October 29, 2025 • 38 mins
About 45 credit unions are customers of JUDI.AI and they are drawn by the company’s simple promise: Eliminate friction in business lending.
There is money to be made serving small businesses and JUDI.AI has the formula, says Gord Baizley, the company’s CEO.
A key is automating a lot of the process and that’s JUDI.AI’s business.
Here’s why the company exists: “Small businesses are the backbone of our economy, but 40-50% of small businesses rank access to capital as their top challenge. At JUDI, we’re transforming small business dreams into vibrant communities by increasing access to capital.”
That’s a big promise but, says Baizley, the company has had very little churn among credit unions who sign up. That’s because they believe they are getting value and also that in fact they are finding ways to serve their communities’ small businesses.
Just about every credit union, certainly all bigger than $500 million in assets, dreams about making small business lending a bigger slice of their portfolio.
Baizley insists that dream can be reality. On the show he tells how.
Listen up.
Like what you are hearing? Find out how you can help sponsor this podcast here. Very affordable sponsorship packages are available. Email rjmcgarvey@gmail.com
And like this podcast on whatever service you use to stream it. That matters.
Find out more about CU2.0 and the digital transformation of credit unions here. It's a journey every credit union needs to take. Pronto
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
SPEAKER_01 (00:00):
Welcome to the CU2.0 podcast.
SPEAKER_00 (00:05):
Hi, and welcome to the CU2.0 podcast with big new
ideas about credit unions andconversations about innovative
technology with credit union andfintech leaders.
This podcast is brought to youby Quillo, the real-time loan
syndication network for creditunions, and by your host,
long-time credit union andfinancial technology journalist
(00:27):
Robert McGarvey.
And now the CU 2.0 podcast withRobert McGarvey.
SPEAKER_01 (00:35):
About 45 credit unions are customers of Judy.ai
and they're drawn by thecompany's simple promise.
Eliminate friction in businesslending.
There's money to be madestirring small businesses in.
(01:03):
Quote, small businesses are thebackbone of our economy.
But 40 to 50 percent of smallbusinesses rank access to
capital as their chiefchallenge.
At Judy, we're transformingsmall business dreams into
vibrant communities byincreasing access to capital.
That's a big promise.
(01:23):
But since basically the companyhas had very, very little turn
among credit unions who signedup.
That's because they believethey're getting value.
And also, in fact, they'refinding ways to serve the
community's small business.
Just about every credit unit,certainly all bigger than$500
million in assets, dreams aboutmaking small business, lending
(01:45):
them bigger slides for theirportfolio.
Credit unions are finding thatthey can do it.
And some of those credit unionsas are as small as maybe$25
million in assets.
(02:09):
And historically, as I'm sureyou know, small business lending
has been something credit unionshave wanted to do, but they
haven't done it for manyreasons.
One of which is that it's notusually profitable for them
because the amounts in theirlending are too small to justify
the labor that goes into bettingthe app loan application to
(02:34):
simplify matters.
I mean, and that's true with bigbanks too.
If I walk in the chase and say,hey, I'd like you to give my
small business a$50,000 loan, mybanker there would say, um,
we'll give you a personal loan,but not your business.
SPEAKER_02 (02:50):
Yeah.
And I mean, it you're right,it's not just credit unions,
it's the broader market.
And if you look at the you know,sentiment data out of any survey
of small businesses, they'll saythey're underserved from a
banking perspective and a creditperspective in particular.
It's just a hard segment toserve.
And so what's happened is youknow, fintech lenders have
stepped in to fill that void.
The problem is they're notdeposit-taking institutions, so
(03:13):
their cost of capital is prettyhigh, and so they have to pass
that along.
And next thing you know, youryou know, your APRs are 30, 40
percent.
SPEAKER_01 (03:22):
Yeah, I talked with a guy uh a couple of years ago
who does just small businesslending, and he told me
traditional financialinstitutions we're kind of dead
at the starting block becausethey they ask for the wrong
data.
Um what's your Dun andBradstreet writing?
My what?
And furthermore, yes, the DunhamBradstreet doesn't know who this
(03:44):
company is.
Two, what's what's your FICOscore?
Well, some of them are havingmoney problems, and their FICO
score is really taking a tumble.
So he said, all I look at iscash flow.
Right, exactly.
And he I assume his interestrate, he wasn't specific, but it
was over 20, significantly over20 percent.
SPEAKER_02 (04:02):
Yeah, and that's that's the the fintech lenders
sort of figured this out uh uh awhile back.
And we actually we were born outof an in uh a fintech lender
ourselves and then developed aplatform and spun out a separate
business for for that business.
But exactly what we learnedalong the way is that cash flow
was really the the the bestpredictor of of what's going on
(04:23):
and and actually most recentcash flow is really important.
So um, you know, the we look at12 months of cash flow data, and
I think that's all this guywanted was 12 months.
SPEAKER_01 (04:33):
If he had 24, he might look at it.
But yeah, but his main concernwas if I lend you this money,
can you make the monthlypayments?
And the cash flow told him yesor no.
SPEAKER_02 (04:44):
Well, and there are some fintech lenders that will
lend on the back of only threemonths of cash flow data.
Whoa, yeah, and and we've evenseen it in our data that you
know, months one to three aremuch more predictive than months
nine to twelve from a from await a minute though.
SPEAKER_01 (04:58):
If I open the right kind of store uh and I give you
cash flow data for October,November, December, I might have
zero cash flow in Januarybecause I'm selling Christmas
gifts, you know.
SPEAKER_02 (05:11):
It's uh so yeah, and and and and I'm speaking in
general out generalities acrossthe small business population.
Obviously, there's seasonalityin some businesses, but but for
the most part, uh recent data ismore predictive than old data.
SPEAKER_01 (05:25):
Um sure.
SPEAKER_02 (05:27):
That's uh and it's pretty acute actually um across
the general population of smallbusinesses.
SPEAKER_01 (05:33):
I mean, that's that's essentially why I really
don't like FICO scores as ameasure of much of anything
except a pretty good picture ofmy past behavior as a credit
card user.
But I don't think it tells you athing about my future behavior.
SPEAKER_02 (05:46):
Yeah, I mean, I think you know, a certain uh a
certain baseline uh uh of a FICOscore will tell you what kind of
a person you're dealing with,and are they likely to you know
completely screw up or or evenworse, you know, defraud you so
you get a sort of sort of abaseline character perspective
out of that, I guess.
Um and then and then from therethough, you really want to know
(06:07):
how their business is doing inin real time.
SPEAKER_01 (06:10):
Now, how many credit union customers do you have?
And how many US-based creditunions do you have?
SPEAKER_02 (06:18):
Yeah, we're working with uh I think we're creeping
up towards 45 customers.
SPEAKER_01 (06:22):
Uh that that's outstanding.
How how long has it taken you toget 45?
SPEAKER_02 (06:27):
Yeah, we launched, so our first customer um we was
Van City Credit Union, so thelargest credit union in Canada.
Um, we launched with them in uhMay of 2017.
So that was that was when thefirst application went through
our first credit union customer.
Um, and so that's uh you knoweight years ago now um that
we've been doing this and uhadding them steadily.
(06:50):
We we sort of got into the USmarket a couple of years ago.
That's been going well.
We're I think up to 18 customersin the US and uh more coming on,
obviously, lots of potentialthere.
But um, we really, to yourpoint, we've the credit unions
in both Canada and US, andthey're very similar, have
historically not done a lot ofsmall business lending for all
of the usual reasons.
It's expensive and unprofitable.
(07:11):
And um uh so they've migrated,you know, they've always focused
on CRE to the extent they'vedone commercial lending, but
they are we we're reallynoticing a pattern now where
they want to get into this.
They know they should servetheir members in this segment,
huge important segment of theeconomy.
And so they're they're startingto run at it, uh, which has been
fun for us.
SPEAKER_01 (07:28):
What's the size range approximately of your US
credit units?
SPEAKER_02 (07:34):
Yeah, so we've got so we work with um uh uh through
through a couple of CUSOs, weworked with some very small
credit unions, uh, as low as 25million in assets.
Um, and then uh like I say, ourlargest client is 30 billion in
assets, and in there's a we havea couple of those in Canada and
and pretty much everywhere inbetween.
(07:55):
Um uh we do sort of notice thatsort of 500 million in assets
and up is where uh if you're notgoing through a CUSO, where it
where it kind of works, wherethey can actually um absorb the
cost and execute uh on a programthat that makes sense.
So that's a kind of a thresholdthat we focus on.
But um if smaller ones come tous, we certainly will work with
(08:15):
them and we'll uh sometimessuggest that we pull in a CUSO
to work with them and help them.
SPEAKER_01 (08:20):
Now, remind me, my my memory tells me that you know
the US has approximately 6,000credit units, Canada has
approximately six.
I mean, I exaggerate, but yeah,Canada has really big credit
units, right?
SPEAKER_02 (08:35):
Yeah, we're so yeah, we're um about 250 credit unions
in Canada, you know, and and soum 10% of the population.
