StarZero All shows Try the podcast clipper →

Oct 24, 2022 · 26m · mad

Behavioral Data Creation for AI | Snowplow Co-Founder & CEO Alex Dean

Alex Dean · 21m spoken Matt Turck · 51s 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 Data Driven NYC presentation, Snowplow Co-Founder and CEO Alex Dean explains why AI and machine learning models require purposefully created behavioral data rather than low-quality data exhaust. He outlines how Snowplow's real-time data creation architecture enables enterprises to capture granular customer intent while maintaining data governance and compliance.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 3.4% of the talking time here. How this is scored →

Matt as informed peer 0.8 Guest teaching 0.4 Guest disagreement 0.2 Matt pushing back 0.2
05100:0010:0020:001:33–4:10 · Matt as informed peer 0/10 Data Apps and the Value of Better Data This is a solo presentation segment by Alex Dean introducing himself, Snowplow, and early customer use cases like Strava. The host does not participate, resulting in zero scores across all dynamics.4:10–6:57 · Matt as informed peer 0/10 Data Creation vs. Traditional Data Exhaust Workflows Alex Dean presents the conceptual difference between traditional data exhaust and intentional data creation. As this is entirely a solo monologue, no host interaction takes place.6:57–10:17 · Matt as informed peer 0/10 The Power of Behavioral Data for AI Models Alex Dean explains how behavioral data provides superior predictive signals for machine learning compared to demographic or transactional data. The host remains silent throughout the monologue.10:17–14:58 · Matt as informed peer 0/10 Primary Use Cases and Industry Applications for Snowplow Alex Dean outlines core enterprise use cases including composable CDPs, digital analytics, and retail adoption. The host does not speak in this segment.14:58–25:54 · Matt as informed peer 4/10 Commercial Platform Overview and Q&A Session Host Matt Turck joins to lead Q&A, demonstrating domain knowledge by contextualizing Snowplow's early stack within the pre-Redshift Hadoop ecosystem. The interaction is friendly and collaborative, with mild clarifying responses from the guest.1:33–4:10 · Guest teaching 0/10 Data Apps and the Value of Better Data This is a solo presentation segment by Alex Dean introducing himself, Snowplow, and early customer use cases like Strava. The host does not participate, resulting in zero scores across all dynamics.4:10–6:57 · Guest teaching 0/10 Data Creation vs. Traditional Data Exhaust Workflows Alex Dean presents the conceptual difference between traditional data exhaust and intentional data creation. As this is entirely a solo monologue, no host interaction takes place.6:57–10:17 · Guest teaching 0/10 The Power of Behavioral Data for AI Models Alex Dean explains how behavioral data provides superior predictive signals for machine learning compared to demographic or transactional data. The host remains silent throughout the monologue.10:17–14:58 · Guest teaching 0/10 Primary Use Cases and Industry Applications for Snowplow Alex Dean outlines core enterprise use cases including composable CDPs, digital analytics, and retail adoption. The host does not speak in this segment.14:58–25:54 · Guest teaching 2/10 Commercial Platform Overview and Q&A Session Host Matt Turck joins to lead Q&A, demonstrating domain knowledge by contextualizing Snowplow's early stack within the pre-Redshift Hadoop ecosystem. The interaction is friendly and collaborative, with mild clarifying responses from the guest.1:33–4:10 · Guest disagreement 0/10 Data Apps and the Value of Better Data This is a solo presentation segment by Alex Dean introducing himself, Snowplow, and early customer use cases like Strava. The host does not participate, resulting in zero scores across all dynamics.4:10–6:57 · Guest disagreement 0/10 Data Creation vs. Traditional Data Exhaust Workflows Alex Dean presents the conceptual difference between traditional data exhaust and intentional data creation. As this is entirely a solo monologue, no host interaction takes place.6:57–10:17 · Guest disagreement 0/10 The Power of Behavioral Data for AI Models Alex Dean explains how behavioral data provides superior predictive signals for machine learning compared to demographic or transactional data. The host remains silent throughout the monologue.10:17–14:58 · Guest disagreement 0/10 Primary Use Cases and Industry Applications for Snowplow Alex Dean outlines core enterprise use cases including composable CDPs, digital analytics, and retail adoption. The host does not speak in this segment.14:58–25:54 · Guest disagreement 1/10 Commercial Platform Overview and Q&A Session Host Matt Turck joins to lead Q&A, demonstrating domain knowledge by contextualizing Snowplow's early stack within the pre-Redshift Hadoop ecosystem. The interaction is friendly and collaborative, with mild clarifying responses from the guest.1:33–4:10 · Matt pushing back 0/10 Data Apps and the Value of Better Data This is a solo presentation segment by Alex Dean introducing himself, Snowplow, and early customer use cases like Strava. The host does not participate, resulting in zero scores across all dynamics.4:10–6:57 · Matt pushing back 0/10 Data Creation vs. Traditional Data Exhaust Workflows Alex Dean presents the conceptual difference between traditional data exhaust and intentional data creation. As this is entirely a solo monologue, no host interaction takes place.6:57–10:17 · Matt pushing back 0/10 The Power of Behavioral Data for AI Models Alex Dean explains how behavioral data provides superior predictive signals for machine learning compared to demographic or transactional data. The host remains silent throughout the monologue.10:17–14:58 · Matt pushing back 0/10 Primary Use Cases and Industry Applications for Snowplow Alex Dean outlines core enterprise use cases including composable CDPs, digital analytics, and retail adoption. The host does not speak in this segment.14:58–25:54 · Matt pushing back 1/10 Commercial Platform Overview and Q&A Session Host Matt Turck joins to lead Q&A, demonstrating domain knowledge by contextualizing Snowplow's early stack within the pre-Redshift Hadoop ecosystem. The interaction is friendly and collaborative, with mild clarifying responses from the guest.

