StarZero All shows Try the podcast clipper →

May 28, 2014 · 21m · mad

Noah Breslow & Abhra Mitra, onDeck // Data Driven #27 // May 2014 (Hosted by FirstMark Capital)

Noah Breslow · 11m spoken Abhra Mitra · 5m spoken Tony Baer · 32s spoken Matt Turck · 12s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this FirstMark Capital presentation, OnDeck CEO Noah Breslow and Data Analytics Manager Abhra Mitra detail how OnDeck leverages big data, entity resolution, and machine learning to address the $100 billion small business lending gap. They showcase OnDeck's integrated platform, proprietary scoring model, and data science architecture followed by an audience Q&A.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →

Matt as informed peer 0.1 Guest teaching 2.1 Guest disagreement 0.0 Matt pushing back 0.0
05100:0010:0020:000:17–3:00 · Matt as informed peer 0/10 Presentation Agenda Overview Noah Breslow provides a structured overview of the presentation agenda and introduces the small business lending gap. As this is a pure presentation monologue without host participation, host scores are zero.3:00–5:13 · Matt as informed peer 0/10 Disrupting Traditional Financial Services Breslow explains how traditional banking models fail small dollar loans and details OnDeck's consolidation of origination, servicing, credit bureau, and scoring functions. This is a solo presentation monologue.5:13–8:00 · Matt as informed peer 0/10 The Small Business Digital Footprint Breslow discusses the heterogeneity of small business digital footprints compared to standardized consumer data. The segment is a solo speaker presentation.8:00–10:37 · Matt as informed peer 0/10 Personal Credit Score vs. The OnDeck Score Breslow contrasts negative personal credit indicators in consumer FICO scores with positive business cash flow metrics evaluated by OnDeck. The host does not participate.10:37–14:08 · Matt as informed peer 0/10 Data Aggregation and Model Performance Evolution Abhra Mitra takes over to present OnDeck's data aggregation platform and entity resolution techniques using a practical Mags Plumbing example. This is an uninterrupted technical presentation.14:08–16:19 · Matt as informed peer 0/10 Multi-Source Voting in Predictive Credit Models Mitra explains multi-source data voting across different business verticals like restaurants versus sparse footprint businesses like funeral homes. The segment is monologue presentation.16:19–21:15 · Matt as informed peer 1/10 Audience Q&A Session The event host facilitates Q&A while audience members ask about risk pricing, sector default rates, data monetization, and human data verification. The dynamic is collaborative and informational throughout.0:17–3:00 · Guest teaching 2/10 Presentation Agenda Overview Noah Breslow provides a structured overview of the presentation agenda and introduces the small business lending gap. As this is a pure presentation monologue without host participation, host scores are zero.3:00–5:13 · Guest teaching 2/10 Disrupting Traditional Financial Services Breslow explains how traditional banking models fail small dollar loans and details OnDeck's consolidation of origination, servicing, credit bureau, and scoring functions. This is a solo presentation monologue.5:13–8:00 · Guest teaching 2/10 The Small Business Digital Footprint Breslow discusses the heterogeneity of small business digital footprints compared to standardized consumer data. The segment is a solo speaker presentation.8:00–10:37 · Guest teaching 2/10 Personal Credit Score vs. The OnDeck Score Breslow contrasts negative personal credit indicators in consumer FICO scores with positive business cash flow metrics evaluated by OnDeck. The host does not participate.10:37–14:08 · Guest teaching 3/10 Data Aggregation and Model Performance Evolution Abhra Mitra takes over to present OnDeck's data aggregation platform and entity resolution techniques using a practical Mags Plumbing example. This is an uninterrupted technical presentation.14:08–16:19 · Guest teaching 2/10 Multi-Source Voting in Predictive Credit Models Mitra explains multi-source data voting across different business verticals like restaurants versus sparse footprint businesses like funeral homes. The segment is monologue presentation.16:19–21:15 · Guest teaching 2/10 