So if you were to sort of do alinear extrapolation, we should
have um you there should be2,500 in the US to be
equivalent, and there's at leastdouble that or something like
double that.
So yeah, you're right, we'reless fragmented and generally
(08:58):
have larger credit unions umhere uh and fewer targets for
us.
But uh so it's a bit of adifferent dynamic.
SPEAKER_01 (09:05):
How how does the QZO come into the picture when
you're serving a credit unionthrough a QZO?
SPEAKER_02 (09:11):
Yeah, so some of the credit unions, uh because as you
say, they haven't historicallydone a lot of commercial lending
and certainly smaller C and Ilending, some of them maybe
don't have the expertise to dothe underwriting or the
servicing or the reviews.
And so um they can outsourcesome or all of that process to
uh uh to a CUSO.
(09:31):
And so where we can partner witha CUSO and bring both technology
and the expertise and servicesto the table, it enables someone
to get up and running veryquickly without building out a
whole department.
SPEAKER_01 (09:41):
So take tell me, what do you bring to the table
and what does the credit unionhave to bring to the table,
either on its own or through aCUSO?
SPEAKER_02 (09:52):
Yeah, so we bring the technology and a little bit
of advice around how toconfigure the technology for um
a program based on whatever therisk and business objectives are
of the of the credit union.
Um and we help them through thatjourney.
But but what we don't do is wedon't do underwriting, um, we
don't write credit policies, uh,we don't do annual reviews, um,
(10:14):
but we can provide thetechnology to enable all of
that.
So uh they either need um peopleinternally that are capable of
that and that have the capacity.
That's another thing.
Sometimes it's it's capability,sometimes it's capacity.
Um, and if they don't have itinternally, then they then they
need a CUSO, and there areseveral around that provide
those services.
SPEAKER_01 (10:33):
So your technology automates the lending process.
Yeah, yeah.
So it does a few things.
And so in in the in the pandemicera, and there was a lot of
government money to lend in uhthe US, government backed loans.
Uh the big banks jumped on that.
Like Wells Fargo did a lot ofthose loans.
(10:55):
When I investigated, it's it'severything was automated.
I don't know what the dollaramount was, was their maximum,
but say it was 100 grand.
Anything under that was gonna beautomated.
No human being would touch theapplication.
Credit unions struggled withthat, they didn't have the
technology, I don't think.
SPEAKER_02 (11:15):
Yeah, and it's interesting, like and in Canada
had a similar program too, sothe dynamics were very similar.
And um, you needed to automateit.
There was so much volume coming,but but it was really workflow,
not underwriting automation, youknow, because there wasn't
really underwriting, like theywere right.
SPEAKER_01 (11:31):
There was essentially no, but again, it's
how much labor can you afford toput into this twenty thousand
dollar loan?
Yeah, very little, basically.
SPEAKER_02 (11:41):
Yeah, no, we've had uh had uh people say to us like
once we've once we've touchedthe file twice, we've lost
money.
SPEAKER_01 (11:47):
Yeah.
So uh absolutely it's theythey'd much rather do a car loan
for thirty thousand dollars.
Yeah.
SPEAKER_02 (11:53):
So the the flip side is that um, although they're
small amounts, there tends to bepretty good margin in the
segment um uh for a number ofreasons.
And so if you can figure out howto control the cost of delivery
and do it at scale, it canactually, like some of our
clients, this is their mostprofitable segment.
So um once you get to thatpoint, it it's it's very
worthwhile.
SPEAKER_01 (12:15):
Now, you don't, I'm assuming you don't provide
marketing advice to the creditunion.
In other words, here's how yougo after small business lots, A,
B, C, and D.
SPEAKER_02 (12:28):
Yeah, we actually do a few things on that front.
Um, so so um the the big thingthat we've noticed, um I always
say, you know, if you build it,they won't come.
Like you can't just stand up aplatform and expect that you're
gonna get applications all of asudden.
So you do have to activate it.
And and one of the mostsuccessful ways to activate it
is to actually dopre-qualifications of your
members.
(12:48):
So most credit unions aresitting on a few thousand uh
business members um and andthey're sitting on all their
account data, that cash flowdata that is very predictive.
And we can go through that andgive them a picture of who they
could make offers to.
And the the conversion rates andthe uptake from small businesses
has been incredible on thosetypes of exercises, I think, for
two reasons.
One, um, there's a high degreeof trust between the business
(13:12):
members and the credit union,and two, business members,
unlike consumers, aren't used togetting these kinds of offers.
And so we've had um we had oneclient do do one of these Libro
credit union in in Ontario andCanada, and they had a 30%
conversion rate on offers madeout of one campaign, uh, just to
give you a feel for howsuccessful that was.
But again, you do have to beproactive.
(13:33):
Pre-quall is one way to do it.
You can do digital emailcampaigns to your members, and
of course, you can get out anddo you know more traditional
digital marketing beyond yourmembership base.
What we've seen with the creditunions is they have so much
low-hanging fruit in theirbusiness member base that they
haven't had to go outside yet.
They can focus on that.
The other area that's alsointeresting is um, and we've got
(13:55):
a partner, uh, Fingle out ofColorado, that can do some
analysis on this, uh, but butthey can comb through uh a
credit union's consumer accountsand predict which ones are uh in
all likelihood running abusiness out of those accounts.
And it's usually somewherearound 10% of your consumer
members are are running abusiness out of the account.
So um there's an opportunity tosort of migrate those over to
(14:17):
business services.
SPEAKER_01 (14:18):
That's that's interesting.
You know, credit unions in theUS is for some years now have
technology that lets them, andit happened to me with my credit
union.
I didn't have a credit card.
They sent me an email or aletter saying, hey, you know,
what we we we really want togive you this credit, this
credit card, like a$20,000limit.
(14:40):
There's no fee.
Do you want it?
I said, huh?
Yeah, I guess so.
But but they're they're cutthey're accustomed to doing
that.
So this is just doing that samething with with small business
members.
SPEAKER_02 (14:55):
It is, and and and it's easy on the consumer side
because if you have a bureauscore, you're comfortable doing
it on the back of that,typically.
Um, but it with a smallbusiness, that doesn't quite cut
it per your previous points thatyou were making around that.
So you have to look into thecash flow and uh and be able to
make sense of the cash flow inthe in the operating account to
decide whether that's a an offeryou can make.
SPEAKER_01 (15:17):
I I don't think my credit union even had a Euro
score when they offered me that.
It's uh I think they just theyjust looked at um my accounts to
balance in them, et cetera.
And so by heavens, he's paying$4,000 a month to American
Express.
Why don't we get a piece ofthis?
SPEAKER_02 (15:36):
Yeah, there could be some of that.
I know a lot of them um uh forcompliance reasons at least do
soft polls on their on theirconsumer members.
So a lot of them will have thatthat on file.
SPEAKER_01 (15:46):
And I could be a soft poll that I never would
have seen.
That's that's true.
That's uh and how long does ittake a credit union?
A credit union calls you up whenthey hear this podcast and they
say, we want to get started onthis.
Now that won't happen.
They're probably gonna ask youto convince them to get started
on it.
But we'll we'll just imaginethat.
It's uh how long does theprocess take?
SPEAKER_02 (16:10):
Yeah, so we usually guide people to sort of allow
for eight to 12 weeks to getlive.
We've done them in as little asfour weeks.
They've also gone longer,depending on some uh uh some of
the choices and dynamics in theaccount.
But um, if we keep it simple andwe really want to just get up
with uh uh a basicimplementation, it can be pretty
quick.
And if in fact, if someone cameto us and said we need to be
(16:32):
live in a week, we we could, youknow, in theory, uh get
everybody uh aligned veryquickly and be live that quick.
So it it is um it it doesn'thave to be a massive project.
You know, I think people thinkabout core conversions or
massive LOS implementations andthey get scared off, but this
this problem can be actuallysolved quite quickly.
SPEAKER_01 (16:53):
Do you need core system access?
SPEAKER_02 (16:56):
You don't need it out of the gate.
No, so um the way we can uhaccess the cash flow data uh is
through the account aggregatorsin the market.
So there's a bunch of them outthere that you know, PLAD, et
cetera.
We use Yodly um uh as ourprimary access point.
And the reason we do that is umA, you don't need to integrate
out of the out of the gate,which uh expedites the process,
(17:16):
but B, you can also accessaccounts at you know other
institutions.
And about 20% of theapplications we receive have
accounts, you know, multipleaccounts at multiple
institutions.
SPEAKER_01 (17:26):
And increasingly more and more of us are familiar
with Plat at least.
Yeah.
SPEAKER_02 (17:33):
Yeah, it's a it's a journey, and I always always say
it's you know, open banking andquotations, we're not there yet,
but there are increase anincreasing number of open
banking-like, you know, APIconnections into these
institutions that are, you know,uh one day this this technology
will be ubiquitous and hopefullyon a standard protocol.
SPEAKER_01 (17:53):
And what's your fee structure?
