How does Upstart actually work!

[DSG13]

Saul,
Thank you for the explanation. I agree with you that the long term risk of CRB is probably not a major issue for Upstart as long as the loans they are writing are profitable for the loan providers.

As disclosed in TMF Deep Dive that I recommended and posted last week, I would think the concentration risk is probably greater with the relationship with Credit Karma. 52% of the loans that have been approved by Upstart on their platform have been recommended to Upstart through Credit Karma. The agreement between Credit Karma and Upstart can be terminated by either party at any time. Credit Karma has recently been purchased by Intuit. Do you have concerns in this area?

DSG
UPST 5% position

[ZoroSGInvesting]

Saul,

Great summary and I totally agree with your understanding on UPST’s business model and believe their customer concentration issue is way overblown since their end loan buyers (institutional investors and a few banks) are well diversified.

Thank you.

Zoro

[ZoroSGInvesting]

Here is what the CEO was saying in Q4 CC on Credit Karma:

Credit Karma is a really important partner of ours, has been for a while, and we expect will continue to be there now part of Intuit, who we’ve had a relationship with previously. The scale of our partnership continues to grow. we see ourselves as co-leaders in kind of adjacent areas. So as they’ve been successful, we’ve been successful, the relationship continues to become more meaningful to both of us. We don’t anticipate that changing. There’s certainly a lot of different approaches they take and ways we can work with them. It is a great partnership. We anticipate it will continue to be a great partnership.

Our goal, as much as we have talked about having a lot of our loan sort of lead volume coming through Credit Karma, it’s actually important for us to continue to build other channels. We have no desire to reduce the scale of the partnership we have with Credit Karma. In fact, we want to keep building it. We don’t essentially view this as a real issue for us. We are a unique company in that we have a proprietary product. Most people in FinTech or in lending FinTech don’t really have a proprietary product so they seek out proprietary distribution. We actually do have a proprietary product, so we’re very happy to have a partnership with somebody like Credit Karma to take advantage of that.

Sort of like back and forth between us and Credit Karma deciding how we work together is a little bit of, I don’t know, a sideshow compared to what is really just the emergence of a product that is distinct in the market. Our ability to model and help our banks successfully underwrite the torso of the market, the breadth of the market is really what is unique. And Credit Karma is an important partner.

But as Sanjay said, our channel mix has not really changed. And what I think that really means is, increasingly, consumers find their way to the best product. And that’s our belief.
We believe that like the sort of origins of the Internet itself, the foundations of Google and Amazon that consumers find their way to what they want, the best product for them. And I think that’s what you really see happening here is, particularly in this challenging time, Upstart has the best product for most consumers. And whether it’s through Credit Karma or whether it’s through other channels, they’re finding their way to it. So I think that’s really the summary of it.

Based on the comments given by the CEO above, I don’t believe Credit Karma is a major risk as well. And in one of the recent interviews, CEO said that the fastest growing marketing channel is actually UPST their own, not Credit Karma anymore.

Zoro

[SMCappraiser]

Sorry about the last post. Still trying to figure out how formatting works on this board.

Upstarts 10K states:

A significant number of consumers that apply for a loan on Upstart.com learn about and access Upstart.com through the website of a loan aggregator, Credit Karma. The percentage of loan originations that were derived from traffic from Credit Karma was 38%, 38% and 52% in 2018, 2019 and 2020.

It looks like the number of customers that get a loan from Upstart actually originate from Credit Karma. The percentages have been increasing the last few years and is actually over 50% for 2020. I could be misinterpreting this but it looks like Credit Karma may be more of a customer concentration risk that we realize.

SMC

[nilvest]

this is great discussion… thanks for sharing insights.

To me, it is clear that UPST has strong technology benefit and the fact that UPST brings in borrowers to this ecosystem of lenders means that UPST has least of risk in this food chain… he who owns the end customer (in this case borrower) controls its own destiny…

having said that, a high concentration of business with a service provider e.g. CRB or a lead generator e.g. Kredit Karma… is a single point of failure that should give a pause to investors.

A bit of quick research into CRB suggests that they are also one of the most aggressive shops in their domain and work with multiple companies likes of UPST… so for CRB, if they find ways to keep bigger piece of pie working with someone else, they would do so without much hesitation…

All in all, my perception is that risk (of coming across sudden surprise) is higher with UPST business than our “usual” SaaS companies like CRWD, NET, ZS, OKTA… these SaaS companies have wide customer base of enterprise companies with recurring revenue and not really have a single point of failure on supply / service providers side either.

