The Growth Org Triad: People & Process, Strategy, Data
A growth org only works when all three of people/process, strategy and data are healthy at once.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 7
- Confidence
- 93%
Asked what matters in building a growth team, Williams buckets everything into three interdependent topics: people and process, strategy, and data. All three must be working for the org to be effective — strong people with no loop strategy chase ideas, a great strategy with untrustworthy data can't be executed, and rich data with a broken process produces learnings that gather dust. He is explicit that growth teams must work differently from core R&D teams, and that handing growth to an existing team without changing how they work is a common trap.
Origin
Ben Williams (VP of Product, Snyk), synthesizing what he saw work and fail while formalizing Snyk's Developer Growth Group. He credits Reforge for team education and Brian Balfour's HubSpot-era writing for the learnings ceremony.
Core principles
- 01Balanced, cross-functional teams: the ability to generate great ideas AND to execute them.
- 02Potential is unbounded; situational factors — role, dynamics, mindset — cap it. Leaders must surface those.
- 03Growth process exists to facilitate a rapid learning cadence and to socialize the learnings.
- 04Experimentation delivers learnings, not outcomes; learnings pave the path to outcomes.
- 05Invest in data infrastructure early — data takes calendar time to accrue.
- 06Start simple with experimentation; sophistication before fluency is a recipe for mistakes.
How to run it
- 1
Build truly cross-functional growth teams
Snyk resolved the friction between R&D and growth marketing by making each growth team genuinely cross-functional and aligned on common objectives and KPIs. Each team has engineers, an engineering manager, a PM, a designer, a growth marketer, and decision science support.
Pro tip Embedding a growth marketer in the product team is uncommon and, Williams argues, a large missed opportunity: it widens both the idea palette and the execution toolbox, letting the team run more parallel threads.
Watch out Misaligned incentives across functions produce great ideas that never ship, and a lot of friction and frustration.
- 2
Select for growth-fit, especially on internal moves
Ask of each person: can this person do their best work in this environment? Devs who thrive in growth move fast, iterate for measurable impact, are unattached to their work, embrace imperfection, happily discard their code and ideas, and stay close to users. Brilliant engineers motivated by deep technical challenges who love the process as much as the outcome often struggle.
Pro tip External hiring tests for this explicitly; internal moves are where the mismatch gets missed.
Watch out A poor growth fit is not a poor human. Reframe it as environment, not talent.
- 3
Equip the team with common vocabulary and skills
The ideal state: every growth team member shares vocabulary, is comfortable with the growth process, and can work with data and the experimentation platform. Snyk uses Reforge plus internal programs to align and level up.
Pro tip Start simple and go deeper as the team builds experience — do not introduce multivariate testing or sequential sampling to a team still dipping its toes in.
- 4
Align the chain: execution → growth strategy → company direction
Execution aligns to an evolving growth strategy; the growth strategy aligns to and influences where the company is going, has hooks into product strategy, and overlaps on KPIs so growth and core product teams swim in the same direction. Skills in the team must match the current strategic focus (a pricing-and-packaging expert is the wrong hire when the focus is acquisition and activation).
Pro tip Plan ahead for the skills you'll need as the growth focus inevitably shifts.
Watch out Everyone must be able to answer why they are there and why their work matters.
- 5
Install a distinct growth process built for a rapid learning cadence
Document growth processes, practices and working cadences that differ from core R&D. The single most important property is a rapid learning cadence plus the means to socialize learnings at the right place and time so they can be leveraged elsewhere.
Pro tip You know it is working when learnings are shared enthusiastically, ideas arrive as well-crafted data-based hypotheses from everyone, and a wide variety of people (not just the PM or tech lead) own experiments end to end.
Watch out Giving growth responsibility to an existing team without changing how they work is a common trap.
- 6
Make behavioral data trustworthy before you scale
Snyk's problem was never too little data — it collected everything — but too little intentional, behavior-specific data, which made it hard to trust. The fix: event tracking plans, plus CI tests that check the instrumented code conforms to schema.
Pro tip Invest early: it takes calendar time to accrue enough data to define activation metrics, run regressions, and reason about retention.
Watch out Untrustworthy data quietly poisons every decision downstream of it.
- 7
Judge the growth org on how far it levels up everyone else
Beyond KPI impact, measure whether growth enables the rest of the org: Snyk built a paved road for behavioral analytics and experimentation, coached other teams onto it, ran internal education on data-driven development, built metric-design tooling, and shipped platform services (e.g. contextual onboarding) usable anywhere in the product.
Pro tip Insights from growth often have utility far beyond growth — but people cannot use what you haven't shared with them.
In the wild
A growth marketer inside Snyk's acquisition team ran lightweight experimentation on the website, creating an SEO-optimized page that performed strongly on both traffic and conversion — and required zero engineering resource, so it shipped while engineers worked on other things. On the back of that success the team built a sidecar product letting users try Snyk without signing up, by pasting code in for an instant scan.
→ Visitor traffic rose significantly; signup rate dipped slightly as expected, but new users had a big bump and much higher intent, which showed up as increased activation rates.
On joining, Williams found Snyk collected absolutely everything yet lacked intentional, behavior-specific data — which made the data hard to trust. The team built event tracking plans and added CI checks that test the instrumented code's conformance to schema.
→ Absolute confidence in the behavioral data, which later enabled the ML-driven activation habit-moment analysis and the product-led sales scoring model.
Common mistakes
Bolting growth onto an existing team unchanged
Growth teams need different cadences, ownership models and processes than core product teams. Assuming otherwise and simply reassigning responsibility is a common trap that yields neither speed nor learning.
Letting learnings gather dust
Without a process to surface learnings in the right place at the right time, they go unused — and then the experiment was pointless. The whole point of experimentation is generating learnings the organization can leverage.
Over-engineering experimentation on day one
Introducing multivariate testing or sequential sampling to a team still learning the basics is a recipe for a lot of mistakes.
Is it for you?
Best for
VPs and heads of growth standing up or repairing a growth org inside a company that already has product-market fit and a bottom-up motion.
Not ideal for
Sub-10-person startups still hunting for retention — this is an operating model for scale, not a substitute for PMF.
From the transcript
“I think I can broadly bucket things here into maybe three main topics so the first will be people and process the second would be…”
“the growth teams were truly cross-functional in nature with everyone in each team aligned around common objectives and kpis every team has Engineers an engineering…”
“it's really important to try to answer the question can this person do their best work in this environment”
“what I've found I think most singularly most important in a growth process is facilitating a rapid learning Cadence”
“the problem I identified very early was that we didn't have enough behavioral specific data and we weren't intentional enough about the data that we…”
“we invested in tooling and processes for building out event tracking plans and now we test conformance to schema of the instrumented code in our…”
From the episode
How Snyk built a product-led growth juggernaut
Ben Williams (VP of Product at Snyk)