.440 --> 00:17:07.279
but it's not all the way solved by any means.

312
00:17:07.559 --> 00:17:11.960
Even once you've solved that really rich analytical problem, well,

313
00:17:12.079 --> 00:17:16.319
then you can sort of think about, oh, well, are

314
00:17:16.400 --> 00:17:19.960
the kind of people who demonstrate the greatest causal impact.

315
00:17:20.240 --> 00:17:22.680
What kind of background do they have? You know, do

316
00:17:22.720 --> 00:17:24.440
we want to actually hire for that background? Do we

317
00:17:24.480 --> 00:17:25.480
want to train for it?

318
00:17:26.279 --> 00:17:26.519
You know?

319
00:17:26.640 --> 00:17:29.720
What are we actually willing to pay? You know, of

320
00:17:29.720 --> 00:17:33.039
course you can get generally a different kind of person

321
00:17:33.160 --> 00:17:35.519
if you're willing to pay a lot more. And you know,

322
00:17:35.799 --> 00:17:39.680
I aspire for us to maybe have some collections roles

323
00:17:40.160 --> 00:17:43.319
that are highly competitive with inside sales. You know, I

324
00:17:43.359 --> 00:17:46.119
think that generally, if you think about the profile of

325
00:17:46.279 --> 00:17:48.839
a collector, we really want that profile to look more

326
00:17:48.880 --> 00:17:53.119
like an inside salesperson and less like you know, a

327
00:17:53.160 --> 00:17:56.079
call center operator, you know, and then the third thing

328
00:17:56.599 --> 00:17:58.599
really how are you incentivizing them?

329
00:17:58.599 --> 00:18:00.440
What actions are you encouraging them to take?

330
00:18:00.519 --> 00:18:02.680
And and what are you rewarding them for and and

331
00:18:02.759 --> 00:18:05.039
so I think the whole structure on the human side

332
00:18:05.759 --> 00:18:10.039
is really really nuanced and complex in terms of combining, uh,

333
00:18:10.279 --> 00:18:13.839
the background of your collectors, the overall compensation, and then

334
00:18:14.279 --> 00:18:17.680
sort of how you mix based comp variable comp with

335
00:18:17.759 --> 00:18:21.319
specific collections based incentives. If we shift to the l

336
00:18:21.519 --> 00:18:25.119
M side of things, l lms are, are, of course,

337
00:18:25.559 --> 00:18:30.880
uh transformative, They're They're an unbelievable technology layer on on

338
00:18:30.880 --> 00:18:33.720
our work and I sort of imagine that they will

339
00:18:33.720 --> 00:18:38.200
eventually be diffused into almost all of our activities. That

340
00:18:38.319 --> 00:18:40.240
that is, like, we want them to be part of

341
00:18:40.279 --> 00:18:41.960
our software, We want them to be part of our

342
00:18:42.000 --> 00:18:44.680
internal tooling, you know, and of course we want to

343
00:18:44.759 --> 00:18:51.799
be part of our operational interactions, uh with our operational interactions.

344
00:18:51.480 --> 00:18:53.519
Uh with borrowers, you know.

345
00:18:53.640 --> 00:18:57.920
And the way that that's manifested today, well, we generally

346
00:18:58.519 --> 00:19:03.880
have already rolled out programs that allow for our own

347
00:19:03.960 --> 00:19:08.519
frontline agents to benefit from better M assisted tooling. We

348
00:19:08.599 --> 00:19:13.839
have LMS analyzing, analyzing callscripts, maybe helping to identify where

349
00:19:13.839 --> 00:19:17.759
there are complaints, helping to categorize where we have opportunities.

350
00:19:18.359 --> 00:19:22.599
You know, we have lms assisting in our training, trying

351
00:19:22.640 --> 00:19:26.319
to you know, role play with some of our incoming

352
00:19:26.359 --> 00:19:30.559
operational agents, and of course, as with others, you know,

353
00:19:30.559 --> 00:19:34.240
we really have started to put lms in a place

354
00:19:34.240 --> 00:19:38.359
where they're taking some of these ropes or routine high

355
00:19:38.440 --> 00:19:43.519
volume tasks like summarizing notes, triaging inbounded messages, or guiding

356
00:19:43.960 --> 00:19:46.839
income in some cases next best actions. And our frontline

357
00:19:46.839 --> 00:19:50.000
team has generally responded really really positively to this. That is, like,

358
00:19:50.720 --> 00:19:54.920
if you are a skilled and influential collector, you don't

359
00:19:54.960 --> 00:19:56.880
really want to be taking notes all day. That's not

360
00:19:56.920 --> 00:19:59.519
the part of your job that you enjoy, you know.