SPEAKER_02 (17:56):
Yeah, so we charge um uh a subscription fee for the
platform that that varies bysize of credit union, and then
we charge a fee per completedloan application.
Um, so anything that we scorebasically, we we charge for.
And in that in that fee, wewould absorb the cost of the
Bureau scores and the dataconnections.
And so it's fully loaded, theDocuSign envelope for the loan
(18:17):
agreements.
SPEAKER_01 (18:18):
Um what's what's a range for the the monthly fee?
SPEAKER_02 (18:25):
Yeah, the the range is uh as low as 4,000 a month
and as high as 30,000 a monthfor the the massive credit
unions and uh everywhere inbetween, and then the
application fees$140 for the thecompleted and what's what
staffing specifically does thecredit union need on its end to
(18:46):
make this work?
Yeah, we sort of we we recommendlike three primary stakeholders
in a successful implementation.
And the first one is some sortof an executive sponsor that's
really going to make sure thatthere's you know strategic
prioritization at the top of theorganization.
And if you have that, thatthat's a key marker of success.
Um, the second one would be um aforward-thinking subject matter
(19:09):
expert around lending andcommercial lending that maybe
wants to make a change.
So somebody that's comfortablewith technology and changing
process, and that can thatperson would be the driver.
And then a project manager justto get it going and they don't
have to stay with it, you know,much beyond the implementation,
but having somebody justcoordinate with you know
marketing and legal and um andcredit risk, et cetera, to make
(19:32):
sure that it gets off theground.
And it's not a ton of work uhwith all of those stakeholders,
but they do need to be informedat least.
SPEAKER_01 (19:39):
Have you had customers who are doing this and
then decide to quit?
And if so, why?
SPEAKER_02 (19:47):
Yeah, good question.
We've had um to the extent we'vehad churn, um, we had a couple
that used it for uh the SIBAprogram in Canada, um, which was
like the PP program and the PPPprogram in the US, and then they
didn't continue on.
Uh, they tended to be thesmaller credit unions that
didn't have the resources to doit.
So that that's really the thebiggest blocker that we've seen
(20:09):
is if you just don't have thestrategic prioritization and the
resources to to move forward.
Um, and so that's where we'vesort of noticed there's a line
there at around 500 million inassets, where below that you may
need the help of a CUSO to pushforward.
SPEAKER_01 (20:25):
I've often asked credit union CEOs, why don't you
have blah, blah, blah?
And the answer often in manycases is I'd love to, but I
don't have the staff.
And these these are smallercredit units, these are not
credit units, and certainly notones over a billion dollars in
assets.
Although I'll tell you, in thelast couple of months, twice
(20:48):
now, CEOs of billion-dollarcredit units have referred to
themselves as running a smallcredit unit.
Twice.
Two different guys.
So billions now small.
Okay.
SPEAKER_02 (21:02):
Yeah, the landscape's changing for sure.
And um, yeah, those thoseresources are are critical.
And if you don't have them, itwon't move forward.
And one of the challenges thatwe see quite frequently is, you
know, like you say, the legacy,sort of the heritage of of
credit unions is on the consumerside typically.
So so these the commercial sideof the house tends to have it
struggle a little more to getthe resources for these kinds of
(21:24):
projects to to grow that side.
So a bit of a chicken and egg.
SPEAKER_01 (21:28):
That said, um many credit unions recognize that
basically they're forcing theircustomer out, their member out
the door.
Yeah, if I start getting mybusiness loans from Wells Fargo
uh or a fintech, what what makesyou think I'm gonna stay around
as a credit union member forvery long?
(21:48):
And what credit unions don'tlike to analyze a heck of a lot
is share of wallet.
And a generation ago, they oftenhad a hundred percent share of
wallet from many of theirmembers.
That's uncommon today.
Um my credit union has maybe 15%of my wallet, maybe it's uh and
(22:11):
uh and I'm perfectly happy withit.
SPEAKER_02 (22:14):
Well, and a couple of things to that point.
One is um if you look at the ifyou deconstruct the data, about
25% of U.S.
households have a small businessattached to them.
Um, so it's important to be ableto serve those small businesses
because a big chunk of yourpopulation is going to need that
service.
Uh, secondly, and and you know,this has come to light with some
of our more successful clientslike Van City, that's now been
(22:37):
doing this for eight years andreally leaned into it.
You know, they're now doing,they've done in they've had
months where they've done 500applications through our
platform.
Uh, but they treat it as a verystrategic segment.
They view it because of thatsort of high incidence rate in
the population.
Uh, but also the the the I guessthe the sort of emotional
component to a small business,like if if if you fund someone's
(22:58):
small business dream, chancesare they're sticking with you
and they're probably bringingeverything else with them.
Um so there's a really strategiccomponent to the relationship
out of that segment.
SPEAKER_01 (23:08):
Right.
I mean, if you do a smallbusiness loan, almost certainly
the business will ask for asmall business credit card too,
eventually, if you have that tooffer.
Um I mean it's now will yoursystem accommodate gig workers,
like a full-time Uber driver,that sort of thing.
SPEAKER_02 (23:30):
The yeah, the where where our system works is it if
if the business has a businessoperating account, if it's a if
it's a you know a side businessrun out of a consumer account,
um, we we sort of think of thatas a consumer loan.
So it's it really needs to havea business operating account.
So if that gig worker setsthemselves up as a business, we
could do it.
But if it's sort of a sidething, then it you'd probably
(23:51):
treat it as a consumer loan.
Does Canada have a lot of gigworkers?
We do, yeah.
I can attest to it personallywith my with my DoorDash bill.
SPEAKER_01 (24:02):
Yeah, the US has an increasing number of gig
workers, and I know a lot ofinstitutions are kicking around
conversationally, how can webetter serve these micro
businesses?
Let's call it not even a smallbusiness, but a micro business.
SPEAKER_02 (24:16):
And if you look at the the data in the small
business world, you know, ourlike our platform was underwrite
uh and our model extends up to250,000.
But if you look at the data, um,you know, 90% of applications
are still under 100,000.
Like that is just where themarket need is.
It is really a lot of reallysmall businesses, typically, you
know, five people or less, uh,and and they don't need huge
(24:39):
sums of money, which makes ithard.
SPEAKER_01 (24:42):
But typically, what kind of default rate do do you
see on the loans?
SPEAKER_02 (24:48):
Yeah, so it varies obviously um you know by credit
union because there's differentdiffering populations and
different credit policies.
So we measure the risk, but wedon't tell credit unions how
much to take.
So they're all doing slightlydifferent things, but in
general, they play in the in asimilar zone.
And what we've seen over thecourse of time is an average
predicted default rate uh uh outof our model based on the policy
(25:11):
chosen by the credit unit ofaround 2%.
Um, so far the default rateshave been less than that.
Um uh so the the actual has beenless than predicted uh for a
variety of reasons.
Um but we do say that you know,believe the model, believe the
predictions, build your pricingand your business model around
the predictions, and and uhyou'll be safe in the long run.
(25:32):
So, you know, that we think ofthis more like um less like a
commercial real estate portfoliowhere you're shooting for a zero
percent loss rate and more likea consumer credit card portfolio
where you might have a two orthree percent loss rate and
you're pricing for it.
SPEAKER_01 (25:47):
Yeah, I to me that would be much more logical.
Uh the credit union can set itsown lending criteria.
They can decide, well, we're notgoing to lend into this sector,
for instance.
True?
SPEAKER_02 (26:03):
Correct.
Yeah.
So there's kind of twocomponents in in simple terms
for uh to the to theimplementation.
One is, you know, we we roll outthe credit model, which doesn't
change by credit union, uh, itproduces a probability default.
That becomes a rule that the thecredit union sets for you know
what what what's the thresholdfor auto approval.
And then beyond that, there's aseries of configurable business
(26:24):
rules uh that the credit unionsets and we work through it with
them.
So you mentioned excludedindustries as one, but it could
be minimum time in business, itcould be um minimum revenue
thresholds.
There's there's about 25 of themthat we look at that um the
credit union can consider.
SPEAKER_01 (26:41):
And the the credit union can also determine how
many months of cash flow theyneed.
SPEAKER_02 (26:47):
That's right.
Yeah, so um we we our minimum issix, so we we won't underwrite
on less than six.
But if uh but if they want tosay we we we won't underwrite on
less than 12 under anycircumstance, they can say that.
SPEAKER_01 (27:01):
I know in the US, quite a few the vast majority of
credit units wouldn't want tolend a dime on to a cannabis
business.
But there's also about a hundredor so that really want to lend
money to cannabis.
SPEAKER_02 (27:14):
Yeah.
Yeah, it's it's funny.
I was um we have a coupleclients that do, but like you
say, most don't.
I was just at a at the CUBGconference this week uh uh in
San Diego, got back last night,and one of the credit unions I
was talking to there was askingme if I knew of any credit
unions that would refer out thatbusiness to them.
So they're quite keen on it.
SPEAKER_01 (27:33):
Oh, yeah, there's a handful of that, and these are
these are smart credit unions.
