In partnership with

Happy Friday, Fintech Takers!

To start us off today, I have an observation and a question.

First, the observation — Marketing, when it’s done well, tends to be both expensive and difficult to measure. As John Wanamaker famously observed, “Half the money I spend on advertising is wasted; the trouble is I don't know which half.” Because of this, when things go badly, marketers tend to take more blame than other groups inside a company. I’ve seen this pattern play out many times over my career, and it strikes me as both unfair and extremely commonplace; like a coach being fired when their team loses in the playoffs.

And now for my question — I have a couple of terrific B2B marketers in my network who are looking for new roles. One is a senior marketing ops and strategy executive who would be very well suited to a VP of Marketing or CMO role. The other is a more junior content marketer, with strong writing skills and a mind for creative thought leadership initiatives. Both are looking to work remotely, but are happy to travel on a regular basis. Do you know of any good opportunities I should send their way?

Thank you for your help!

- Alex

P.S. — Compliance, reliability, developer experience. The companies that scaled embedded payments had all three locked in before it mattered. Now it matters.

Where does that leave the companies still trying to catch up?

Come find out on May 27th. 

Was this email forwarded to you?

Subscribe Here

Sponsored by Method Financial

A 21-point FICO lift within 30 days of closing isn't a coincidence; it's what happens when debt gets retired at funding.

Most lenders know borrowers intend to use proceeds for debt payoff, but they usually can't verify it happened or underwrite against that intent.

Method'sconnectivity layer closes that gap.

With Method, lenders can identify what a borrower owes, then retire that debt directly at funding (without the borrower doing anything).

Figure, the largest non-bank HELOC originator, built a direct payoff tool called Intellidebt on Method's foundation, and the DTI credits it unlocked materially changed the underwriting math.

Figure lent 11% more to direct-pay borrowers, funded them at twice the rate, and saw 6-month delinquency drop by half.

Borrowers got $500/month back and a 21-point FICO lift within the first month.

See for yourself

DEEP DIVE

Three New Fair Lending Stories

Today marks the third time in four weeks that I’ve written about fair lending, which is a frequency that has caught me by surprise.

I’m not an expert on the subject, by any means, but I do have lots of questions about it (here are 10 of them!) More importantly, between the current regulatory environment and everything happening in AI, I think fair lending may, at this moment, be one of the most important topics in fintech.

So, here are three new fintech news stories that are, in different ways, relevant to fair lending. I’ll leave it to you to decide what these stories tell us about the current state of the ecosystem.

#1: PayPal Settles

What happened?

The Department of Justice announced that it has entered into a settlement with PayPal to resolve an investigation into fair lending violations:

Today, the Justice Department announced a settlement with PayPal Inc. to resolve a fair lending investigation into a discriminatory investment program created for black and minority-owned businesses.

So what?

There’s a contradiction in that sentence from the DOJ’s press release, but it’s intentionally subtle. You might not have noticed it. 

Let’s try to hone in on it by pulling some additional quotes from the press release:

The Equal Credit Opportunity Act prohibits creditors from discriminating against credit applicants on the basis of race, color, religion, national origin, sex, marital status, age, because an applicant receives income from a public assistance program, or because an applicant has in good faith exercised any right under the Consumer Credit Protection Act.

Right, yep. ECOA prohibits creditors from discriminating against credit applicants on the basis of a number of protected characteristics, including race, color, religion, national origin, sex, marital status, and age. 

As I wrote about a few weeks ago, ECOA applies to all of us, including the demographic groups that the current administration seems most concerned about protecting from the excesses of DEI. However, and this is the crucial part, it only applies to creditors that are making loans to consumers and small businesses.

So, it’s safe to assume that the PayPal program that the DOJ found to be discriminatory was, at least in part, a lending program, right?

Here’s the DOJ again:

PayPal announced the discriminatory Economic Opportunity Fund in 2020 to invest in black and minority-owned businesses. While the program gave a preference to businesses based on race, color, and national origin, it was not implemented to remediate any specific instances of past discrimination.  

Whoa, wait. It was an investment program? Was there any lending involved in it at all?

Nope.

PayPal's program was made up five pieces: a $500M Economic Opportunity Fund that deployed capital indirectly through minority depository institutions (MDIs), community development financial institutions (CDFIs), and minority-led VC funds; $15M in Empowerment Grants of up to $10,000 for Black-owned small businesses hit by COVID; $5M for nonprofit community partners doing microloans and technical assistance; $15M for internal D&I spend on hiring and employee programs; and $500K in separate civil rights donations to groups like the NAACP Legal Defense Fund and National Urban League.