361
00:19:59.559 --> 00:20:02.200
And so some of the feedback that we've gotten from

362
00:20:02.240 --> 00:20:04.680
our team is that the system generally feels more organized,

363
00:20:04.720 --> 00:20:07.839
feels more understandable, feel more clear, you know. And our

364
00:20:07.839 --> 00:20:11.319
goal here is for AI to play a pretty central

365
00:20:11.400 --> 00:20:15.559
role in helping to surface relevant context instantly, which we

366
00:20:15.599 --> 00:20:18.960
think will let our best people focus on empathy, on negotiation,

367
00:20:19.240 --> 00:20:22.640
as I said, on persuasion, you know, and exercise that

368
00:20:22.640 --> 00:20:25.319
that human judgment, which we think is is still pretty

369
00:20:25.319 --> 00:20:26.279
pretty unique.

370
00:20:27.079 --> 00:20:30.680
How do you think about maybe then finding that talent

371
00:20:30.759 --> 00:20:32.680
so as you think, you know that is the role

372
00:20:32.759 --> 00:20:35.240
is changing. And I talked a lot about and I

373
00:20:35.279 --> 00:20:38.440
think it's a pretty uh maybe a unique idea that

374
00:20:38.480 --> 00:20:42.839
you want to think about. Uh, people who you know

375
00:20:42.960 --> 00:20:46.160
maybe are are currently paid more like contact center specialists,

376
00:20:46.160 --> 00:20:48.720
and in the future it could be paid and compensated

377
00:20:48.759 --> 00:20:52.400
more like an inside sale like a BDR inside sales representative.

378
00:20:53.359 --> 00:20:56.319
How do you think about identifying or and or upskilling

379
00:20:56.440 --> 00:20:58.720
people that are in these roles today, so as you

380
00:20:58.720 --> 00:21:01.599
think about helping them get comfortable with with AI l

381
00:21:01.720 --> 00:21:04.240
ll ms and and have the right skills and then

382
00:21:05.279 --> 00:21:08.039
and and that I think that mindset of that this

383
00:21:08.119 --> 00:21:10.519
is an enhancement, this will make my job better, not

384
00:21:11.000 --> 00:21:13.839
like this will just replace my job. Uh. And you know,

385
00:21:13.880 --> 00:21:15.920
focus on the opportunity and out the fear, Like how

386
00:21:15.920 --> 00:21:18.720
do you think about like identifying that talent in the

387
00:21:18.759 --> 00:21:21.599
existing pool, but then also you know, as you're as

388
00:21:21.640 --> 00:21:25.519
you're hiring and looking for that outside with with new roles.

389
00:21:26.039 --> 00:21:26.400
Yeah.

390
00:21:26.680 --> 00:21:30.160
Uh, it's it's such a good question because of course, uh,

391
00:21:30.240 --> 00:21:32.400
you know, if you're going to go out and and

392
00:21:32.440 --> 00:21:34.359
you sort of say, I bet there are a bunch

393
00:21:34.400 --> 00:21:36.240
of people who would be really good at this job,

394
00:21:36.480 --> 00:21:39.839
who would never think of themselves as good at this job.

395
00:21:40.240 --> 00:21:40.599
Uh.

