Lenny's PodcastThe ultimate guide to Martech | Austin Hay (Reforge, Ramp, Runway)
CHAPTERS
- 0:00 – 0:52
Cold open: The shift from deterministic to probabilistic attribution
Austin kicks off with a look at how marketing measurement has changed—especially after the end of “deterministic matching.” He frames the modern challenge as making confident decisions with incomplete, probabilistic data.
- •2010–2020 as the “golden years” of deterministic matching (e.g., IDFA-based tracking)
- •Why it’s now harder to tie ad spend to user outcomes with precision
- •Ad networks are getting more sophisticated while measurement gets murkier
- •How probabilistic models can extrapolate from partial signal (e.g., 30% observed → 100% inferred)
- 0:52 – 4:21
Who Austin Hay is and what this episode will cover
Lenny introduces Austin’s background across Branch, mParticle, Runway, and Ramp, and positions him as a leading MarTech thinker. The episode roadmap is set: definitions, org design, tools, attribution, hiring, and frameworks.
- •Austin’s experience advising companies like Notion, Airbnb, Walmart, and more
- •His current role leading Marketing Technology at Ramp
- •Why MarTech is a missing piece in many growth/product conversations
- •What listeners will walk away with: tools, structures, interview questions, frameworks
- 4:21 – 6:17
What MarTech actually is: a systems-focused product mindset
Austin defines MarTech as a cross-functional discipline at the intersection of marketing, product, engineering, and growth. He emphasizes that MarTech is less about a tool list and more about owning the system/platform that enables growth.
- •MarTech spans people/process + systems/platforms
- •MarTech as a “product manager for internal marketing systems”
- •Third-party tools plus first-party solutions built on top
- •How the definition changes with company size and stage
- 6:17 – 10:24
MarTech vs growth roles: when ‘the village’ stops scaling
Austin explains how early-stage companies often distribute MarTech responsibilities across growth/engineering, but this breaks as complexity grows. The tipping point is when someone must centrally own schemas, data flow, tooling decisions, and risk.
- •At small startups, growth roles often “are” MarTech by necessity
- •Around ~100–150 employees, centralized ownership becomes critical
- •Responsibilities expand: schema/data flow, tool governance, procurement/legal risk
- •Different homes for MarTech: product ops, IT, standalone MarTech, marketing products
- 10:24 – 14:03
Signals you need a MarTech hire: pain, complexity, and tool risk
The strongest hiring trigger is operational pain: teams can’t move because nobody understands how the stack works or changing tools feels too risky. Austin highlights the hidden leverage in architecture decisions and contract management.
- •Common pattern: one overwhelmed tool-keeper (often an engineer) becomes a bottleneck
- •MarTech work includes standing up tools and building ‘the last 10%’ on top
- •“Build and buy” as the scalable reality (vs build vs buy)
- •Contract scrutiny becomes a major lever as spend and scale increase
- 14:03 – 21:15
Where MarTech should live: B2C vs B2B, centralized vs decentralized
Austin provides an org-design lens: business model (B2C/B2B/B2B2C) plus centralized vs decentralized ownership. He shares why decentralized models often create more fragmentation—unless you’re at massive scale.
- •B2C centralized: MarTech typically serves growth/marketing as the primary customer
- •B2C decentralized: ‘ops people everywhere’ often increases system sprawl
- •B2B/B2B2C is messier; org placement depends on customer, resourcing, leadership profile
- •Ramp’s own evolution: centralized → decentralized → centralized again
- 21:15 – 28:38
Day-to-day MarTech: governance + architecture + persuasion
Austin breaks the role into two halves: high-leverage administration (permissions, PII workflows, access controls) and the fun part—designing a future-state system and getting the org to invest in it. Much of the work is influencing without direct control of resources.
- •Governance tasks prevent costly mistakes (permissions, compliance, access controls)
- •Automation of admin workflows as a leverage multiplier
- •Long-horizon stack planning (1–2 years), with financial and resourcing models
- •MarTech as a cross-functional ‘quarterback’—selling and persuading to get work done
- 28:38 – 31:14
How MarTech should be goaled: growth outcomes + tooling efficiency
Austin outlines different ways to set goals for MarTech: align directly with the teams you support (e.g., CAC goals) and/or own efficiency and cost goals. He also describes discrete “capability goals” like migrations and platform upgrades.
- •Option 1: tie MarTech goals to UA/growth goals (CAC, growth targets)
- •Option 2: track tooling cost efficiency over time (cost per user/seat)
- •Capability OKRs: migrations (e.g., email platform upgrades) without revenue disruption
- •Balancing future-proofing with not over-investing too early
- 31:14 – 35:40
Tooling, B2C ‘then vs now’: CDPs to warehouses and reverse ETL
Austin walks through how the B2C stack evolved from CDP-centered architectures (2016–2020) to warehouse-centered architectures as storage and modeling got cheaper. Reverse ETL enables “build your own CDP” patterns and introduces new governance decisions.
