StarZero All shows Try the podcast clipper →

Mar 18, 2016 · 22m · mad

Using AI to Predict the Performance of Text // Kieran Snyder, Textio (Data Driven NYC / FirstMark)

Kieran Snyder · 20m spoken Matt Turck · 14s 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, Textio CEO Kieran Snyder explains how machine learning and natural language processing can predict text performance before publication to significantly improve recruiting outcomes and workplace diversity. Through live demonstrations and enterprise case studies, she illustrates how real-time linguistic guidance transforms corporate writing into an effective driver for talent acquisition.

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 1.2% of the talking time here. How this is scored →

Matt as informed peer 0.4 Guest teaching 6.0 Guest disagreement 0.0 Matt pushing back 0.2
05100:0010:0020:000:15–2:39 · Matt as informed peer 0/10 The Core Premise of Predicting Text Performance Kieran presents a monologue detailing her natural language processing background and Textio's premise of predicting text performance before publication. The host does not speak in this segment.2:39–6:35 · Matt as informed peer 0/10 Why Candidate Communication Matters in Recruitment Kieran explains how job posting language directly influences candidate engagement and details how Textio bootstrapped its early models using publicly observable signals from job boards like AngelList and Indeed. The host remains silent.6:35–14:42 · Matt as informed peer 0/10 Live Product Demonstration: Analyzing a Job Posting in Textio Kieran walks through a live demonstration of Textio analyzing a JPMorgan job ad, explaining complex data-driven findings such as how bullet point density affects male vs female applicant ratios. The host does not speak during the demonstration.14:42–18:10 · Matt as informed peer 0/10 Business Outcomes, Diversity Lift, and Corporate Jargon Kieran highlights business outcomes such as reduced time-to-hire and explains how corporate jargon disproportionately deters underrepresented applicant groups. The host is not involved in this presentation segment.18:10–22:54 · Matt as informed peer 2/10 Q&A Session: Data Feedback Loops and Retraining Models Matt Turck asks a brief, direct question about how Textio gets ground-truth outcome feedback to validate model predictions. Kieran elaborates on ATS integrations and audience questions regarding retraining models as language evolves.0:15–2:39 · Guest teaching 5/10 The Core Premise of Predicting Text Performance Kieran presents a monologue detailing her natural language processing background and Textio's premise of predicting text performance before publication. The host does not speak in this segment.2:39–6:35 · Guest teaching 6/10 Why Candidate Communication Matters in Recruitment Kieran explains how job posting language directly influences candidate engagement and details how Textio bootstrapped its early models using publicly observable signals from job boards like AngelList and Indeed. The host remains silent.6:35–14:42 · Guest teaching 7/10 Live Product Demonstration: Analyzing a Job Posting in Textio Kieran walks through a live demonstration of Textio analyzing a JPMorgan job ad, explaining complex data-driven findings such as how bullet point density affects male vs female applicant ratios. The host does not speak during the demonstration.14:42–18:10 · Guest teaching 6/10 Business Outcomes, Diversity Lift, and Corporate Jargon Kieran highlights business outcomes such as reduced time-to-hire and explains how corporate jargon disproportionately deters underrepresented applicant groups. The host is not involved in this presentation segment.18:10–22:54 · Guest teaching 6/10 Q&A Session: Data Feedback Loops and Retraining Models Matt Turck asks a brief, direct question about how Textio gets ground-truth outcome feedback to validate model predictions. Kieran elaborates on ATS integrations and audience questions regarding retraining models as language evolves.0:15–2:39 · Guest disagreement 0/10 The Core Premise of Predicting Text Performance Kieran presents a monologue detailing her natural language processing background and Textio's premise of predicting text performance before publication. The host does not speak in this segment.2:39–6:35 · Guest disagreement 0/10 Why Candidate Communication Matters in Recruitment Kieran explains how job posting language directly influences candidate engagement and details how Textio bootstrapped its early models using publicly observable signals from job boards like AngelList and Indeed. The host remains silent.6:35–14:42 · Guest disagreement 0/10 Live Product Demonstration: Analyzing a Job Posting in Textio Kieran walks through a live demonstration of Textio analyzing a JPMorgan job ad, explaining complex data-driven findings such as how bullet point density affects male vs female applicant ratios. The host does not speak during the demonstration.14:42–18:10 · Guest disagreement 0/10 Business Outcomes, Diversity Lift, and Corporate Jargon Kieran highlights business outcomes such as reduced time-to-hire and explains how corporate jargon disproportionately deters underrepresented applicant groups. The host is not involved in this presentation segment.18:10–22:54 · Guest disagreement 0/10 Q&A Session: Data Feedback Loops and Retraining Models Matt Turck asks a brief, direct question about how Textio gets ground-truth outcome feedback to validate model predictions. Kieran elaborates on ATS integrations and audience questions regarding retraining models as language evolves.0:15–2:39 · Matt pushing back 0/10 The Core Premise of Predicting Text Performance Kieran presents a monologue detailing her natural language processing background and Textio's premise of predicting text performance before publication. The host does not speak in this segment.2:39–6:35 · Matt pushing back 0/10 Why Candidate Communication Matters in Recruitment Kieran explains how job posting language directly influences candidate engagement and details how Textio bootstrapped its early models using publicly observable signals from job boards like AngelList and Indeed. The host remains silent.6:35–14:42 · Matt pushing back 0/10 Live Product Demonstration: Analyzing a Job Posting in Textio Kieran walks through a live demonstration of Textio analyzing a JPMorgan job ad, explaining complex data-driven findings such as how bullet point density affects male vs female applicant ratios. The host does not speak during the demonstration.14:42–18:10 · Matt pushing back 0/10 Business Outcomes, Diversity Lift, and Corporate Jargon Kieran highlights business outcomes such as reduced time-to-hire and explains how corporate jargon disproportionately deters underrepresented applicant groups. The host is not involved in this presentation segment.18:10–22:54 · Matt pushing back 1/10 Q&A Session: Data Feedback Loops and Retraining Models Matt Turck asks a brief, direct question about how Textio gets ground-truth outcome feedback to validate model predictions. Kieran elaborates on ATS integrations and audience questions regarding retraining models as language evolves.