396
00:21:40.680 --> 00:21:43.920
Well, that's it's a really really hard search problem. And

397
00:21:44.119 --> 00:21:45.960
I guess the way that we've thought about it is

398
00:21:46.160 --> 00:21:49.720
by really creating a virtuous cycle, a little bit of

399
00:21:49.720 --> 00:21:52.319
a flywheel. That is, Oh, first we put in place

400
00:21:52.359 --> 00:21:56.440
this measurement, trying to see how, uh you know, how

401
00:21:56.480 --> 00:22:00.559
each person is contributing to the overall impact of the team,

402
00:22:01.000 --> 00:22:03.759
and then maybe looking back at those people who are

403
00:22:04.039 --> 00:22:08.640
most impactful and trying to isolate which skills or behaviors

404
00:22:09.440 --> 00:22:13.480
they demonstrate that maybe other folks don't demonstrate, and then

405
00:22:13.640 --> 00:22:17.160
of course taking that both into our training and into

406
00:22:17.759 --> 00:22:20.599
and into our hiring, you know, so holding a higher

407
00:22:20.640 --> 00:22:27.079
bar for those skills and behaviors, you know, generally rewarding

408
00:22:27.200 --> 00:22:29.920
those skills and behaviors. And of course, as I mentioned,

409
00:22:29.920 --> 00:22:32.440
all of this is really tied into compensations. So if

410
00:22:32.440 --> 00:22:37.119
you're looking for either higher quality expression of the same

411
00:22:37.200 --> 00:22:41.119
skills and abilities or novel skills and abilities that maybe

412
00:22:41.559 --> 00:22:44.160
someone has and another person doesn't have, well you have

413
00:22:44.200 --> 00:22:46.799
to be willing to pay for that. And generally, my

414
00:22:47.000 --> 00:22:50.319
hope is that our collections team over time is better

415
00:22:50.400 --> 00:22:54.279
and better compensated because they are so good at helping

416
00:22:54.319 --> 00:22:57.559
barbers find paths back to current they're playing a really

417
00:22:57.640 --> 00:23:02.119
pivotal role in shaping people's minds, their behaviors, and ultimately

418
00:23:02.599 --> 00:23:03.799
the repayment on their loans.

419
00:23:04.559 --> 00:23:06.640
Yeah, and I think it's a unique way that we've

420
00:23:06.680 --> 00:23:11.960
talked about this in UH in in servicing here at Upstart,

421
00:23:12.000 --> 00:23:14.160
where you know, a lot of times folks can think

422
00:23:14.200 --> 00:23:17.880
about operations teams as more of a cost center and

423
00:23:17.920 --> 00:23:22.759
it's just you know, preventative or defensive in servicing UH.

424
00:23:22.799 --> 00:23:25.039
And I think the way that you've approached it and

425
00:23:25.039 --> 00:23:27.079
are thinking about it is more dynamic than that that

426
00:23:27.160 --> 00:23:30.559
this is a place where where there's returns to be

427
00:23:30.799 --> 00:23:32.799
to be had because there are people, as you said,

428
00:23:32.799 --> 00:23:36.480
who there's some people that will that will pay if

429
00:23:36.480 --> 00:23:38.960
we contact them, or will pay if we don't contact them,

430
00:23:39.039 --> 00:23:40.599
and then there are people who do the opposite, like

431
00:23:40.640 --> 00:23:43.079
we can contact them and they won't. But there's you know,

432
00:23:43.119 --> 00:23:45.359
I think you have a pretty strong belief that there's

433
00:23:45.400 --> 00:23:48.799
a subset of folks that maybe won't pay if we

434
00:23:48.799 --> 00:23:51.000
don't contact them in the right way or the right frequency.

435
00:23:51.160 --> 00:23:54.599
But our behaviors can differentiate that and can drive a

436
00:23:54.599 --> 00:23:57.200
better return. You know, maybe talk to me a little

437
00:23:57.240 --> 00:23:58.839
bit more. You know, we've already talked quite a bit

438
00:23:58.839 --> 00:24:01.799
about machine learning and service and kind of being a

439
00:24:01.799 --> 00:24:05.720
hard problem, and I don't think it's an area that

440
00:24:05.759 --> 00:24:08.079
really anyone else is spending a lot of time focusing

441
00:24:08.119 --> 00:24:10.000
on and using machine learning to solve some of these

442
00:24:10.079 --> 00:24:13.519
servicing challenges. Maybe just expound a little bit on like

443
00:24:13.599 --> 00:24:16.519
why is it so hard? And why is it so

444
00:24:16.599 --> 00:24:19.160
hard to take something that that we know how to

445
00:24:19.200 --> 00:24:21.279
do really well machine learning and apply it to this

446
00:24:21.440 --> 00:24:24.000
particular kind of set of problems.

447
00:24:24.680 --> 00:24:27.039
Yeah, it definitely is hard.