May be the risk here is somewhat higher even compared to non-SaaS but wide user base companies like ROKU, PINS, SQ etc… these companies depend on consumers as does UPST… yet do not have single point of failures vs UPST having two.

Does not mean UPST is bad investment, its actual growth makes it really compelling despite risks… it just becomes a question of position sizing and portfolio management.

[nilvest]

Zoro, I read Kredit Karma commentary on the call as more conciliatory rather than convincing…
It is management job to put things in positive light… and draw attention to what their strength is… does not mean that their dependency on Kredit Karma is reduced.

May be today Intuit or Kredit Karma does not have a good reason to divert traffic to someone other than Upstart… however, if there is someone else coming in and offering them better terms, they have no can change in no time…

so IMO the large and increasing % of origination from this source is a sustained risk

[FoolishJeff]

I am trying to summarize for my own understanding, tell me if I’m wrong. Instead of saying, Customer concentration risk, it is more accurate to say:

1. Supplier Concentration Risk - CRB writes the majority (63%) of loans for Upstart but owns only 21% of the loans themselves. If they stop writing loans, Upstart is in big trouble.

2. Marketing Risk - The majority (52%) of loans that eventually get written come from Credit Karma (Intuit). If things fall apart with Credit Karma, then Upstart is in big trouble.

[CMFFrankDip]

I look at this all a bit differently - Upstart’s value add is its ability to enable higher quality loans using artificial intelligence AI technology - the company is a “pick 'n shovel” play to the lending business.

Although in the near-term Upstart’s stock price may contain headline risk around the front end channel of Credit Karma or the back end lending “middleman”, Cross River Bank (CRB), I do not see either as essential to Upstart’s business model.

As long as Upstart can demonstrably show that its platform will enable higher quality loans with less credit risk for the eventual holder of the loan the company should do well over time. The fact that Upstart takes minimal credit risk is essential to the investing thesis around this company.

Keep in mind that the short-term front end reliance on Credit Karma is about to be dramatically reduced as Upstart enters the market to enable loans for car buyers through its recent acquisition of Prodigy Software. The auto loan market is much bigger than the personal loan market.

As far as CRB it is just a “middleman” - CRB needs Upstart much more than Upstart needs CRB, IMHO.

If Upstart has better loan quality the institutions will find a way to buy Upstart generated loans whether or not CRB is involved.

Upstart may or may not work in the short-term, I have no idea. Long-term if it can create better quality loans with a lower default rate than the competition I feel it should do well.

Frank - long UPST, see profile for all holdings

[TMFCheesehead]

Great discussion going here.

I’m working on an article on this very topic and wanted to throw my 2 cents in.

CRB doesn’t seem like too big of a deal to me. The concentration may come and go. If UPST’s product really does what it says it does (increase acceptance, reduce losses), any bank with half a brain will eventually sign on for those loans. If it doesn’t, then it was never fit to be an investment in the first place.

Credit Karma is – to me – a much bigger risk. But it’s worth stating, a risk doesn’t mean something will come to fruition. Intuit’s new CEO is focused on AI, and I don’t know enough about if Upstart’s AI really is differentiated, or if they were just the first to package it like this (and it’s easily replicable). I also have no idea what the arrangement is between CK and UPST when it comes to fees CK gets paid for referrals.

That said, the risk would be two-fold for UPST. First, they would lose a key source of traffic. More importantly, because INTU has so much data already (Credit Karma, QuickBooks, TurboTax), it could develop a competing product and keep all that traffic for itself.

None of that is to say it would happen. There’s a tension between the vested interest in working together and the risk Intuit would take on by trying to strike out on its own – disrupting what many might see as an already-good thing.

Either way, it’s a relationship any shareholder needs to be aware of.

Best,
Brian

[tamhas]

however, if there is someone else coming in and offering them better terms, they have no can change in no time…

It seems to me that the key issue here is our perception of Upstart’s product.

If their product is fairly easily duplicated so that someone else can deliver equivalent performance at a lower cost, then yes there seems little technical lock in and those large customers could quit. Of course, if that is true, then the issue is less customer concentration as it is the weakness of the moat for the product.

On the other hand, if we think that Upstart has a significant lead over potential competitors such that it is unlikely to be displaced any time soon and we think that the significance of this product is the increase in profitability by minimizing risk, then it seems unlikely that they would be changed out, suggesting that the customer concentration is not necessarily much of an issue.

[Fooledbydesign]

This is a great discussion.
One of the things that concerns me is how do many of us invest in Upstart when its business and risks are not clear, and are not easy to understand…

Among the risks listed in their 10-k, the following seems significant to me. That is, if the regulatory authority, central or any state, decides against upstart to operate, or to limit it in many ways, then it could take significant time to get that corrected and restart their business. Some such sudden news could come out one day and lead to a significant loss of stock price --is my interpretation and concern.

