Case study ⬩ 4 min

How ODF handled 1000s of monthly applications with their custom CRM

[ joinodf.com ]

Applications a year
15K+
Saved a year on setup
$40K
Program capacity
2x

About ODF

ODF (previously On Deck) helps ambitious founders start companies through online bootcamps. Founders from 1,000+ startups, including Traba, Cal.com, and Luma, launched their ventures through ODF, collectively raising $2B from top VCs.

System Mapping

We focus here on ODF’s application management system (excluding bootcamp operations).

FIG.1⬩ODFOS04LAYERS01⬩CUSTOMER-FACINGTOOLSWebflowWEBSITEFilloutFORMSLumaEVENTS02⬩DATAFEEDSAirtableCENTRALDATABASE03⬩INTERNALTOOLSREADANDWRITESlackNOTIFICATIONS⬩BOTStripeINVOICING⬩PAYMENTSCustomer.ioEMAILINGClayENRICHMENT⬩AISCREENING04⬩WORKFLOWSAUTOMATEZapierTrigger.dev
ODF OS

System Overview:

  • Leads apply via Fillout forms, synced with Airtable via Zapier.
  • Data from Luma events flows into Airtable.
  • Airtable is the central hub, managing 80,000+ records for applications, screening, and reporting.
  • Zapier automates connections between Airtable, Slack, Stripe, and Customer.io.
  • Trigger.dev powers a custom Slack bot.
  • Clay enriches Airtable data and simplifies screening via AI.

Before vs. After

Legacy System After Optimization
One Airtable table with 20 duplicated Zapier workflows per cohort 3-table structure (Pipeline, Contacts, Cohorts) completely remove the need for Zap duplication
2-5 days of setup per program (25 programs per year, 15,000+ applications) 1 day per program setup (from 1 week), saving $40k per year
2 full-time employees managing the system 1 part-time person (~5 hours per week) manages the entire system
Complex funnel with optional steps, leading to inconsistent tracking 2x program capacity & revenue potential without additional hires

Data restructuring

Goal: Save time
Impact: Estimated $40K/year saved in reduced setup overhead.

Airtable data structuring (one to 3 separate tables):

  • Eliminated Zap duplication for program setup.
  • Enabled automation that wasn’t possible before.

Zapier updates

Goal: Save time
Impact: Reduced setup time from 1 week to 1 day per program, saving $40K/year.

Replaced standard Zaps with webhooks & custom scripts:

  • One URL to do setup instead of manual work.
  • Switched triggers from view-based (unreliable) to button-based (predictable, interactive UX, rich error tracking).

Email params

Goal: Save time & increase conversion
Impact: Increased lead conversion rate by 5%, adding $375K/year (assuming 100 people/program, 25 programs, $3,000 per lead).

Transition from Gmail to Customer.io:

  • Dates as variables for automated follow-ups & copy changes in emails before, after and during cohorts.
  • Implemented event-based email tracking → Conversion-focused analytics.
  • Added UTM-tracked unique links → No more mismatched email-to-contact issues.

KPI tracking

Goal: Increase conversion
Impact: Improved funnel optimization, adding $250K/year in revenue by identifying and reducing drop-offs.

Analytics are the most complete analytics you can dream of for a cohort-based system and are 100% automated:

  • KPI tracking table updates every Monday.
  • Weekly cohort-based performance dashboard.
  • Funnel cycle comparisons to optimize conversions.

Custom interfaces

Goal: Reduce training & screening time
Impact: From 1 month of training → 0 training needed for 2 operators, saving $10K/year in onboarding costs

Custom interfaces for application screening instead of raw Airtable DB that required tons of team training:

  • Automated team allocation for lead screening.
  • Multi-step screening.
  • Queues for decision submission emails to save button clicks & loading time (-90%) decreasing application review by 15%.

Top-of-the-funnel

Goal: Increase conversion & generate leads
Impact: Minimum +5% lead conversion & tracking accuracy, adding at $375K/year in lead revenue.

  • Automated referral system (4 referral methods tracked), resulting in +10% referral conversion while referral represented 50% of customers.
  • Luma event-attendee mapping → Application completion impact analysis.
  • Automated Stripe payment follow-ups.
  • Clay enrichment → Fetches Fillout API data even for incomplete forms.

Custom Slack bot

Goal: Save time
Impact: Reduced manual screening coordination by 20% ($15K/year saved in operational costs).

Deep integration with Slack for notifications but also actions via a custom Slack bot we built to:

  • Extracts email data and suggests actions directly in Slack.
  • Auto-creates screening threads for interview feedback.

AI implementation

Goal: Save time
Impact: 80% faster application screening, saving $100K/year in manual labor costs.

AI research agent:

  • AI scores applications based on LinkedIn & application signals.
  • Tracks job changes on LinkedIn to measure bootcamp impact.
  • Slack-integrated AI bots extract & process email data for faster decisions.
  • Clay enrichment for people who don’t submit their apps.
  • Fetch partial application submission data even when they start but don’t submit.

Highlights

Impact

With this OS, ODF achieved remarkable results:

  • 15,000 applications are screened annually by one part-time person.
  • $1.2M per year in estimated added value through cost savings & revenue generation.

Features

The system’s greatest strengths include:

  • Simplicity: Centralized system in Airtable, pushing it to its limits resulting in an extremely simple OS for an organisation with such operational complexity
  • Data viz: Complete, automated data visualization giving in-depth insights into performance and conversion to optimize sales, marketing & ops decision-making
  • Customization: Completely custom on Airtable, with no noticeable limitations
  • Centralized Data: Clean, accurate and accessible information in Airtable
  • Team Adoption: Simple custom interfaces and actions from Slack
All posts Railblocks ⬩ Blog