They often in some cases they'vebeen asked to get into the
business by state government,which wants to get a little bit
of the cash out of that businessand into banks and credit
unions.
Yeah, yeah.
SPEAKER_02 (27:51):
It's interesting, and there's some margin in it, I
think.
So yeah.
SPEAKER_01 (27:55):
So what's what's what's your next step in this
business?
What's the next evolution?
SPEAKER_02 (28:00):
It's sort of continuing to do two things.
Obviously, grow the customerbase and the size of our data
set, because that's one of ourreal assets, is just the the
fact that we've been doing thisfor now you know 10 years,
including our time as a lender,and we've accumulated a pretty
big set of performance data.
So we want to keep continue todo that, but also add
functionality.
So, you know, things like uhwe're working on an embedded
(28:22):
lending project where you canput uh loan applications inside
of different softwareapplications where small
businesses live on the internet.
Uh, we're we're working on asecond look program where we can
actually evaluate a board orbased on two credit policies.
So if they don't pass the creditunion policy, um there's an
instantaneous uh application ofit to the to the second look
(28:46):
partner, and and you'll be ableto, you know, on the spot be
able to tell the small business,hey, we can't help you, but our
partner can.
Um so we're really trying to getto the point where credit unions
can can serve their members uhin a really broadly in terms of
you know, if if they don't havea solution, they can say yes
somehow.
SPEAKER_01 (29:05):
Has um friction with Canada in the US White House had
any impact on your business?
SPEAKER_02 (29:14):
Yeah, very interesting question.
Um, I would say no, it hasn't todate.
Um, and for I guess a couple ofreasons.
Um, you know, there's been notariffs on digital services so
far in either direction.
There was a little bit of a talkabout one about it at one point,
but I think the the US is such anet exporter of digital services
that that probably won't happen.
(29:35):
So that hasn't come come tofruition.
Um, so so it hasn't affected usin terms of the ability to work
with credit unions.
I you know, you could debatewhether there's an economic
impact happening, which we'reyou know, maybe seeing in terms
of credit quality andapplication volume.
Canada, we've seen a uh a bit ofa drop in application volume of
(29:57):
late, and that may be related toeconomic circumstances that.
May be related to tariffs, butyou know, different people will
have different points of view onthat.
SPEAKER_01 (30:05):
Might also, at least in the US, be related to what
increasingly looks like aninflationary spiral.
SPEAKER_02 (30:11):
Yeah, yeah.
And we we haven't seen thechange as acute in the US yet as
as Canada.
Uh, but actually just this weekI got an email from uh uh about
a client that was noticing uhsome some changes in in sort of
the the foundations of theeconomy.
So we'll see.
It's uh interesting time asalways.
(30:32):
Well and one of the one of thethings that um you know we like
about the way we underwrite inour platform is to the extent
things are changes are happeningin real time, we see the changes
to cash flow in real time.
Um so and this was true in thepandemic as well, where well
that's the beauty of doing cashflow.
SPEAKER_01 (30:49):
Yeah, you know, one thing that happens to a small
business, and I've run a smallbusiness for most of my life, is
if when the economy gets routerocky, suddenly those bills that
pay like clockwork on 30 days or60 days or sometimes 90 days.
It's uh so you see like analmost instantaneous impact on
your cash flow if you're a smallbusiness and the economy begins
(31:13):
to go into a recession or everyinflation is out of control,
etc.
It's it's it's so quick.
So you you you would be able tosee a lot of that data too.
SPEAKER_02 (31:24):
Yeah, that's that's the nice thing about the
methodology is when the when theeconomy is dynamic and in flux,
you're you're gonna catch thingssooner than you would with tax
returns or financial statementsor a bureau score, like you say.
SPEAKER_01 (31:37):
Now, how does AI figure into what you do?
And it's in the company websiteaddresses then.
So what what's the where's AIplaying here?
SPEAKER_02 (31:48):
Yeah, there's a couple of spots where we use it.
So, and it's all related to thecash flow data.
The first is just to make senseof the cash flow data.
So you get a bank statement andyou have to turn that into
something that's legible,basically.
So we have a categorizationengine um that that transforms a
bank statement into a syntheticcash flow statement.
So there's some there's somebasic AI in there to do that.
(32:11):
Uh, and then the main the mainapplication of it is in our
credit model itself.
So um the way I was explainedthis is we're we're
transitioning from sort of arules-based underwriting
process, which is what you seewith scorecards that use, say,
financial statement ratios, to abehavioral-based predictive
model.
(32:31):
So we're looking at thebehaviors around cash flow,
figuring out how they interactwith each other uh in a multi
and then producing them a scoreout of a multivariate model.
Um, and so that's a verydifferent approach than saying,
you know, if you if you thinkabout the the the move from that
rules-based approach, like acommon rule would be a 1.25 debt
(32:53):
service coverage ratio.
Um now that that's beenestablished over the course of
time based on intuition andrough experience, but it's not
scientific.
Why isn't it 1.2 or 1.35 or1.31?
Like the it's just sort ofarbitrary, whereas with a
predictive model, you canactually figure out what exactly
that should be.
(33:14):
The second thing is you know,rules or or model variables
don't operate in isolation.
So there may be you know caseswhere um a 1.1 debt service
coverage ratio is totally fineif you have an 800 credit score
or if you keep a million dollarsof cash in your account at all
times.
So there's there's an interplaythat that rules don't capture,
(33:35):
that behavioral models cancapture.
And then the last one that Ialways talk about is what if you
just have the wrong rule?
Um, and this one's reallyinteresting.
So debt service coverage is isfascinating because that's the
way that commercial lending'salways been done.
But with small businesses,they're actually cash out
businesses in our experience.
And we have lots of data to showthis.
So they don't run profitably.
(33:56):
Um, uh, they they theydeliberately commingle personal
expenses and conflate capex anddividends.
And so so they're they'redriving to roughly a break-even
point uh uh on a net operatingcash flow perspective.
And so when you when you apply atraditional commercial lending
metric like debt servicecoverage, it's gonna lead you to
say no most of the time.
(34:18):
And but that's not necessarilythe right answer.
And so what the AI can do, andit can look for different
variables, different behaviorsthat are predictive of risk uh
beyond the traditional metrics.
SPEAKER_01 (34:30):
And what you're saying makes sense to me.
I was a partner in a business,and when you say that, I
remember my partner and I wouldpay ourselves more or less,
depending upon the cash flow.
And and we wanted to have acertain amount of money kicking
around the company accounts, butin the good months, we take more
(34:52):
out for ourselves.
It's um in a bad month, we takea lot less out.
So it's so if you're using thatmetric, it's not always gonna
make a lot of sense with abusiness that runs like that.
And I think a lot of them do runthat way.
SPEAKER_02 (35:06):
So and that's why we really, you know, one of the
sort of our mantras is to thinkabout small business as a
segment.
It's not commercial lending andit's not consumer lending, it's
it's in the middle, and it's itreally is different.
It's big enough to be its ownsegment, and you and it needs a
different mindset, it needsdifferent tech, it needs
different credit criteria, itneeds a different owner inside
(35:26):
the credit union, all that kindof stuff.
And it's underserved, it's agreat opportunity.
Yeah, hugely underserved.
Um, and and the data shows itall everywhere you look.
SPEAKER_01 (35:40):
And there's all kinds of small businesses.
There are farms, for instance.
I mean, fewer and fewer, to behonest, that are small
businesses, but there's still inparts of the United States,
there are a lot of small, smallfarms that probably have trouble
getting loans.
It's um, I mean, it's awonderful opportunity there.
SPEAKER_02 (35:59):
Yeah, there's no question that it's it's a a
massively underservedopportunity.
Um, one that could, you know,serving it better could have
huge impact.
Um, there was a study out atMcKinsey recently that talked
about how um small businessesare about half as productive as
larger businesses, and part ofthe challenge uh there has been
they're undercapitalized, sothey can't get as productive.
(36:21):
And and the punchline in thewhole thing was the impact of
making those businesses moreproductive with more capital
could could be something like uhto the tune of five to ten
percent of GDP in the US.
So it's just a an enormous issuefor the economy that that uh and
if we solve it would have hugeimpact, but also could be quite
(36:42):
profitable.
Again, you know, years ago youcouldn't do this profitably, but
now you can.
It actually can be one of yourbetter lines of business.
SPEAKER_01 (36:49):
Yeah, yeah.
And for the doubters, we sawexactly this during the pandemic
government loans.
It was purely automated,streamlined.
Yeah, there was no risk becausethey were all government backed,
but nonetheless, it was theycould make loans really quickly.
SPEAKER_02 (37:05):
And aside from the workflow that with the
technology now and the you know,the models, the AI models, you
can you can manage the creditrisk quite effectively too
without spending day hours anddays on a file.
SPEAKER_01 (37:18):
Before we go, think hard about how you can help
support this podcast so we cando more interviews with more
thoughtful leaders in the creditunion world.
What we're trying to figure outhere in these podcasts is what's
next for credit unions.
What can they do to really,really, really make a difference
in the financial scene?
Can't all be mega banks, can it?