"
Our business is subject to a wide range of laws and regulations, many of which are evolving, and failure or perceived failure to comply with such
laws and regulations could harm our business, financial condition and results of operations.

The legal and regulatory environment surrounding our AI lending platform is relatively new, susceptible to change and may require clarification or
interpretive guidance with respect to existing laws and regulations. The body of laws and regulations applicable to our business are complex and subject to
varying interpretations, in many cases due to the lack of specificity regarding the application of AI and related technologies to the already highly regulated
consumer lending industry. As a result, the application of such laws and regulations in practice may change or develop over time through judicial decisions
or as new guidance or interpretations are provided by regulatory and governing bodies, such as federal, state and local administrative agencies.

Since we launched our AI lending platform, we have been proactively working with the federal government and regulatory bodies to ensure that
our AI lending platform and other services are in compliance with applicable laws and regulations. For example, after significant collaboration with the
Consumer Financial Protection Bureau, or CFPB, the CFPB issued Upstart the first no-action letter in 2017 and, upon its expiration, we received a second
no-action letter regarding the use of our Al model to underwrite and price unsecured closed-end loans. The current no-action letter expires on November
30, 2023, unless terminated by the CFPB earlier for one of the bases provided for by the no-action letter, and we can provide no assurance that the CFPB
will continue to permit us to operate under its current no-action letter policies beyond that date, or will not change its position regarding supervisory or enforcement action against us in the future. Further, this no-action letter does not extend to other credit products offered on Upstart’s platform. We plan to
continue working and collaborating closely with regulators to provide visibility into AI and related emerging technologies and the potential benefits such
technologies can have on the consumer lending industry, while also addressing the related risks. New laws and regulations and changes to existing laws and
regulations continue to be adopted, implemented and interpreted in response to our industry and the emergence of AI and related technologies. As we
expand our business into new markets, introduce new loan products on our platform and continue to improve and evolve our AI models, regulatory bodies
or courts may claim that we are subject to additional requirements. Such regulatory bodies could reject our applications for licenses or deny renewals, delay
or impede our ability to operate, charge us fees or levy fines or penalties, or otherwise disrupt our ability to operate our AI lending platform, any of which
could adversely affect our business, financial condition and results of operations.

Recent financial, political and other events may increase the level of regulatory scrutiny on financial technology companies. Regulatory bodies
may enact new laws or promulgate new regulations or view matters or interpret laws and regulations differently than they have in the past, or commence
investigations or inquiries into our business practices. For example, in February 2020, we received a letter from five members of the U.S. Senate asking
questions in connection with claims of discriminatory lending made by an advocacy group. We responded to this inquiry, and in July 2020, three of the
Senators issued their findings from this inquiry, writing a letter to the Director of the CFPB recommending the CFPB further review Upstart’s use of
educational variables in its model and requesting that the CFPB stop issuing no-action letters related to the Equal Credit Opportunity Act, or ECOA. We
have been subject to other governmental inquiries on this topic including an inquiry in June 2020 from the North Carolina Department of Justice. See the
risk factor titled “—We have been in the past and may in the future be subject to federal and state regulatory inquiries regarding our business” for more
information. Any such investigations or inquiries, whether or not accurate or warranted, or whether concerning us or one of our competitors, could
negatively affect our brand and reputation and the overall market acceptance of and trust in our AI lending platform. Any of the foregoing could harm our
business, financial condition and results of operations.

".

[Fooledbydesign]

Couple of points to highlight in the regulatory risk information above.

“Such regulatory bodies could reject our applications for licenses or deny renewals, delay
or impede our ability to operate, charge us fees or levy fines or penalties, or otherwise disrupt our ability to operate our AI lending platform”.

“in July 2020, three of the Senators issued their findings from this inquiry, writing a letter to the Director of the CFPB recommending the CFPB further review Upstart’s use of
educational variables in its model and requesting that the CFPB stop issuing no-action letters related to the Equal Credit Opportunity Act, or ECOA”
.

[Fooledbydesign]

“On the other hand, if we think that Upstart has a significant lead over potential competitors such that it is unlikely to be displaced any time soon and we think that the significance of this product is the increase in profitability by minimizing risk, then it seems unlikely that they would be changed out, suggesting that the customer concentration is not necessarily much of an issue.”