448
00:24:27.720 --> 00:24:31.680
And you know, I think there are increasingly folks who

449
00:24:31.880 --> 00:24:36.319
have awareness that that servicing is an exciting opportunity space.

450
00:24:36.359 --> 00:24:39.079
But you know, much as I hate to say it,

451
00:24:39.640 --> 00:24:42.160
when I was a little boy, I definitely did not

452
00:24:42.240 --> 00:24:43.960
grow up saying, oh, what I want to do in

453
00:24:44.000 --> 00:24:46.240
my life is run a large servicing organization.

454
00:24:46.279 --> 00:24:48.640
That that just isn't It's not it's not super.

455
00:24:48.400 --> 00:24:52.599
Common, you know, And and I think if you you know,

456
00:24:52.640 --> 00:24:57.480
if you think about the space and the relevance for

457
00:24:57.720 --> 00:25:02.200
machine learning, well, the servicing data, it's it's fundamentally, there's

458
00:25:02.200 --> 00:25:04.559
fundamentally a lot of it that's really good, you know,

459
00:25:04.599 --> 00:25:07.319
that's that's one sort of core premise for being a

460
00:25:07.319 --> 00:25:10.000
good machine learning problem. But the servicing data is it's

461
00:25:10.039 --> 00:25:14.720
really messy and it's really behavioral. That is, you're dealing

462
00:25:14.720 --> 00:25:18.000
with really complex interactions. Somebody can pay this month, not

463
00:25:18.079 --> 00:25:22.279
pay next month. You're dealing with multi channel interactions. Hardly

464
00:25:22.319 --> 00:25:25.880
anybody wants to be the lender that is doing a

465
00:25:25.920 --> 00:25:30.279
holdout where you're like, well, I wonder how impactful calling is,

466
00:25:30.680 --> 00:25:32.480
and so I'm not going to call fifty percent of

467
00:25:32.480 --> 00:25:34.599
my barbers and I'm gonna see which one's default. That's

468
00:25:34.720 --> 00:25:38.799
that's really not what anybody wants to be doing. And

469
00:25:38.880 --> 00:25:42.839
so trying to sort of inch your way toward understanding

470
00:25:42.920 --> 00:25:48.680
causality across payments, calls, hardship plans, you know, third party collections,

471
00:25:48.680 --> 00:25:51.920
first party collections. You know, the sort of evolution of

472
00:25:51.960 --> 00:25:56.799
that data space over time is really really quite complex,

473
00:25:56.880 --> 00:25:59.039
and in many ways, you know, it feels to me

474
00:25:59.200 --> 00:26:03.599
like we are inching toward an understanding, and every time

475
00:26:03.680 --> 00:26:06.000
we take a little bit of a step forward, well,

476
00:26:06.039 --> 00:26:08.400
it broadens our horizon to the future. We have a

477
00:26:08.400 --> 00:26:11.000
little bit of a sense, but there's still so much

478
00:26:11.200 --> 00:26:13.599
that we don't know that we can't quite see too yet.

479
00:26:14.119 --> 00:26:14.359
Now.

480
00:26:14.640 --> 00:26:17.000
I think like if we look at machine learning either,

481
00:26:17.119 --> 00:26:19.920
and that's sort of that's a servicing view. If we

482
00:26:19.920 --> 00:26:22.759
look at machine learning, traditional mL models, well they're mostly

483
00:26:23.160 --> 00:26:25.680
built to predict relatively static outcomes.

484
00:26:26.200 --> 00:26:26.359
You know.

485
00:26:26.559 --> 00:26:28.680
They want to predict something at a point in time.

486
00:26:29.559 --> 00:26:32.359
And the challenge for us is like, well, we're trying

487
00:26:32.359 --> 00:26:35.839
to build models that then adapt in relatively real time

488
00:26:35.920 --> 00:26:39.240
and predict a wide range of hypotheticals. That is, like

489
00:26:39.720 --> 00:26:44.279
which intervention today has the greatest chance of improving recovery

490
00:26:44.319 --> 00:26:48.920
or improving customer satisfaction tomorrow. And this is an enormous

491
00:26:48.960 --> 00:26:52.519
modeling challenge, but it's also, I think, really a natural

492
00:26:52.599 --> 00:26:55.000
fit for what Upstart is good at, and a natural

493
00:26:55.039 --> 00:26:57.920
extension of the skill set that we've been building.