(37:40):
It's my hope it won't all bemega banks.
It's it'll always be a place forcredit unions.
That's what we're discussinghere.
To figure out how you can help,get in touch with me.
This is RJMegarvey at gmail.com.
Robert McGarvey again.
That's RJMegarvey at gmail.com.
Get in touch, we'll figure out away that you can help.
We need your support, we wantyour support, we thank you for
(38:03):
your support.
The CU2.0 Podcast.
Welcome to the CU2.0 podcast.
SPEAKER_00 (00:05):
Hi, and welcome to the CU2.0 podcast with big new
ideas about credit unions andconversations about innovative
technology with credit union andfintech leaders.
This podcast is brought to youby Quillo, the real-time loan
syndication network for creditunions, and by your host,
long-time credit union andfinancial technology journalist
(00:27):
Robert McGarvey.
And now the CU 2.0 podcast withRobert McGarvey.
SPEAKER_01 (00:35):
About 45 credit unions are customers of Judy.ai
and they're drawn by thecompany's simple promise.
Eliminate friction in businesslending.
There's money to be madestirring small businesses in.
(01:03):
Quote, small businesses are thebackbone of our economy.
But 40 to 50 percent of smallbusinesses rank access to
capital as their chiefchallenge.
At Judy, we're transformingsmall business dreams into
vibrant communities byincreasing access to capital.
That's a big promise.
(01:23):
But since basically the companyhas had very, very little turn
among credit unions who signedup.
That's because they believethey're getting value.
And also, in fact, they'refinding ways to serve the
community's small business.
Just about every credit unit,certainly all bigger than$500
million in assets, dreams aboutmaking small business, lending
(01:45):
them bigger slides for theirportfolio.
Credit unions are finding thatthey can do it.
And some of those credit unionsas are as small as maybe$25
million in assets.
(02:09):
And historically, as I'm sureyou know, small business lending
has been something credit unionshave wanted to do, but they
haven't done it for manyreasons.
One of which is that it's notusually profitable for them
because the amounts in theirlending are too small to justify
the labor that goes into bettingthe app loan application to
(02:34):
simplify matters.
I mean, and that's true with bigbanks too.
If I walk in the chase and say,hey, I'd like you to give my
small business a$50,000 loan, mybanker there would say, um,
we'll give you a personal loan,but not your business.
SPEAKER_02 (02:50):
Yeah.
And I mean, it you're right,it's not just credit unions,
it's the broader market.
And if you look at the you know,sentiment data out of any survey
of small businesses, they'll saythey're underserved from a
banking perspective and a creditperspective in particular.
It's just a hard segment toserve.
And so what's happened is youknow, fintech lenders have
stepped in to fill that void.
The problem is they're notdeposit-taking institutions, so
(03:13):
their cost of capital is prettyhigh, and so they have to pass
that along.
And next thing you know, youryou know, your APRs are 30, 40
percent.
SPEAKER_01 (03:22):
Yeah, I talked with a guy uh a couple of years ago
who does just small businesslending, and he told me
traditional financialinstitutions we're kind of dead
at the starting block becausethey they ask for the wrong
data.
Um what's your Dun andBradstreet writing?
My what?
And furthermore, yes, the DunhamBradstreet doesn't know who this
(03:44):
company is.
Two, what's what's your FICOscore?
Well, some of them are havingmoney problems, and their FICO
score is really taking a tumble.
So he said, all I look at iscash flow.
Right, exactly.
And he I assume his interestrate, he wasn't specific, but it
was over 20, significantly over20 percent.
SPEAKER_02 (04:02):
Yeah, and that's that's the the fintech lenders
sort of figured this out uh uh awhile back.
And we actually we were born outof an in uh a fintech lender
ourselves and then developed aplatform and spun out a separate
business for for that business.
But exactly what we learnedalong the way is that cash flow
was really the the the bestpredictor of of what's going on
(04:23):
and and actually most recentcash flow is really important.
So um, you know, the we look at12 months of cash flow data, and
I think that's all this guywanted was 12 months.
SPEAKER_01 (04:33):
If he had 24, he might look at it.
But yeah, but his main concernwas if I lend you this money,
can you make the monthlypayments?
And the cash flow told him yesor no.
SPEAKER_02 (04:44):
Well, and there are some fintech lenders that will
lend on the back of only threemonths of cash flow data.
Whoa, yeah, and and we've evenseen it in our data that you
know, months one to three aremuch more predictive than months
nine to twelve from a from await a minute though.
SPEAKER_01 (04:58):
If I open the right kind of store uh and I give you
cash flow data for October,November, December, I might have
zero cash flow in Januarybecause I'm selling Christmas
gifts, you know.
SPEAKER_02 (05:11):
It's uh so yeah, and and and and I'm speaking in
general out generalities acrossthe small business population.
Obviously, there's seasonalityin some businesses, but but for
the most part, uh recent data ismore predictive than old data.
SPEAKER_01 (05:25):
Um sure.
SPEAKER_02 (05:27):
That's uh and it's pretty acute actually um across
the general population of smallbusinesses.
SPEAKER_01 (05:33):
I mean, that's that's essentially why I really
don't like FICO scores as ameasure of much of anything
except a pretty good picture ofmy past behavior as a credit
card user.
But I don't think it tells you athing about my future behavior.
SPEAKER_02 (05:46):
Yeah, I mean, I think you know, a certain uh a
certain baseline uh uh of a FICOscore will tell you what kind of
a person you're dealing with,and are they likely to you know
completely screw up or or evenworse, you know, defraud you so
you get a sort of sort of abaseline character perspective
out of that, I guess.
Um and then and then from therethough, you really want to know
(06:07):
how their business is doing inin real time.
SPEAKER_01 (06:10):
Now, how many credit union customers do you have?
And how many US-based creditunions do you have?
SPEAKER_02 (06:18):
Yeah, we're working with uh I think we're creeping
up towards 45 customers.
SPEAKER_01 (06:22):
Uh that that's outstanding.
How how long has it taken you toget 45?
SPEAKER_02 (06:27):
Yeah, we launched, so our first customer um we was
Van City Credit Union, so thelargest credit union in Canada.
Um, we launched with them in uhMay of 2017.
So that was that was when thefirst application went through
our first credit union customer.
Um, and so that's uh you knoweight years ago now um that
we've been doing this and uhadding them steadily.
(06:50):
We we sort of got into the USmarket a couple of years ago.
That's been going well.
We're I think up to 18 customersin the US and uh more coming on,
obviously, lots of potentialthere.
But um, we really, to yourpoint, we've the credit unions
in both Canada and US, andthey're very similar, have
historically not done a lot ofsmall business lending for all
of the usual reasons.
It's expensive and unprofitable.
(07:11):
And um uh so they've migrated,you know, they've always focused
on CRE to the extent they'vedone commercial lending, but
they are we we're reallynoticing a pattern now where
they want to get into this.
They know they should servetheir members in this segment,
huge important segment of theeconomy.
And so they're they're startingto run at it, uh, which has been
fun for us.
SPEAKER_01 (07:28):
What's the size range approximately of your US
credit units?
SPEAKER_02 (07:34):
Yeah, so we've got so we work with um uh uh through
through a couple of CUSOs, weworked with some very small
credit unions, uh, as low as 25million in assets.
Um, and then uh like I say, ourlargest client is 30 billion in
assets, and in there's a we havea couple of those in Canada and
and pretty much everywhere inbetween.
(07:55):
Um uh we do sort of notice thatsort of 500 million in assets
and up is where uh if you're notgoing through a CUSO, where it
where it kind of works, wherethey can actually um absorb the
cost and execute uh on a programthat that makes sense.
So that's a kind of a thresholdthat we focus on.
But um if smaller ones come tous, we certainly will work with
(08:15):
them and we'll uh sometimessuggest that we pull in a CUSO
to work with them and help them.
SPEAKER_01 (08:20):
Now, remind me, my my memory tells me that you know
the US has approximately 6,000credit units, Canada has
approximately six.
I mean, I exaggerate, but yeah,Canada has really big credit
units, right?
SPEAKER_02 (08:35):
Yeah, we're so yeah, we're um about 250 credit unions
in Canada, you know, and and soum 10% of the population.
So if you were to sort of do alinear extrapolation, we should
have um you there should be2,500 in the US to be
equivalent, and there's at leastdouble that or something like
double that.
So yeah, you're right, we'reless fragmented and generally
(08:58):
have larger credit unions umhere uh and fewer targets for
us.
But uh so it's a bit of adifferent dynamic.
SPEAKER_01 (09:05):
How how does the QZO come into the picture when
you're serving a credit unionthrough a QZO?
SPEAKER_02 (09:11):
Yeah, so some of the credit unions, uh because as you
say, they haven't historicallydone a lot of commercial lending
and certainly smaller C and Ilending, some of them maybe
don't have the expertise to dothe underwriting or the
servicing or the reviews.
And so um they can outsourcesome or all of that process to
uh uh to a CUSO.