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 7.2% · guest 92.8%18:00 · Matt 7.2% · guest 92.8%21:00 · Matt 3.3% · guest 96.7%21:00 · Matt 3.3% · guest 96.7%
Sharpest disagreement ▶ 20:37 Rejecting Auto-Generation Framing

Kieran firmly reframes the product philosophy away from fully automated text generation, emphasizing that users insist on authorial autonomy.

 Hardest push from Matt ▶ 18:18 Asking for Feedback Loop Proof

Matt pushes Kieran to explain how Textio empirically verifies that its predictive suggestions produce real-world hiring results.

 Biggest teaching moment ▶ 11:15 Formatting Impact on Gender Demographics

Kieran reveals non-intuitive statistical findings showing how subtle shifts in bullet point percentage drastically affect female and male application rates.

 Matt holds his own ▶ 18:18 Targeting Practical Efficacy

Matt zeroes in on the core data problem by asking how Textio acquires the feedback loop needed to know its score predictions are correct.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
The Core Premise of Predicting Text Performance 0500 Kieran presents a monologue detailing her natural language processing background and Textio's premise of predicting text performance before publication. The host does not speak in this segment.
Why Candidate Communication Matters in Recruitment 0600 Kieran explains how job posting language directly influences candidate engagement and details how Textio bootstrapped its early models using publicly observable signals from job boards like AngelList and Indeed. The host remains silent.
Live Product Demonstration: Analyzing a Job Posting in Textio 0700 Kieran walks through a live demonstration of Textio analyzing a JPMorgan job ad, explaining complex data-driven findings such as how bullet point density affects male vs female applicant ratios. The host does not speak during the demonstration.
Business Outcomes, Diversity Lift, and Corporate Jargon 0600 Kieran highlights business outcomes such as reduced time-to-hire and explains how corporate jargon disproportionately deters underrepresented applicant groups. The host is not involved in this presentation segment.
Q&A Session: Data Feedback Loops and Retraining Models 2601 Matt Turck asks a brief, direct question about how Textio gets ground-truth outcome feedback to validate model predictions. Kieran elaborates on ATS integrations and audience questions regarding retraining models as language evolves.

Statements from this episode (9)

Insight
Machine learning models often predict outcomes without explaining why they happen
“The models that you can create can often tell you what might happen, but they can't always tell you why.”
Kieran Snyder Mar 18, 2016 ▶ 1:04
Insight
Text is the primary daily volume output of almost every business
“Whatever business you are in, whether you're making software, or you're making hamburgers, or you're making playground equipment, The thing you're actually making the most of every day at work is text, ah, almost certainly.”
Kieran Snyder Mar 18, 2016 ▶ 1:18
Assertion Not checkable as stated
Job seekers scan listings for only a couple seconds before deciding
“Turns out people scan a listing for only a couple seconds before deciding whether to engage.”
Kieran Snyder Mar 18, 2016 ▶ 3:07
Assertion Not checkable as stated
Over a third of recruited candidates abandon applications after reading the listing
“Over a third of people that you're reaching out to who are looking for jobs walk away when they see your job listing.”
Kieran Snyder Mar 18, 2016 ▶ 3:16
Disclosure
Textio collected 15 million job listings tagged with performance outcome metrics
“At this point we have about fifteen million job listings across industries and geographies that are tagged with, in many cases, very rich outcomes.”
Kieran Snyder Mar 18, 2016 ▶ 4:39
Insight
Diversity of an applicant pool strongly predicts how quickly jobs fill
“It turns out that the percentage of underrepresented groups who apply to a job is a very good predictor of how quickly the role will fill”
Kieran Snyder Mar 18, 2016 ▶ 9:07
Assertion Not checkable as stated
Job listings with over 50% bullet points significantly reduce female applicants
“If you go above 50% bulleted content in a listing, you quickly reduce the proportion of women who are likely to apply for the job, statistically.”
Kieran Snyder Mar 18, 2016 ▶ 12:09
Assertion Partly supported
The average corporate cost per hire is approximately $2,000
“Considering your average cost per hire is about 2000 dollars, you're saving a fairly substantial amount”
Kieran Snyder Mar 18, 2016 ▶ 17:03
Assertion Supported
Corporate jargon in job postings disproportionately deters applicants of color
“Like some of that corporate jargon, the words like synergy and stakeholders and ROI. It turns out everybody hates them, but underrepresented groups, especially people of color, hate them even more. So nobody is as likely to apply when you include that language…”
Kieran Snyder Mar 18, 2016 ▶ 17:37
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