- •Classic CDP promise: one SDK, many downstream integrations
- •Warehouse-first shift as Snowflake/DBT become more accessible
- •Reverse ETL as the mechanism to activate modeled warehouse data
- •Need a philosophy for when to move data from ingestion tools vs the warehouse
- 35:40 – 36:35
Reverse ETL landscape: capability, not just a category
Austin clarifies what reverse ETL means and names key vendors across both CDPs and standalone products. He reiterates the principle that tools should be chosen for problem fit, not trendiness.
- •Reverse ETL = moving data from the warehouse into operational tools
- •Examples inside CDPs: Segment, mParticle, RudderStack
- •Standalone reverse ETL players: Census, Hightouch
- •Choosing tools based on the problem you’re solving
- 36:35 – 41:34
B2B and B2B2C stacks: Salesforce gravity and object-model conflicts
Austin explains the enduring reality that many B2B systems revolve around Salesforce, with layers of tools bolted on. B2B2C adds consumer-style event/user tracking plus CRM entity complexity, making identity and data mapping particularly hard.
- •Early B2B stack blueprint: lead capture → Salesforce → outbound (e.g., Outreach) → enrichment
- •Salesforce as the center of gravity in B2B stacks
- •B2B2C complexity: tying users/events to companies/entities at the right moment
- •HubSpot + Salesforce coexistence can create ‘data mapping hell’ without careful architecture
- 41:34 – 50:48
Attribution: future-proofing from first/last touch to MTA
Austin shares practical guidance for setting up attribution so you aren’t stuck later when you want multi-touch attribution. The core is capturing referrers, UTMs, and ad IDs correctly—and storing both first-touch and last-touch values at the user and event level.
- •First-touch vs last-touch vs multi-touch attribution and why teams fight over credit
- •Collect referrer URL + UTMs + ad network IDs (e.g., GCLID, Meta parameters) from day one
- •Persist first and last touch locally (and update last touch over time)
- •Record attribution on both the user profile and events to enable later modeling
- 50:48 – 55:26
Emerging channels and measurement reality: Threads, Reddit, and ATT fallout
Austin highlights platforms he’s watching (Threads, Reddit) and connects them to the broader trend: measurement is getting harder, especially in mobile. He argues MarTech and marketers must become fluent in probabilistic thinking and decision-making.
- •Threads: advertising API and cross-app attribution questions inside Meta’s ecosystem
- •Reddit: conversion API and more native-feeling ad formats
- •ATT/iOS changes: deterministic app attribution is largely gone
- •Growing importance of probabilistic attribution and modeling for decision-making
- 55:26 – 57:42
MMM vs MTA: when modern modeling helps (and when it doesn’t)
Austin cautions that many teams jump to MMM without having the data maturity or clear decision path to benefit from it. He recommends focusing on stronger MTA and pragmatic experimentation (e.g., geo tests) before treating MMM as a cure-all.
- •MMM basics and how it differs from MTA in purpose and requirements
- •Why most businesses aren’t ready for MMM despite the hype
- •Operating with uncertainty: broad-strokes performance + expected organic leakage
- •Experiment options: geo-based tests (e.g., billboards) and the coordination burden
- 57:42 – 1:04:49
Hiring MarTech: traits, interview prompts, and avoiding false signals
Austin emphasizes intellectual curiosity and engineering scrappiness as core traits, since tools and ecosystems change constantly. He shares interview questions that reveal systems thinking and tool-agnostic problem solving, plus how to interpret ‘false flags.’
- •Look for: curiosity + ability to learn tools quickly; ‘scrappy’ coding/technical fluency
- •Great question: “How did you prepare for this interview?” to reveal process and thinking
- •Scenario prompt to detect tool bias vs problem-first reasoning
- •False flags: resume gaps and school pedigree often mislead more than they help
- 1:04:49 – 1:13:37
Favorite frameworks: PPS, build-and-buy, and ‘thinking gray’
Austin shares the frameworks he leans on to avoid jumping prematurely to tools and to make better decisions under uncertainty. He closes with “thinking gray,” a leadership principle about delaying decisions until you truly must commit.
- •“Tools solve problems” as a guiding principle
- •PPS framework: Problem → People → System (don’t start with the tool)
- •Build-and-buy vs build-vs-buy for faster, more collaborative outcomes
- •Thinking gray: defer binary decisions to improve decision quality and reduce bias
- 1:13:37 – 1:24:36
Lightning round + Austin’s ‘golden stack’ + drones and closing
The episode wraps with rapid-fire recommendations (books, shows, interview questions, products), then Austin delivers the promised ‘golden stack’ for B2C and B2B. A personal detour into drone piloting closes the conversation, plus where to find Austin and his upcoming Reforge course.
- •Lightning round: books, shows, favorite interview question, Cal.com
- •Golden B2C stack example: Amplitude, Customer.io→Braze, Snowflake, Hightouch, AppsFlyer
- •Golden B2B adjustments: Branch for web attribution, Salesforce-centric activation
- •Drone pilot story: trying (and failing) to fly legally near DC’s restricted airspace