Two main advantages that Upstart quotes about themselves-

1- AI with lots of parameters in their model. Source of data is various, and public information too.
2- Make it very easy to apply loan - loan application screen/workflow for borrowers is very easy. Focused on benefiting borrowers by offering best possible loan rate, without using FICO score, for those whose loan application would be rejected in traditional routes.

[MarinoT]

I came across this excellent podcast from ‘Leaders in Lending’. Paul Gu, co-founder and Head of Product from UPST is on. He really addresses how they are different to the market, different to the FICO score and explain the different AI systems they leverage.

It doesn’t cover the commercials, just focuses on the tech.

Link is here - https://podcasts.apple.com/au/podcast/the-opportunity-with-a…

[KP1981]

Here is some information from the 10K report

Business model - 98% of the revenue comes from fees. What are those fees?

Here are the fees for the years, 2018 2019 2020 in millions
Referral fees, net - $ 53.869; $ 90.672 (68% increase); $ 133.425 (47% increase)
Platform fees, net - 29.512; 53.383 (81% increase); 66.832 (25% increase)
Servicing fees, net - 5.101; 15.792 (210% increase); 28.343 (79% increase)

Contribution Profit - $ 13.098; $ 48.940 (273% increase); $ 105.088 (115% increase)
Contribution Margin - 15%; 31%; 46 %.

Number of loans - 114,125; 214,556 (calculated from info that says 2020 had 40% increase); 300,379

What does each fee mean?

1. Referral fee - this is the fee that Upstart gets from the bank when someone gets loan from that bank, but the process started in Upstart.com or through the Credit Karma website or any one of the other sites where people go to find personal loans.
2. Platform fee - Upstart has an agreement with banks where the banks use the Upstart algorithm and pay a fee for each loan that is approved.
3. Servicing fee - this is a standard fee for servicing the loan after it has been issued. Basically a feel to collect the monthly payment and pass it on to the investors. - According the the 10K report, for most loans, banks choose to let Upstart service the loan, regardless of where it originates.

Other interesting observations:

1. Fee received from Cross River Bank (CRB)- it is 63% in 2020 and has been pointed out by some as something to be concerned about. However, it was 80% in 2019, and 81% in 2018. - declining trend while the business in increasing, so the revenue base is growing.
2. Platform fee increase in 2019 - agreement with CRB went into effect on Jan 1, 2019. They are in the process of expanding this service to other banks, they already have multiple agreements.
3. Growth from Servicing Fee - their typical loan is for 36 months or 60 months. In 2019 they had issued about 114K loans for which they received a fee of about 5 million. in 2019, they issued about 214.5K loans, so they are servicing about 328.5K (114+214.5)loans, which is nearly a 3X increase, and is reflected in the increase in fee. In 2020, they issue about 300K loans, so they would have serviced 628.5K (328.5 + 300) loans, which is a bit under 100% growth, and the service fee has grown by about 79%. - this discrepancy is likely because of the time of the year when the loan was issued. - However, the trend is clear - the servicing fee does have an exponential growth.
4. Loans retained by the banks - In 2020, 21% of the loans originated by the banks were retained by the banks. Though they have not given specifics, they have stated that the general trend is increasing % of loans being held by banks for their own investors.
5. Credit Karma & Referral Fee - Between 2019 and 2020, the revenue from Credit Karma has increased, but it started reducing in Nov 2020 because Credit Karma starting showing other lenders on the site. While the 10K report that was released in March says that they chose to not participate in that, today when I checked Credit Karma for personal loan, Upstart is an option for a loan. So something must have changed.
6. Intuit & Credit Karma & Upstart - In Dec 2020, Intuit acquired Credit Karma. On Mint, Upstart is one of the 4 options for personal loan.
7. Auto loan & Student Loans & Platform Fee - In June 2020, Upstart started offering auto loan evaluation for banks, but the first auto loan was issued in September 2020. They have also started processing Student Loans - but no details on that yet.
8. Other products - They have indicated that their product can be expanded to include credit cards, mortgages, HELOC, student loan and point of sale loans.

Important points from the 1st quarter report and how that compares to the 10K report

1. In 1Q of 2021, they had processed nearly 170K loans compared to 300K in all of 2020
2. Contribution margin for 1Q is 48% compared to 42% for the whole year of 2020.
3. The other bank - there is mention of “other bank” which accounts for 25% of the revenue.

So what does this all mean?
Note that this is just my personal opinion/interpretation - if you think is is off base, let me know, I want to hear other perspectives.