494
00:26:58.039 --> 00:26:59.720
We spend a decade really.

495
00:26:59.599 --> 00:27:03.319
Trying to get these models to be good at understanding

496
00:27:03.480 --> 00:27:07.319
complex behavioral data over long term outcomes, you know, and

497
00:27:07.559 --> 00:27:11.000
we think that incorporating some of these operational feedback loops

498
00:27:11.559 --> 00:27:14.279
is a is a really really exciting next challenge.

499
00:27:14.599 --> 00:27:17.319
Yeah, and I think it definitely goes into you know,

500
00:27:17.440 --> 00:27:19.000
kind of what we started on as we were talking

501
00:27:19.039 --> 00:27:24.000
about underwriting and origination and some of the things that

502
00:27:24.039 --> 00:27:27.720
impact folks on that end, like the macroeconomic cycle, things

503
00:27:27.720 --> 00:27:30.079
like the you know, a recent government shutdown of course,

504
00:27:30.160 --> 00:27:33.119
and certain people being furloughed and not receiving their money.

505
00:27:33.519 --> 00:27:36.960
You may have borrowers who had a desire to pay

506
00:27:37.039 --> 00:27:39.079
but not an ability to pay, which is very different

507
00:27:39.119 --> 00:27:43.519
than somebody who doesn't have a willingness to pay, or

508
00:27:43.559 --> 00:27:46.039
it doesn't have either a willingness or or an ability

509
00:27:46.079 --> 00:27:46.440
to pay.

510
00:27:46.880 --> 00:27:49.240
At the end of the day, servicing is oh, I

511
00:27:49.240 --> 00:27:53.160
should say, loan performance is really a function of your underwriting,

512
00:27:53.359 --> 00:27:57.000
your macro and your servicing. And of course, you know,

513
00:27:57.079 --> 00:28:01.920
we want to play a central role in upstarts overall

514
00:28:01.960 --> 00:28:04.960
long performance. But you're absolutely right to highlight that there

515
00:28:05.000 --> 00:28:10.319
are such complex and confounding factors that understanding which part

516
00:28:10.400 --> 00:28:14.440
of that impact came from us is a really ongoing challenge.

517
00:28:15.039 --> 00:28:15.400
All right.

518
00:28:15.519 --> 00:28:17.640
So we've talked a lot about what you've been doing,

519
00:28:17.759 --> 00:28:20.200
you know, and I mean, I think it's fair to

520
00:28:20.240 --> 00:28:22.279
say that most people don't as you said, is it

521
00:28:22.720 --> 00:28:24.359
you know, six year old and aren't thinking I'm going

522
00:28:24.440 --> 00:28:26.880
to run a large servicing organization when I grow up,

523
00:28:26.920 --> 00:28:29.960
and I would imagine that six years ago you probably

524
00:28:29.960 --> 00:28:33.119
were not thinking that either before you you became the

525
00:28:33.160 --> 00:28:37.319
general manager of servicing here at Upstart. But certainly looking ahead,

526
00:28:37.519 --> 00:28:40.640
I know one of the big topics on the horizon

527
00:28:40.720 --> 00:28:43.880
for us as servicing as a service. So maybe talk

528
00:28:43.880 --> 00:28:46.160
about that, like what does that mean, and what does

529
00:28:46.160 --> 00:28:48.559
that look like, and what would I would kind of

530
00:28:48.640 --> 00:28:52.200
excite you about it, and what do you feel like

531
00:28:52.240 --> 00:28:55.559
you really need to focus on to be very really

532
00:28:55.559 --> 00:28:56.279
effective there.

533
00:28:56.799 --> 00:28:59.680
Yeah, maybe one of the first things is just to

534
00:28:59.799 --> 00:29:04.240
say I really think of good servicing as good stewardship

535
00:29:04.359 --> 00:29:08.599
of our barbers. That is, like, oh, I believe that

536
00:29:08.640 --> 00:29:13.440
by offering customers better experiences, you make them more loyal

537
00:29:13.440 --> 00:29:16.559
to you. I believe that by offering lower friction experiences,

538
00:29:16.880 --> 00:29:20.519
you make them more likely to pay you back, you know.