Audience Q&A Session The event host facilitates Q&A while audience members ask about risk pricing, sector default rates, data monetization, and human data verification. The dynamic is collaborative and informational throughout.0:17–3:00 · Guest disagreement 0/10 Presentation Agenda Overview Noah Breslow provides a structured overview of the presentation agenda and introduces the small business lending gap. As this is a pure presentation monologue without host participation, host scores are zero.3:00–5:13 · Guest disagreement 0/10 Disrupting Traditional Financial Services Breslow explains how traditional banking models fail small dollar loans and details OnDeck's consolidation of origination, servicing, credit bureau, and scoring functions. This is a solo presentation monologue.5:13–8:00 · Guest disagreement 0/10 The Small Business Digital Footprint Breslow discusses the heterogeneity of small business digital footprints compared to standardized consumer data. The segment is a solo speaker presentation.8:00–10:37 · Guest disagreement 0/10 Personal Credit Score vs. The OnDeck Score Breslow contrasts negative personal credit indicators in consumer FICO scores with positive business cash flow metrics evaluated by OnDeck. The host does not participate.10:37–14:08 · Guest disagreement 0/10 Data Aggregation and Model Performance Evolution Abhra Mitra takes over to present OnDeck's data aggregation platform and entity resolution techniques using a practical Mags Plumbing example. This is an uninterrupted technical presentation.14:08–16:19 · Guest disagreement 0/10 Multi-Source Voting in Predictive Credit Models Mitra explains multi-source data voting across different business verticals like restaurants versus sparse footprint businesses like funeral homes. The segment is monologue presentation.16:19–21:15 · Guest disagreement 0/10 Audience Q&A Session The event host facilitates Q&A while audience members ask about risk pricing, sector default rates, data monetization, and human data verification. The dynamic is collaborative and informational throughout.0:17–3:00 · Matt pushing back 0/10 Presentation Agenda Overview Noah Breslow provides a structured overview of the presentation agenda and introduces the small business lending gap. As this is a pure presentation monologue without host participation, host scores are zero.3:00–5:13 · Matt pushing back 0/10 Disrupting Traditional Financial Services Breslow explains how traditional banking models fail small dollar loans and details OnDeck's consolidation of origination, servicing, credit bureau, and scoring functions. This is a solo presentation monologue.5:13–8:00 · Matt pushing back 0/10 The Small Business Digital Footprint Breslow discusses the heterogeneity of small business digital footprints compared to standardized consumer data. The segment is a solo speaker presentation.8:00–10:37 · Matt pushing back 0/10 Personal Credit Score vs. The OnDeck Score Breslow contrasts negative personal credit indicators in consumer FICO scores with positive business cash flow metrics evaluated by OnDeck. The host does not participate.10:37–14:08 · Matt pushing back 0/10 Data Aggregation and Model Performance Evolution Abhra Mitra takes over to present OnDeck's data aggregation platform and entity resolution techniques using a practical Mags Plumbing example. This is an uninterrupted technical presentation.14:08–16:19 · Matt pushing back 0/10 Multi-Source Voting in Predictive Credit Models Mitra explains multi-source data voting across different business verticals like restaurants versus sparse footprint businesses like funeral homes. The segment is monologue presentation.16:19–21:15 · Matt pushing back 0/10 Audience Q&A Session The event host facilitates Q&A while audience members ask about risk pricing, sector default rates, data monetization, and human data verification. The dynamic is collaborative and informational throughout.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%9:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%12:00 · Matt 0% · guest 100%15:00 · Matt 0% · guest 100%15:00 · Matt 0% · guest 100%18:00 · Matt 0% · guest 100%18:00 · Matt 0% · guest 100%21:00 · Matt 0% · guest 100%21:00 · Matt 0% · guest 100%
Sharpest disagreement ▶ 17:26 Declining to disclose specific loss rates