There were no loans! And yet the DOJ investigated PayPal for fair lending violations! And PayPal “settled” the investigation by agreeing to launch a new small business initiative:

The settlement requires PayPal to launch a new Small Business Initiative that excludes criteria based on race, national origin, or other protected characteristics. As part of the initiative, PayPal will waive processing fees for $1 billion of transactions – a value of approximately $30 million – for eligible American small businesses that are veteran-owned or engaged in farming, manufacturing, or technology.

In addition to fee waivers, as part of the settlement PayPal will designate a director of the Small Business Initiative, conduct an assessment of the needs of American small businesses and determine how PayPal can best support them, submit plans and proposals for the initiative to the United States, provide training to employees on the Equal Credit Opportunity Act, and report on the initiative annually.  

Read that last bit again.

As part of a voluntary settlement that PayPal has entered into with the U.S. government to resolve a fair lending investigation over a program in which no loans were made, PayPal has agreed to “provide training to employees on the Equal Credit Opportunity Act.”

I think I’m actually losing my mind.

I can’t believe PayPal settled.

I mean, I can believe it. We’ve seen lots of other companies settle under similar circumstances over the last 18 months. And in this specific case, PayPal’s economic opportunity fund had already been wound down because all the funds had been invested. And the terms of the “settlement” with the DOJ aren’t punitive at all. And it’s in PayPal’s interest to be on the Trump Administration’s good side because it probably wants some stuff from regulators over the next couple of years (for example, PayPal has applied for an ILC bank charter in Utah and it will need FDIC sign-off).

But still, I wish we lived in a universe in which PayPal just told the DOJ, “Look, the economic opportunity fund was clearly legal at the time that it was launched and active. We think it probably would still be legal today, if it were active, but it’s not. We did nothing wrong. Please go away.”  

But, alas, that’s not the universe we live in. 

#2: Newsom Hires Chopra

What happened?

Rohit Chopra, former Director of the CFPB, got a new job:

Governor Gavin Newsom today announced the appointment of Rohit Chopra to serve as Secretary of California’s new Business and Consumer Services Agency (BCSA), bringing one of the nation’s most prominent consumer protection leaders to California state government.

The new agency, which was established by Governor Newsom through a government reorganization last year will bring together a broad range of licensing, enforcement and other functions that ensure fair competition and treatment for consumers and businesses across a number of sectors of California’s economy. The agency officially launches July 1, 2026.

So what?

Well, for every action, there is an equal and opposite reaction.

The Trump Administration weakens consumer protection regulation and guts the CFPB. Governor Newsom hires the Director of President Biden’s CFPB to run a new agency tasked with protecting California consumers.

A few things are worth pointing out here:

• The Business and Consumer Services Agency (BCSA) isn't a California CFPB. It's a cabinet-level umbrella agency sitting above California's Department of Financial Protection and Innovation, Department of Consumer Affairs, Department of Real Estate, and several other departments (including those in charge of regulating alcohol, cannabis, and horse racing!). This is broader in remit than the CFPB, but without the same standalone rulemaking authority. Chopra's leverage is directional and coordinative: he’ll set priorities across multiple existing departments, but formal rulemaking power remains with each underlying agency.

• California's market size means state-level enforcement actions routinely become de facto national policy. Financial services providers often implement California-compliant practices everywhere rather than maintain separate systems. The closest historical parallel to this Chopra appointment may be Eliot Spitzer's tenure as New York Attorney General in the early 2000s, when state-level enforcement against Wall Street reshaped financial industry compliance nationally during a period of perceived federal laxity. Chopra has more institutional infrastructure than Spitzer did and California's financial services footprint is even larger than New York's was, which means the playbook is even more powerful this time around.

• Governor Newsom is term-limited in 2027 and widely expected to run for president. Thus, both he and Chopra have a narrow window to build a record of success with this new agency. I expect Chopra to swing for the fences, prioritizing high-visibility enforcement actions against major fintech companies, banks, and tech platforms that generate headlines and contrast favorably with federal deregulation. The incentive structure favors marquee cases over incremental supervision, which means California's most prominent financial services providers should expect to be near the top of the target list.

• I expect Chopra to use California law to pursue exactly the fair lending theories the federal government is in the process of abandoning. Disparate impact remains a viable enforcement theory under California statutes even after the CFPB eliminated it. Special purpose credit programs (SPCPs) that the new CFPB rule restricts may find safer harbor under California guidance, creating a state-federal split where the same program is permitted in Sacramento and prosecuted in Washington. And algorithmic underwriting — a long-time Chopra obsession — is very likely to be a focus, since AI models that produce racially disparate outcomes are exactly the kind of high-visibility, technically complex case that generates headlines and reshapes industry best practices nationally.