(09:31):
And so where we can partner witha CUSO and bring both technology
and the expertise and servicesto the table, it enables someone
to get up and running veryquickly without building out a
whole department.
SPEAKER_01 (09:41):
So take tell me, what do you bring to the table
and what does the credit unionhave to bring to the table,
either on its own or through aCUSO?
SPEAKER_02 (09:52):
Yeah, so we bring the technology and a little bit
of advice around how toconfigure the technology for um
a program based on whatever therisk and business objectives are
of the of the credit union.
Um and we help them through thatjourney.
But but what we don't do is wedon't do underwriting, um, we
don't write credit policies, uh,we don't do annual reviews, um,
(10:14):
but we can provide thetechnology to enable all of
that.
So uh they either need um peopleinternally that are capable of
that and that have the capacity.
That's another thing.
Sometimes it's it's capability,sometimes it's capacity.
Um, and if they don't have itinternally, then they then they
need a CUSO, and there areseveral around that provide
those services.
SPEAKER_01 (10:33):
So your technology automates the lending process.
Yeah, yeah.
So it does a few things.
And so in in the in the pandemicera, and there was a lot of
government money to lend in uhthe US, government backed loans.
Uh the big banks jumped on that.
Like Wells Fargo did a lot ofthose loans.
(10:55):
When I investigated, it's it'severything was automated.
I don't know what the dollaramount was, was their maximum,
but say it was 100 grand.
Anything under that was gonna beautomated.
No human being would touch theapplication.
Credit unions struggled withthat, they didn't have the
technology, I don't think.
SPEAKER_02 (11:15):
Yeah, and it's interesting, like and in Canada
had a similar program too, sothe dynamics were very similar.
And um, you needed to automateit.
There was so much volume coming,but but it was really workflow,
not underwriting automation, youknow, because there wasn't
really underwriting, like theywere right.
SPEAKER_01 (11:31):
There was essentially no, but again, it's
how much labor can you afford toput into this twenty thousand
dollar loan?
Yeah, very little, basically.
SPEAKER_02 (11:41):
Yeah, no, we've had uh had uh people say to us like
once we've once we've touchedthe file twice, we've lost
money.
SPEAKER_01 (11:47):
Yeah.
So uh absolutely it's theythey'd much rather do a car loan
for thirty thousand dollars.
Yeah.
SPEAKER_02 (11:53):
So the the flip side is that um, although they're
small amounts, there tends to bepretty good margin in the
segment um uh for a number ofreasons.
And so if you can figure out howto control the cost of delivery
and do it at scale, it canactually, like some of our
clients, this is their mostprofitable segment.
So um once you get to thatpoint, it it's it's very
worthwhile.
SPEAKER_01 (12:15):
Now, you don't, I'm assuming you don't provide
marketing advice to the creditunion.
In other words, here's how yougo after small business lots, A,
B, C, and D.
SPEAKER_02 (12:28):
Yeah, we actually do a few things on that front.
Um, so so um the the big thingthat we've noticed, um I always
say, you know, if you build it,they won't come.
Like you can't just stand up aplatform and expect that you're
gonna get applications all of asudden.
So you do have to activate it.
And and one of the mostsuccessful ways to activate it
is to actually dopre-qualifications of your
members.
(12:48):
So most credit unions aresitting on a few thousand uh
business members um and andthey're sitting on all their
account data, that cash flowdata that is very predictive.
And we can go through that andgive them a picture of who they
could make offers to.
And the the conversion rates andthe uptake from small businesses
has been incredible on thosetypes of exercises, I think, for
two reasons.
One, um, there's a high degreeof trust between the business
(13:12):
members and the credit union,and two, business members,
unlike consumers, aren't used togetting these kinds of offers.
And so we've had um we had oneclient do do one of these Libro
credit union in in Ontario andCanada, and they had a 30%
conversion rate on offers madeout of one campaign, uh, just to
give you a feel for howsuccessful that was.
But again, you do have to beproactive.
(13:33):
Pre-quall is one way to do it.
You can do digital emailcampaigns to your members, and
of course, you can get out anddo you know more traditional
digital marketing beyond yourmembership base.
What we've seen with the creditunions is they have so much
low-hanging fruit in theirbusiness member base that they
haven't had to go outside yet.
They can focus on that.
The other area that's alsointeresting is um, and we've got
(13:55):
a partner, uh, Fingle out ofColorado, that can do some
analysis on this, uh, but butthey can comb through uh a
credit union's consumer accountsand predict which ones are uh in
all likelihood running abusiness out of those accounts.
And it's usually somewherearound 10% of your consumer
members are are running abusiness out of the account.
So um there's an opportunity tosort of migrate those over to
(14:17):
business services.
SPEAKER_01 (14:18):
That's that's interesting.
You know, credit unions in theUS is for some years now have
technology that lets them, andit happened to me with my credit
union.
I didn't have a credit card.
They sent me an email or aletter saying, hey, you know,
what we we we really want togive you this credit, this
credit card, like a$20,000limit.
(14:40):
There's no fee.
Do you want it?
I said, huh?
Yeah, I guess so.
But but they're they're cutthey're accustomed to doing
that.
So this is just doing that samething with with small business
members.
SPEAKER_02 (14:55):
It is, and and and it's easy on the consumer side
because if you have a bureauscore, you're comfortable doing
it on the back of that,typically.
Um, but it with a smallbusiness, that doesn't quite cut
it per your previous points thatyou were making around that.
So you have to look into thecash flow and uh and be able to
make sense of the cash flow inthe in the operating account to
decide whether that's a an offeryou can make.
SPEAKER_01 (15:17):
I I don't think my credit union even had a Euro
score when they offered me that.
It's uh I think they just theyjust looked at um my accounts to
balance in them, et cetera.
And so by heavens, he's paying$4,000 a month to American
Express.
Why don't we get a piece ofthis?
SPEAKER_02 (15:36):
Yeah, there could be some of that.
I know a lot of them um uh forcompliance reasons at least do
soft polls on their on theirconsumer members.
So a lot of them will have thatthat on file.
SPEAKER_01 (15:46):
And I could be a soft poll that I never would
have seen.
That's that's true.
That's uh and how long does ittake a credit union?
A credit union calls you up whenthey hear this podcast and they
say, we want to get started onthis.
Now that won't happen.
They're probably gonna ask youto convince them to get started
on it.
But we'll we'll just imaginethat.
It's uh how long does theprocess take?
SPEAKER_02 (16:10):
Yeah, so we usually guide people to sort of allow
for eight to 12 weeks to getlive.
We've done them in as little asfour weeks.
They've also gone longer,depending on some uh uh some of
the choices and dynamics in theaccount.
But um, if we keep it simple andwe really want to just get up
with uh uh a basicimplementation, it can be pretty
quick.
And if in fact, if someone cameto us and said we need to be
(16:32):
live in a week, we we could, youknow, in theory, uh get
everybody uh aligned veryquickly and be live that quick.
So it it is um it it doesn'thave to be a massive project.
You know, I think people thinkabout core conversions or
massive LOS implementations andthey get scared off, but this
this problem can be actuallysolved quite quickly.
SPEAKER_01 (16:53):
Do you need core system access?
SPEAKER_02 (16:56):
You don't need it out of the gate.
No, so um the way we can uhaccess the cash flow data uh is
through the account aggregatorsin the market.
So there's a bunch of them outthere that you know, PLAD, et
cetera.
We use Yodly um uh as ourprimary access point.
And the reason we do that is umA, you don't need to integrate
out of the out of the gate,which uh expedites the process,
(17:16):
but B, you can also accessaccounts at you know other
institutions.
And about 20% of theapplications we receive have
accounts, you know, multipleaccounts at multiple
institutions.
SPEAKER_01 (17:26):
And increasingly more and more of us are familiar
with Plat at least.
Yeah.
SPEAKER_02 (17:33):
Yeah, it's a it's a journey, and I always always say
it's you know, open banking andquotations, we're not there yet,
but there are increase anincreasing number of open
banking-like, you know, APIconnections into these
institutions that are, you know,uh one day this this technology
will be ubiquitous and hopefullyon a standard protocol.
SPEAKER_01 (17:53):
And what's your fee structure?
SPEAKER_02 (17:56):
Yeah, so we charge um uh a subscription fee for the
platform that that varies bysize of credit union, and then
we charge a fee per completedloan application.
Um, so anything that we scorebasically, we we charge for.
And in that in that fee, wewould absorb the cost of the
Bureau scores and the dataconnections.
And so it's fully loaded, theDocuSign envelope for the loan
(18:17):
agreements.
SPEAKER_01 (18:18):
Um what's what's a range for the the monthly fee?
SPEAKER_02 (18:25):
Yeah, the the range is uh as low as 4,000 a month
and as high as 30,000 a monthfor the the massive credit
unions and uh everywhere inbetween, and then the
application fees$140 for the thecompleted and what's what
staffing specifically does thecredit union need on its end to
(18:46):
make this work?
Yeah, we sort of we we recommendlike three primary stakeholders
in a successful implementation.