Abhra Mitra politely demurs on sharing proprietary sector loss rates when asked directly by an audience member, marking the only mild boundary-setting in an otherwise collaborative presentation.

 Hardest push from Matt ▶ 16:36 Audience question probing risk pricing and default metrics

Victoria from the Economist Intelligence Unit presses the speakers on whether they use variable interest rates based on model output and demands specific sector default data.

 Biggest teaching moment ▶ 12:10 Explaining entity resolution and fuzzy logic algorithms

Abhra Mitra delivers an educational breakdown of how OnDeck uses machine learning and human intuition to resolve inconsistent business identity records across multiple data sources.

 Matt holds his own ▶ 18:56 Audience analyst offering Google Flu comparison and business model extension

Tony Baer demonstrates domain expertise by framing OnDeck's data as an early macroeconomic indicator akin to Google Flu Trends and questioning whether they plan to launch a data product business.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Presentation Agenda Overview 0200 Noah Breslow provides a structured overview of the presentation agenda and introduces the small business lending gap. As this is a pure presentation monologue without host participation, host scores are zero.
Disrupting Traditional Financial Services 0200 Breslow explains how traditional banking models fail small dollar loans and details OnDeck's consolidation of origination, servicing, credit bureau, and scoring functions. This is a solo presentation monologue.
The Small Business Digital Footprint 0200 Breslow discusses the heterogeneity of small business digital footprints compared to standardized consumer data. The segment is a solo speaker presentation.
Personal Credit Score vs. The OnDeck Score 0200 Breslow contrasts negative personal credit indicators in consumer FICO scores with positive business cash flow metrics evaluated by OnDeck. The host does not participate.
Data Aggregation and Model Performance Evolution 0300 Abhra Mitra takes over to present OnDeck's data aggregation platform and entity resolution techniques using a practical Mags Plumbing example. This is an uninterrupted technical presentation.
Multi-Source Voting in Predictive Credit Models 0200 Mitra explains multi-source data voting across different business verticals like restaurants versus sparse footprint businesses like funeral homes. The segment is monologue presentation.
Audience Q&A Session 1200 The event host facilitates Q&A while audience members ask about risk pricing, sector default rates, data monetization, and human data verification. The dynamic is collaborative and informational throughout.

Statements from this episode (13)

Assertion Supported
OnDeck has loaned over $1 billion to US small businesses
“We've loaned over a billion dollars to small businesses all across America.”
Noah Breslow May 28, 2014 ▶ 0:18
Assertion Partly supported
Oliver Wyman study: Bank loans under $100K are almost unprofitable
“Oliver Wyman did a study where they looked at loans to small businesses underneath a 100,000 dollars, and they found that the traditional banking process, which can take about 30 days on average is almost unprofitable, ah, for the bank, ah, in most cases.”
Noah Breslow May 28, 2014 ▶ 1:35
Assertion Not checkable as stated
Breslow: US small business loan demand is roughly $250 billion
“When you look at these smaller loans and what the demand looks like in this country it's about a quarter trillion dollars of demand that exists out there for small business loans.”
Noah Breslow May 28, 2014 ▶ 1:57
Assertion Not checkable as stated
Breslow estimates $100B in unmet US small business loan demand
“We think there's another hundred billion or so of unmet demand in this country.”
Noah Breslow May 28, 2014 ▶ 2:11
Assertion Supported
Breslow: US small-dollar business loans fell after the 2008 recession
“The number of small dollar loans to small businesses that has actually fallen since the Great Recession back in 2008, 2009.”
Noah Breslow May 28, 2014 ▶ 2:42
Assertion Not checkable as stated
OnDeck tracks eight million US small businesses from creation to closure
“Essentially, there's about eight million businesses that we're focused on in this country, and we have a database that's tracking all of them from birth Through death.”
Noah Breslow May 28, 2014 ▶ 4:55
Assertion Contradicted
Breslow: Small business online banking adoption rose from 60% to 95%
“Even when we started OnDeck, only 60% of small business owners used online banking. Today that number's 95%.”
Noah Breslow May 28, 2014 ▶ 6:27
Disclosure
Breslow: OnDeck aims to create the FICO score for small businesses
“And just like the FICO score describes an individual borrower, we're looking to make the on-deck score the language that describes the credit worthiness of a main street business trying to borrow money.”
Noah Breslow May 28, 2014 ▶ 8:09
Insight
Breslow: Social media data is too noisy for assessing creditworthiness
“Social data, and Abra said this to me many moons ago is, is very, very noisy.”
Noah Breslow May 28, 2014 ▶ 9:05
Assertion Not checkable as stated
Breslow: OnDeck's typical customers average 10 years in business and $1M revenue
“We really thought that our customer would be a very, very new business that wasn't picked up by the traditional banking system. We've been sort of surprised to see that actually it's much more established business owners are using our product. 10 years in busi…”
Noah Breslow May 28, 2014 ▶ 9:28
Assertion Not checkable as stated
Mitra: OnDeck increased applicant scoring coverage from 60% to over 95%
“When we started, we could only score about 60% of our applicants. Now we can score more than 95%.”
Abhra Mitra May 28, 2014 ▶ 11:03
Assertion Not checkable as stated
Breslow: Restaurants are historically OnDeck's top lending vertical in most states
“Restaurants have been our number one vertical in most states forever.”
Noah Breslow May 28, 2014 ▶ 18:34
Disclosure
Mitra: OnDeck uses two to three human reviewers per loan application
“The second is loan applications. So there it, you know, it's typically two to three people so in the, in, in the case of, for example, industry prediction we would predict an industry, but we're not gonna automatically classify a loan applicant right off the b…”
Abhra Mitra May 28, 2014 ▶ 20:48
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more