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 15.2% · guest 84.8%15:00 · Matt 15.2% · guest 84.8%18:00 · Matt 13% · guest 87%18:00 · Matt 13% · guest 87%21:00 · Matt 0.3% · guest 99.7%21:00 · Matt 0.3% · guest 99.7%24:00 · Matt 2.9% · guest 97.1%24:00 · Matt 2.9% · guest 97.1%
Sharpest disagreement ▶ 20:37 Addressing pushback on behavioral data tracking

Alex Dean addresses an audience question regarding whether tracking behavior aims to manipulate users, acknowledging past pushback that the concept sounded spooky.

 Hardest push from Matt ▶ 18:18 Host probes deeper into community success metrics

Matt Turck pushes back on Alex's general point about community importance by asking for specific metrics used to evaluate success.

 Biggest teaching moment ▶ 17:25 Guest corrects host framing of early data stack

When Matt Turck frames 2012 as early adoption of the modern data stack, Alex Dean gently reframes the premise by noting it did not feel modern back then.

 Matt holds his own ▶ 19:00 Host demonstrates technical context of early data infrastructure

Matt Turck shows clear technical expertise by instantly connecting Alex's mention of running Hive on EMR to the Hadoop world.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Data Apps and the Value of Better Data 0000 This is a solo presentation segment by Alex Dean introducing himself, Snowplow, and early customer use cases like Strava. The host does not participate, resulting in zero scores across all dynamics.
Data Creation vs. Traditional Data Exhaust Workflows 0000 Alex Dean presents the conceptual difference between traditional data exhaust and intentional data creation. As this is entirely a solo monologue, no host interaction takes place.
The Power of Behavioral Data for AI Models 0000 Alex Dean explains how behavioral data provides superior predictive signals for machine learning compared to demographic or transactional data. The host remains silent throughout the monologue.
Primary Use Cases and Industry Applications for Snowplow 0000 Alex Dean outlines core enterprise use cases including composable CDPs, digital analytics, and retail adoption. The host does not speak in this segment.
Commercial Platform Overview and Q&A Session 4211 Host Matt Turck joins to lead Q&A, demonstrating domain knowledge by contextualizing Snowplow's early stack within the pre-Redshift Hadoop ecosystem. The interaction is friendly and collaborative, with mild clarifying responses from the guest.

Statements from this episode (10)

Assertion Not checkable as stated
Alex Dean: Strava generates four billion events per day using Snowplow
“Strava's been a Snowplow customer for some time, and they generate four billion events per day with Snowplow, and that drives into Snowflake, and then they build all sorts of cool data apps on the other side of it, including a lot of the Strava UI and a lot of…”
Alex Dean Oct 24, 2022 ▶ 1:51
Insight
Alex Dean: Data exhaust from SaaS tools is poorly suited for AI models
“Increasingly people are finding that this data exhaust is not well suited to AI, to advanced ML use cases.”
Alex Dean Oct 24, 2022 ▶ 3:10
Insight
Alex Dean: Schematized data creation accelerates feature engineering and model training
“Because you're deliberately creating this data, because you're creating it in a very schematized way, it's much quicker to get to the feature Engineering, it's much quicker to get to the model training.”
Alex Dean Oct 24, 2022 ▶ 4:51
Insight
Alex Dean: Sentence-like syntax makes behavioral data highly predictive for ML
“If you describe behavior, human behavior in that way, you get to very, very granular descriptions of what's happened in the past, and those are highly, highly predictive for machine learning.”
Alex Dean Oct 24, 2022 ▶ 9:14
Assertion Not checkable as stated
Dean: Thousands of organizations use Snowplow's open-source technology
“So we've got thousands of organizations using our open source technology.”
Alex Dean Oct 24, 2022 ▶ 10:22
Assertion Not checkable as stated
Dean: COVID-19 drove widespread retail adoption of Snowplow's platform
“Before COVID there was not very much adoption of snowplow in retail. And then basically what happened is through COVID these big retailers in the UK, US and beyond really started building out data teams, really started taking their kind of digital experiences …”
Alex Dean Oct 24, 2022 ▶ 14:14
Disclosure
Alex Dean: Snowplow bootstrapped until 2019
“And so we basically built a team to do that and charged them for it and bootstrapped up till 2019.”
Alex Dean Oct 24, 2022 ▶ 17:19
Insight
Alex Dean: Early-stage B2B software adoption happens city by city
“I think a lot of tech is adopted almost city by city, and so I think that one of the things we got right in the early years was focusing on a few cities like New York, Berlin, London, to really grow the scene there, and then you get positive word of mouth.”
Alex Dean Oct 24, 2022 ▶ 18:00
Opinion
Alex Dean: Google Analytics is poorly suited outside retail and publishing
“Google Analytics, for example, it's pretty decent if you're a news publisher. It's pretty good if you're kind of a standard retailer. Once you go outside of those you know, it, it's much less kind of tailored to those other industries.”
Alex Dean Oct 24, 2022 ▶ 23:24
Insight
Alex Dean: Behavioral data predicts form fraud better than demographic data
“If you're filling out a form fraudulently, actually, your behaviors are super different from if you're filling it out genuinely. So like, you know, a fraudster will probably copy-paste their name into a box. It's not really their name. And so that signal's bee…”
Alex Dean Oct 24, 2022 ▶ 25:24
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