539
00:29:20.559 --> 00:29:24.039
And I believe that by offering empathetic hardship programs that

540
00:29:24.200 --> 00:29:28.440
match customers real life financial circumstances, you can really help

541
00:29:28.480 --> 00:29:31.880
them get back on their feet and keep a loan

542
00:29:32.480 --> 00:29:34.960
current that otherwise might have rolled to charge off. And

543
00:29:35.000 --> 00:29:39.440
so we've talked a lot about the analytical challenges of servicing,

544
00:29:39.920 --> 00:29:42.599
But at the end of the day, the customer experience

545
00:29:42.839 --> 00:29:46.960
is really front and center. The way that you create

546
00:29:47.079 --> 00:29:51.839
better loan performance outcomes is, in my opinion, by offering

547
00:29:52.400 --> 00:29:55.400
really next level and excellent customer experiences.

548
00:29:55.960 --> 00:29:58.400
And when I think about the magnitude of.

549
00:29:58.599 --> 00:30:02.559
This challenge, you know, almost within I don't know, within

550
00:30:02.599 --> 00:30:06.799
probably two months of joining the servicing team, I was

551
00:30:06.839 --> 00:30:11.200
pretty clear this is a year's long, you know, maybe

552
00:30:11.319 --> 00:30:16.000
decades long kind of problem for us to solve. This

553
00:30:16.039 --> 00:30:19.200
is not a fast problem. And my goodness, if we

554
00:30:19.319 --> 00:30:22.039
build the infrastructure to solve this problem, we shouldn't keep

555
00:30:22.039 --> 00:30:24.759
it to ourselves, you know. And that's really where servicing

556
00:30:24.799 --> 00:30:26.279
as a service comes in for me.

557
00:30:26.559 --> 00:30:26.759
You know.

558
00:30:26.799 --> 00:30:31.519
I believe that the infrastructure, the kinds of analytical approaches

559
00:30:31.559 --> 00:30:34.279
that we're building, you know, and the kinds of experiences

560
00:30:34.279 --> 00:30:37.200
that we're trying to to marry them with are going

561
00:30:37.279 --> 00:30:40.119
to be unique and gonna be really best in class.

562
00:30:40.240 --> 00:30:42.079
And I want to make sure that we're able to

563
00:30:42.440 --> 00:30:46.839
apply those excellent experiences to as many loans as possible,

564
00:30:47.559 --> 00:30:50.839
whether or not they were underwritten by Upstart or not.

565
00:30:51.640 --> 00:30:55.359
Yeah, I think that's a really exciting, you know, exciting

566
00:30:55.400 --> 00:30:58.079
idea as we think about you know, kind of optimizing

567
00:30:58.119 --> 00:31:00.359
what's working here and then as you said, share it

568
00:31:00.480 --> 00:31:02.359
and like we you know, we there's a lot of

569
00:31:02.400 --> 00:31:04.920
things I think we are have very strong competencies in

570
00:31:05.000 --> 00:31:07.079
and and this being one of them, and so kind

571
00:31:07.079 --> 00:31:10.119
of building that for for what's next, and and potentially

572
00:31:10.200 --> 00:31:12.960
servicing you know, loans that didn't even originate on our

573
00:31:12.960 --> 00:31:16.160
platform as an exciting place for us to head uh,

574
00:31:16.200 --> 00:31:18.079
and that we can really bring all of that kind

575
00:31:18.119 --> 00:31:22.599
of machine learning uh and servicing know how and applying

576
00:31:22.599 --> 00:31:25.720
it to to in a kind of a long term way,

577
00:31:25.759 --> 00:31:27.559
which is I think interesting that this is a long

578
00:31:27.640 --> 00:31:32.200
term innovation build. So I think what what are you

579
00:31:32.279 --> 00:31:35.480
most excited about? Like what's next on your agenda? That's

580
00:31:35.480 --> 00:31:39.079
probably to you the most exciting thing, uh, that that

581
00:31:39.119 --> 00:31:40.359
you'll get the chance to build.