• As I pointed out on Twitter, this appointment seems perfectly designed to drive Marc Andreessen out of California. I doubt that was Governor Newsom’s goal, but still. It is amusing.  

#3: Affirm Embraces Transformers for Underwriting

What happened?

Affirm held an investor forum in New York, in which it outlined its growth strategy for reaching $100 billion in annual gross merchandise volume. 

A small detail from the presentations leapt out to me:

[Libor Michalek, Affirm's President] said Affirm is now working with transformer-based artificial intelligence models that are outperforming existing models in experiments and are expected to become a new baseline for underwriting performance.

So what?

Transformer-based model architectures (which underpin generative AI models like GPT-5.5 and Claude Opus 4.7) only accounted for about 90 seconds of the 3-hour forum, but it was the most interesting bit to me, by far. 

Specifically, this chart:

What this chart shows is a head-to-head comparison of three vintages of Affirm's new-user underwriting model, each benchmarked against the same March 2023 production baseline. 

The y-axis is approval rate lift; the x-axis is delinquency rate, with better outcomes to the right. Each curve traces a model's full risk-approval frontier — how many more applicants it can approve than the baseline at any given loss tolerance. The blue Mar-25 and orange May-26 lines are gradient-boosted machine learning models (this is the cutting edge in non-transformer AI architecture). The green Apr-26 line is the transformer-based model. It sits above the orange gradient-boost model that is currently in production, meaning an older model (using a transformer-based architecture) is beating a newer one (using a Gradient-boost architecture) on the same data, same problem, same methodology.

That’s a really big deal. 

According to Michalek, the transformer-based model “is discovering more pockets of opportunity more deeply in our data, especially in making dimension a first-class feature, so using our existing data, we can find new information within it.”

Beyond that quote, Affirm hasn’t released any details on how these new models work yet. However, if I had to guess, it seems like what the company is saying is that the transformer model can look at the full sequence of a consumer's financial history — every past purchase, repayment, and balance change, in the order they actually happened — and learn for itself which patterns across that time series predict repayment. The older gradient-boosted models couldn't do that natively. They needed engineers to hand-pick the time windows worth looking at and then flatten everything inside those windows into a single number. The transformer doesn't need pre-chosen windows, which apparently allows it to find predictive credit risk patterns in Affirm’s existing data that the company hadn’t been able to see before.

Michalek went on to say that this transformer-based model architecture will become “the form we're going to be transitioning all of our models to,” which tells us that this is moving beyond experimentation and will soon become the on-the-ground reality for how consumer credit underwriting works at Affirm, and likely other lenders as well.

This makes perfect sense from a credit risk perspective (Look at that chart! They’d be irresponsible not to embrace this!) but if I was a regulator (working with Chopra at the new BSCA, perhaps?) I’d have lots of questions, especially around foundational fair lending concepts like explainability and replicability.

MORE QUESTIONS TO PONDER TOGETHER

Big news for the endlessly curious (yes, you): I’m collecting your fintech questions on a rolling basis. 

What’s keeping you up at night? What great mysteries in financial services beg to be unraveled? Think of it this way, if a stranger is a friend you just haven't met yet, your question is a Fintech Takes conversation waiting to happen. 

One that could headline a Friday newsletter or be answered in an upcoming Fintech Office Hours event.

Drop your question here, whenever inspiration strikes!

WHERE I'LL BE

There are many fun events — virtual and in-person — coming up in the next few months. Here’s where I’ll be!

💻 Fast Rails, Frozen Books: What Banks Are Missing in the Era of Faster Everything (with Rouzbeh Rotabi at Qolo and Kiah Haslett of Fintech Takes Banking)*

Kiah and Rouzbeh are digging into why most banks can't deliver on modern treasury management (even though the rails already exist), and what to do about it. May 20th @ 12 pm ET (virtual); register here.

✈️ Emerge | May 19 - 21

Next week! If you’ll be in Atlanta, let me know!

🏔️ Fintech Frontier Summit | May 31 - June 3 

Talk about a great topic and a great venue. We’ll be talking about bank - fintech partnerships at a ranch in the mountains in northern Montana. It will be glorious.

✈️ Open Banker Salon | June 5

OK, I’m cheating a bit on this one. I won’t be attending this event (too close to my son’s birthday) but I really wish I was. John, Ashwin, Casey, and the rest of the Open Banker team have done a great job pulling together a compelling group of speakers and topics, and the format (a salon!) sounds exactly like what we need more of in D.C.

*This rec is brought to you by one of our fantastic brand partners.

Thanks for the read! Let me know what you thought by replying back to this email.

— Alex