Upstart is not a loan company, but rather a company that helps banks evaluate risk of each loan. They get a fee for each evaluation. Think about this, if you apply for a loan on Upstart.com (either directly or through Credit Karma), they make money in the form of referral fee. If you apply on a bank website, they make money in the form of platform fee. In both cases, they make money in the form of servicing fee. Basically their business model is, “heads I win, tails you lose”.

[GauchoRico]

Upstart originates most of the loans THEMSELVES on the Upstart platform using a loan platform agreement (about 63% of the loans last year) – but CRB actually ISSUES the loans for them, because CRB is a bank, and Upstart isn’t. So CRB isn’t a customer, it’s a service provider. It provides a service for Upstart.

Somebody at then end of the line has to have confidence that UPST’s AI underwriting isn’t going to lead to much more than expected defaults. The more banks that are buying in and actually using UPST’s AI model, the lower the customer concentration risk. Having 2 banks take on 85% of the business isn’t as good as many more banks taking the risk on the AI model. This is true even if CRB then sells the loans to institutional investors who are obviously looking for high yield fix income. Do we know who’s at the end of the line taking the risk? I’m not sure we know which institutions are ultimately buying the loans from CRB. Or if the institutional investors are reselling to their clients (probably). The risk profiles of the institutions and their clients can vary widely and some investors may be willing to take a lot of risk right now since high yield fixed income is hard to come by. Risk appetite can change quickly. Investors can take on “too much” risk. Perhaps I still have flashbacks to the financial (a.k.a. subprime) crisis. Or to First Marblehead in the student loan business where risk wasn’t quite so known. Yes, for me, the customer concentration risk is real for reasons more than just losing the one big customer. It’s also a risk because a lack of breadth in customer adoption says something about the current aggregate banks’ confidence in AI-based underwriting in general and in UPST AI model specifically. I want to see UPST’s model get more widely adopted.

The above is the main reason why I don’t want my UPST position to be too big (currently it’s 6.6% of my portfolio but it grew into that from a ~3.5% investment). UPST, probably more than any of my stocks, is one that I could see going up 300% or down 50% in a year. I do like what they’re doing and their progress, but with this one I prefer to let my position grow on its own rather than adding to it at higher and higher prices when I don’t see that the main risks have been mitigated.

GR

[nilvest]

just a thought… if Upstart shared some metrics around performance of their value proposition - i.e. % or portion of loans (originated by Upstarts… irrespective of who is holding at the end) in default or delayed vs industry… that would be most interesting metric to see.

Since Upstart itself is not holding the loan, one can not find out these from reserve ratio or similar metrics used by banks holding the loan.

[ibuildthings]

if Upstart shared some metrics around performance of their value proposition - i.e. % or portion of loans (originated by Upstarts… irrespective of who is holding at the end) in default or delayed vs industry.

Since Upstart itself is not holding the loan, one can not find out these from reserve ratio or similar metrics used by banks holding the loan

I wondered that myself. How more effective are their algorithms over traditional FICO or other metrics? They should be able to track the results because they would use them to re-train their neural nets. It is part of how the technology works. There is no way they would intentionally lose track of the results. I understand privacy rules and all, but their bank clients could anonymize the data. Even if the bank resold the loan in many cases, because banks often make agreements with the buyers of the packaged loans to continue servicing the loan. The servicer gets to keep the late fees and extra interest charges.

[KP1981]

https://www.upstart.com/for-banks/

In the link above, they are claiming 75% lower default rate (take it with more than a pinch of salt as it is based on their own simulation).
However, there is a statement by the COO of Customers Bank stating that Upstart was able to increase returns and lower default rates, and the have moved from a small pilot scale to full scale lending program.

[RafesUserName]

While the following has been touched on here, and while much of the discussion and points in this thread are important to form a foundation on which my investment can stand, there is a specific combination of points that convinced me to purchase shares:

Upstart gets paid for the referral, not just when a loan is granted. This means their product is valuable enough to their customers that they are willing to pay just for leads! This is great confirmation of the value of their AI/proprietary product

In other words, Upstart are growing revenue because the CUSTOMERS KNOW the quality of Upstart referrals are higher than other sources, and so WE KNOW their tech WORKS.

I feel a point I’ve learned from Saul is: if the customers love the product do we really have to know how it works under the hood? The customers know the value because they are experts. They pay for it in increasingly larger amounts and that is all I really need to know about the quality of the product.

Now it IS possible I am putting too much weight in this aspect of how things work. I don’t know the lending industry well enough to know for sure, but I do like the dynamic. This is also why I hold around 8% for now while I learn more (this thread is just so great!).

Another quick point. If they are successful in the auto loan segment, which is larger than the under $50k personal loan segment they operate in now, then I assume the concentration in all areas will be spread out.