582
00:31:41.079 --> 00:31:43.839
Yeah, our north star here is really to be able

583
00:31:43.880 --> 00:31:47.240
to service not just UPSTARTE originated loans, but partner loans

584
00:31:47.279 --> 00:31:50.640
as well. And we have a handful of things that

585
00:31:50.640 --> 00:31:54.240
that we still need to work on our core platform

586
00:31:54.359 --> 00:31:58.279
capabilities today. You know, of course our bread and butter

587
00:31:58.519 --> 00:32:01.920
is unsecured personal loans. We have lots of work to

588
00:32:01.960 --> 00:32:06.640
do to maybe UH build out some of those UH

589
00:32:06.680 --> 00:32:12.079
revolving credit capabilities UH for loan types like heylocks or

590
00:32:12.079 --> 00:32:14.319
credit cards, you know, both of which make up a

591
00:32:14.400 --> 00:32:19.759
very large part of the third party servicing UH market today.

592
00:32:19.799 --> 00:32:20.000
You know.

593
00:32:20.200 --> 00:32:23.200
But but to be honest, UH, you know, platform capabilities

594
00:32:23.240 --> 00:32:27.079
are something that we have to do, you know. And UH,

595
00:32:27.119 --> 00:32:29.279
And then that's the part that really gets me excited

596
00:32:29.279 --> 00:32:32.720
that I'm that I'm super thrilled about is really trying

597
00:32:32.720 --> 00:32:37.079
to drive more of this unique value through more personalization.

598
00:32:37.440 --> 00:32:39.519
And if there were any one place that I would

599
00:32:39.519 --> 00:32:42.480
probably want to put that, you know to me, we

600
00:32:42.519 --> 00:32:45.599
already talked about it a little bit. But the opportunity

601
00:32:45.640 --> 00:32:49.640
to create more flexible, more personalized hardship plans UH with

602
00:32:49.839 --> 00:32:55.160
deep awareness of whether we are increasing the loan value

603
00:32:55.559 --> 00:32:58.440
or not UH is UH is something that I just

604
00:32:58.519 --> 00:32:59.839
get continuously excited about.

605
00:32:59.839 --> 00:33:00.160
It.

606
00:33:00.160 --> 00:33:04.240
It's an immense challenge because the search space is so wide.

607
00:33:04.599 --> 00:33:07.000
You can change the length of the loan, you can

608
00:33:07.119 --> 00:33:08.759
change the interest rate of the loan, you can change

609
00:33:08.759 --> 00:33:10.440
the principle of the loan. You can do it when

610
00:33:10.480 --> 00:33:13.240
someone is in early or mid or late delinquency. You

611
00:33:13.279 --> 00:33:16.240
can offer it via email or text or call. There

612
00:33:16.240 --> 00:33:20.160
are so so many options in when and how to

613
00:33:20.240 --> 00:33:23.240
do this best, you know, but I think of this

614
00:33:23.440 --> 00:33:26.359
as really the most exciting thing on our horizon.

615
00:33:26.839 --> 00:33:29.279
Yeah, I love I love that framing, like really thinking

616
00:33:29.319 --> 00:33:34.200
about the deep personalization of it, and I think, you know, certainly,

617
00:33:34.240 --> 00:33:36.400
I mean Jesse, thanks for joining us, giving so look

618
00:33:36.440 --> 00:33:40.079
and how Upstart is really turning Servicing into a true

619
00:33:40.160 --> 00:33:44.680
performance engine for the company. And I too, am excited

620
00:33:44.720 --> 00:33:47.079
about the idea of us being able to make this

621
00:33:47.119 --> 00:33:50.960
available to something that others can make use of as

622
00:33:51.000 --> 00:33:51.799
we continue to grow.

623
00:33:52.279 --> 00:33:53.680
Thanks Lene, Servicing.

624
00:33:53.799 --> 00:33:55.960
You know, it's not the thing that we dreamed of

625
00:33:55.960 --> 00:33:58.640
when we were little kids, but it's a pretty darn

626
00:33:58.680 --> 00:34:00.119
exciting place to be today.

627
00:34:00.680 --> 00:34:03.920
Thanks again to our listeners for tuning in to another

628
00:34:03.960 --> 00:34:06.920
episode of Leaders and Lending. Don't forget to subscribe and

629
00:34:06.920 --> 00:34:08.920
share the episode and we will see you next time.