And the first one is some sortof an executive sponsor that's
really going to make sure thatthere's you know strategic
prioritization at the top of theorganization.
And if you have that, thatthat's a key marker of success.
Um, the second one would be um aforward-thinking subject matter
(19:09):
expert around lending andcommercial lending that maybe
wants to make a change.
So somebody that's comfortablewith technology and changing
process, and that can thatperson would be the driver.
And then a project manager justto get it going and they don't
have to stay with it, you know,much beyond the implementation,
but having somebody justcoordinate with you know
marketing and legal and um andcredit risk, et cetera, to make
(19:32):
sure that it gets off theground.
And it's not a ton of work uhwith all of those stakeholders,
but they do need to be informedat least.
SPEAKER_01 (19:39):
Have you had customers who are doing this and
then decide to quit?
And if so, why?
SPEAKER_02 (19:47):
Yeah, good question.
We've had um to the extent we'vehad churn, um, we had a couple
that used it for uh the SIBAprogram in Canada, um, which was
like the PP program and the PPPprogram in the US, and then they
didn't continue on.
Uh, they tended to be thesmaller credit unions that
didn't have the resources to doit.
So that that's really the thebiggest blocker that we've seen
(20:09):
is if you just don't have thestrategic prioritization and the
resources to to move forward.
Um, and so that's where we'vesort of noticed there's a line
there at around 500 million inassets, where below that you may
need the help of a CUSO to pushforward.
SPEAKER_01 (20:25):
I've often asked credit union CEOs, why don't you
have blah, blah, blah?
And the answer often in manycases is I'd love to, but I
don't have the staff.
And these these are smallercredit units, these are not
credit units, and certainly notones over a billion dollars in
assets.
Although I'll tell you, in thelast couple of months, twice
(20:48):
now, CEOs of billion-dollarcredit units have referred to
themselves as running a smallcredit unit.
Twice.
Two different guys.
So billions now small.
Okay.
SPEAKER_02 (21:02):
Yeah, the landscape's changing for sure.
And um, yeah, those thoseresources are are critical.
And if you don't have them, itwon't move forward.
And one of the challenges thatwe see quite frequently is, you
know, like you say, the legacy,sort of the heritage of of
credit unions is on the consumerside typically.
So so these the commercial sideof the house tends to have it
struggle a little more to getthe resources for these kinds of
(21:24):
projects to to grow that side.
So a bit of a chicken and egg.
SPEAKER_01 (21:28):
That said, um many credit unions recognize that
basically they're forcing theircustomer out, their member out
the door.
Yeah, if I start getting mybusiness loans from Wells Fargo
uh or a fintech, what what makesyou think I'm gonna stay around
as a credit union member forvery long?
(21:48):
And what credit unions don'tlike to analyze a heck of a lot
is share of wallet.
And a generation ago, they oftenhad a hundred percent share of
wallet from many of theirmembers.
That's uncommon today.
Um my credit union has maybe 15%of my wallet, maybe it's uh and
(22:11):
uh and I'm perfectly happy withit.
SPEAKER_02 (22:14):
Well, and a couple of things to that point.
One is um if you look at the ifyou deconstruct the data, about
25% of U.S.
households have a small businessattached to them.
Um, so it's important to be ableto serve those small businesses
because a big chunk of yourpopulation is going to need that
service.
Uh, secondly, and and you know,this has come to light with some
of our more successful clientslike Van City, that's now been
(22:37):
doing this for eight years andreally leaned into it.
You know, they're now doing,they've done in they've had
months where they've done 500applications through our
platform.
Uh, but they treat it as a verystrategic segment.
They view it because of thatsort of high incidence rate in
the population.
Uh, but also the the the I guessthe the sort of emotional
component to a small business,like if if if you fund someone's
(22:58):
small business dream, chancesare they're sticking with you
and they're probably bringingeverything else with them.
Um so there's a really strategiccomponent to the relationship
out of that segment.
SPEAKER_01 (23:08):
Right.
I mean, if you do a smallbusiness loan, almost certainly
the business will ask for asmall business credit card too,
eventually, if you have that tooffer.
Um I mean it's now will yoursystem accommodate gig workers,
like a full-time Uber driver,that sort of thing.
SPEAKER_02 (23:30):
The yeah, the where where our system works is it if
if the business has a businessoperating account, if it's a if
it's a you know a side businessrun out of a consumer account,
um, we we sort of think of thatas a consumer loan.
So it's it really needs to havea business operating account.
So if that gig worker setsthemselves up as a business, we
could do it.
But if it's sort of a sidething, then it you'd probably
(23:51):
treat it as a consumer loan.
Does Canada have a lot of gigworkers?
We do, yeah.
I can attest to it personallywith my with my DoorDash bill.
SPEAKER_01 (24:02):
Yeah, the US has an increasing number of gig
workers, and I know a lot ofinstitutions are kicking around
conversationally, how can webetter serve these micro
businesses?
Let's call it not even a smallbusiness, but a micro business.
SPEAKER_02 (24:16):
And if you look at the the data in the small
business world, you know, ourlike our platform was underwrite
uh and our model extends up to250,000.
But if you look at the data, um,you know, 90% of applications
are still under 100,000.
Like that is just where themarket need is.
It is really a lot of reallysmall businesses, typically, you
know, five people or less, uh,and and they don't need huge
(24:39):
sums of money, which makes ithard.
SPEAKER_01 (24:42):
But typically, what kind of default rate do do you
see on the loans?
SPEAKER_02 (24:48):
Yeah, so it varies obviously um you know by credit
union because there's differentdiffering populations and
different credit policies.
So we measure the risk, but wedon't tell credit unions how
much to take.
So they're all doing slightlydifferent things, but in
general, they play in the in asimilar zone.
And what we've seen over thecourse of time is an average
predicted default rate uh uh outof our model based on the policy
(25:11):
chosen by the credit unit ofaround 2%.
Um, so far the default rateshave been less than that.
Um uh so the the actual has beenless than predicted uh for a
variety of reasons.
Um but we do say that you know,believe the model, believe the
predictions, build your pricingand your business model around
the predictions, and and uhyou'll be safe in the long run.
(25:32):
So, you know, that we think ofthis more like um less like a
commercial real estate portfoliowhere you're shooting for a zero
percent loss rate and more likea consumer credit card portfolio
where you might have a two orthree percent loss rate and
you're pricing for it.
SPEAKER_01 (25:47):
Yeah, I to me that would be much more logical.
Uh the credit union can set itsown lending criteria.
They can decide, well, we're notgoing to lend into this sector,
for instance.
True?
SPEAKER_02 (26:03):
Correct.
Yeah.
So there's kind of twocomponents in in simple terms
for uh to the to theimplementation.
One is, you know, we we roll outthe credit model, which doesn't
change by credit union, uh, itproduces a probability default.
That becomes a rule that the thecredit union sets for you know
what what what's the thresholdfor auto approval.
And then beyond that, there's aseries of configurable business
(26:24):
rules uh that the credit unionsets and we work through it with
them.
So you mentioned excludedindustries as one, but it could
be minimum time in business, itcould be um minimum revenue
thresholds.
There's there's about 25 of themthat we look at that um the
credit union can consider.
SPEAKER_01 (26:41):
And the the credit union can also determine how
many months of cash flow theyneed.
SPEAKER_02 (26:47):
That's right.
Yeah, so um we we our minimum issix, so we we won't underwrite
on less than six.
But if uh but if they want tosay we we we won't underwrite on
less than 12 under anycircumstance, they can say that.
SPEAKER_01 (27:01):
I know in the US, quite a few the vast majority of
credit units wouldn't want tolend a dime on to a cannabis
business.
But there's also about a hundredor so that really want to lend
money to cannabis.
SPEAKER_02 (27:14):
Yeah.
Yeah, it's it's funny.
I was um we have a coupleclients that do, but like you
say, most don't.
I was just at a at the CUBGconference this week uh uh in
San Diego, got back last night,and one of the credit unions I
was talking to there was askingme if I knew of any credit
unions that would refer out thatbusiness to them.
So they're quite keen on it.
SPEAKER_01 (27:33):
Oh, yeah, there's a handful of that, and these are
these are smart credit unions.
They often in some cases they'vebeen asked to get into the
business by state government,which wants to get a little bit
of the cash out of that businessand into banks and credit
unions.
Yeah, yeah.
SPEAKER_02 (27:51):
It's interesting, and there's some margin in it, I
think.
So yeah.
SPEAKER_01 (27:55):
So what's what's what's your next step in this
business?
What's the next evolution?
SPEAKER_02 (28:00):
It's sort of continuing to do two things.
Obviously, grow the customerbase and the size of our data
set, because that's one of ourreal assets, is just the the
fact that we've been doing thisfor now you know 10 years,
including our time as a lender,and we've accumulated a pretty
big set of performance data.
So we want to keep continue todo that, but also add
functionality.
So, you know, things like uhwe're working on an embedded
(28:22):
lending project where you canput uh loan applications inside
of different softwareapplications where small
businesses live on the internet.
Uh, we're we're working on asecond look program where we can
actually evaluate a board orbased on two credit policies.
So if they don't pass the creditunion policy, um there's an
instantaneous uh application ofit to the to the second look
(28:46):
partner, and and you'll be ableto, you know, on the spot be
able to tell the small business,hey, we can't help you, but our
partner can.
Um so we're really trying to getto the point where credit unions
can can serve their members uhin a really broadly in terms of
you know, if if they don't havea solution, they can say yes
somehow.
SPEAKER_01 (29:05):
Has um friction with Canada in the US White House had
any impact on your business?
SPEAKER_02 (29:14):
Yeah, very interesting question.
Um, I would say no, it hasn't todate.
Um, and for I guess a couple ofreasons.
Um, you know, there's been notariffs on digital services so
far in either direction.
There was a little bit of a talkabout one about it at one point,
but I think the the US is such anet exporter of digital services
that that probably won't happen.
(29:35):
So that hasn't come come tofruition.
Um, so so it hasn't affected usin terms of the ability to work
with credit unions.
I you know, you could debatewhether there's an economic
impact happening, which we'reyou know, maybe seeing in terms
of credit quality andapplication volume.
Canada, we've seen a uh a bit ofa drop in application volume of
(29:57):
late, and that may be related toeconomic circumstances that.
May be related to tariffs, butyou know, different people will
have different points of view onthat.
SPEAKER_01 (30:05):
Might also, at least in the US, be related to what
increasingly looks like aninflationary spiral.
SPEAKER_02 (30:11):
Yeah, yeah.
And we we haven't seen thechange as acute in the US yet as
as Canada.
Uh, but actually just this weekI got an email from uh uh about
a client that was noticing uhsome some changes in in sort of
the the foundations of theeconomy.
So we'll see.
It's uh interesting time asalways.
(30:32):
Well and one of the one of thethings that um you know we like
about the way we underwrite inour platform is to the extent
things are changes are happeningin real time, we see the changes
to cash flow in real time.
Um so and this was true in thepandemic as well, where well
that's the beauty of doing cashflow.
SPEAKER_01 (30:49):
Yeah, you know, one thing that happens to a small
business, and I've run a smallbusiness for most of my life, is
if when the economy gets routerocky, suddenly those bills that
pay like clockwork on 30 days or60 days or sometimes 90 days.
It's uh so you see like analmost instantaneous impact on
your cash flow if you're a smallbusiness and the economy begins
(31:13):
to go into a recession or everyinflation is out of control,
etc.
It's it's it's so quick.
So you you you would be able tosee a lot of that data too.
SPEAKER_02 (31:24):
Yeah, that's that's the nice thing about the
methodology is when the when theeconomy is dynamic and in flux,
you're you're gonna catch thingssooner than you would with tax
returns or financial statementsor a bureau score, like you say.
SPEAKER_01 (31:37):
Now, how does AI figure into what you do?
And it's in the company websiteaddresses then.
So what what's the where's AIplaying here?
SPEAKER_02 (31:48):
Yeah, there's a couple of spots where we use it.
So, and it's all related to thecash flow data.
The first is just to make senseof the cash flow data.
So you get a bank statement andyou have to turn that into
something that's legible,basically.
So we have a categorizationengine um that that transforms a
bank statement into a syntheticcash flow statement.
So there's some there's somebasic AI in there to do that.
(32:11):
Uh, and then the main the mainapplication of it is in our
credit model itself.
So um the way I was explainedthis is we're we're
transitioning from sort of arules-based underwriting
process, which is what you seewith scorecards that use, say,
financial statement ratios, to abehavioral-based predictive
model.
(32:31):
So we're looking at thebehaviors around cash flow,
figuring out how they interactwith each other uh in a multi
and then producing them a scoreout of a multivariate model.
Um, and so that's a verydifferent approach than saying,
you know, if you if you thinkabout the the the move from that
rules-based approach, like acommon rule would be a 1.25 debt
(32:53):
service coverage ratio.
Um now that that's beenestablished over the course of
time based on intuition andrough experience, but it's not
scientific.
Why isn't it 1.2 or 1.35 or1.31?
Like the it's just sort ofarbitrary, whereas with a
predictive model, you canactually figure out what exactly
that should be.
(33:14):
The second thing is you know,rules or or model variables
don't operate in isolation.
So there may be you know caseswhere um a 1.1 debt service
coverage ratio is totally fineif you have an 800 credit score
or if you keep a million dollarsof cash in your account at all
times.
So there's there's an interplaythat that rules don't capture,
(33:35):
that behavioral models cancapture.
And then the last one that Ialways talk about is what if you
just have the wrong rule?
Um, and this one's reallyinteresting.
So debt service coverage is isfascinating because that's the
way that commercial lending'salways been done.
But with small businesses,they're actually cash out
businesses in our experience.
And we have lots of data to showthis.
So they don't run profitably.
(33:56):
Um, uh, they they theydeliberately commingle personal
expenses and conflate capex anddividends.
And so so they're they'redriving to roughly a break-even
point uh uh on a net operatingcash flow perspective.
And so when you when you apply atraditional commercial lending
metric like debt servicecoverage, it's gonna lead you to
say no most of the time.
(34:18):
And but that's not necessarilythe right answer.
And so what the AI can do, andit can look for different
variables, different behaviorsthat are predictive of risk uh
beyond the traditional metrics.
SPEAKER_01 (34:30):
And what you're saying makes sense to me.
I was a partner in a business,and when you say that, I
remember my partner and I wouldpay ourselves more or less,
depending upon the cash flow.
And and we wanted to have acertain amount of money kicking
around the company accounts, butin the good months, we take more
(34:52):
out for ourselves.
It's um in a bad month, we takea lot less out.
So it's so if you're using thatmetric, it's not always gonna
make a lot of sense with abusiness that runs like that.
And I think a lot of them do runthat way.
SPEAKER_02 (35:06):
So and that's why we really, you know, one of the
sort of our mantras is to thinkabout small business as a
segment.
It's not commercial lending andit's not consumer lending, it's
it's in the middle, and it's itreally is different.
It's big enough to be its ownsegment, and you and it needs a
different mindset, it needsdifferent tech, it needs
different credit criteria, itneeds a different owner inside
(35:26):
the credit union, all that kindof stuff.
And it's underserved, it's agreat opportunity.
Yeah, hugely underserved.
Um, and and the data shows itall everywhere you look.
SPEAKER_01 (35:40):
And there's all kinds of small businesses.
There are farms, for instance.
I mean, fewer and fewer, to behonest, that are small
businesses, but there's still inparts of the United States,
there are a lot of small, smallfarms that probably have trouble
getting loans.
It's um, I mean, it's awonderful opportunity there.
SPEAKER_02 (35:59):
Yeah, there's no question that it's it's a a
massively underservedopportunity.
Um, one that could, you know,serving it better could have
huge impact.
Um, there was a study out atMcKinsey recently that talked
about how um small businessesare about half as productive as
larger businesses, and part ofthe challenge uh there has been
they're undercapitalized, sothey can't get as productive.
(36:21):
And and the punchline in thewhole thing was the impact of
making those businesses moreproductive with more capital
could could be something like uhto the tune of five to ten
percent of GDP in the US.
So it's just a an enormous issuefor the economy that that uh and
if we solve it would have hugeimpact, but also could be quite
(36:42):
profitable.
Again, you know, years ago youcouldn't do this profitably, but
now you can.
It actually can be one of yourbetter lines of business.
SPEAKER_01 (36:49):
Yeah, yeah.
And for the doubters, we sawexactly this during the pandemic
government loans.
It was purely automated,streamlined.
Yeah, there was no risk becausethey were all government backed,
but nonetheless, it was theycould make loans really quickly.
SPEAKER_02 (37:05):
And aside from the workflow that with the
technology now and the you know,the models, the AI models, you
can you can manage the creditrisk quite effectively too
without spending day hours anddays on a file.
SPEAKER_01 (37:18):
Before we go, think hard about how you can help
support this podcast so we cando more interviews with more
thoughtful leaders in the creditunion world.
What we're trying to figure outhere in these podcasts is what's
next for credit unions.
What can they do to really,really, really make a difference
in the financial scene?
Can't all be mega banks, can it?
(37:40):
It's my hope it won't all bemega banks.
It's it'll always be a place forcredit unions.
That's what we're discussinghere.
To figure out how you can help,get in touch with me.
This is RJMegarvey at gmail.com.
Robert McGarvey again.
That's RJMegarvey at gmail.com.
Get in touch, we'll figure out away that you can help.
We need your support, we wantyour support, we thank you for
(38:03):
your support.
The CU2.0 Podcast.
Popular Podcasts
Las Culturistas with Matt Rogers and Bowen Yang
Ding dong! Join your culture consultants, Matt Rogers and Bowen Yang, on an unforgettable journey into the beating heart of CULTURE. Alongside sizzling special guests, they GET INTO the hottest pop-culture moments of the day and the formative cultural experiences that turned them into Culturistas. Produced by the Big Money Players Network and iHeartRadio.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.