Most PLG companies build their tool stack too fast. They buy a product analytics platform before any events are tracked, or they try to integrate HubSpot while the event definition isn't even nailed down. The result: an expensive, fragmented stack that nobody uses well. This article describes which tools you need per growth stage, in what order to implement them, and which pitfalls that order prevents.
The four layers of a PLG tool stack
A working PLG stack isn't a collection of loose tools — it's a layered architecture where each layer builds on the one below it. The four layers:
- Event tracking (foundation): Which user actions do you capture, and which pipeline carries them into the rest of the stack?
- Product analytics (analysis): Where do users drop off, which features drive conversion, what differentiates activated users from non-activated ones?
- CRM (commercial memory): Which product data is visible to sales, and how are PQL triggers translated into action?
- Automation (orchestration): Which workflows connect the layers, and how do you scale PQL follow-up without manual work?
The order is deliberate. Anyone who starts with automation while tracking is still broken is automating noise. Anyone who starts with product analytics on top of a dirty event layer draws conclusions from incomplete data. The stack works from the bottom up.
Layer 1: Event tracking
Event tracking is the infrastructure layer that captures every user action and forwards it to downstream tools. This isn't about pageviews — it's about meaningful actions: a user creating a report, inviting a teammate, opening a paid feature, or logging in for the fifth time this week.
Segment: the standard for mid-market and enterprise
Segment is the most-used event tracking platform in the PLG world, and not without reason. Segment acts as a central event bus: you implement one SDK (web, mobile, or server-side), and Segment handles distribution to every downstream tool — analytics, CRM, data warehouse, email platforms.
The strength of Segment lies in the separation between event production and event consumption. If you want to move from Mixpanel to Amplitude, you don't re-instrument anything in your product — you connect a new destination in Segment. That makes Segment increasingly valuable as the stack grows.
Segment isn't free. Costs scale with Monthly Tracked Users (MTUs). For early-stage startups that can hurt. Segment offers a free tier up to 1,000 MTUs, but anyone above that quickly pays hundreds of euros per month.
RudderStack: the open-source variant
RudderStack is functionally comparable to Segment but offers a self-hosted option. For PLG companies with strong technical teams and privacy-sensitive customers (think enterprise compliance or EU data residency requirements), RudderStack is the better choice. The developer experience is slightly less polished than Segment's, but the cost structure is significantly better at high volumes.
PostHog as an early-stage alternative
For seed-stage companies that don't yet have a dedicated data engineer, PostHog is a practical alternative. PostHog combines event tracking, product analytics, session replay, and feature flags in one platform. You give up some flexibility and downstream integrations, but you get a fully working analysis layer without Segment-level costs. Once the team grows and the stack expands, moving to Segment/RudderStack plus a dedicated analytics tool is the logical next step.
Layer 2: Product analytics
Product analytics answers the questions no regular web analytics tool can: which users hit the activation moment, how long is time-to-value, which feature combinations predict conversion to paid, and where do this month's cohorts drop off earlier than three-month-old cohorts?
Mixpanel: the standard for PLG companies
Mixpanel is the most-used product analytics platform among mid-market PLG companies. Its strength is the combination of event-based funnels, retention curves per cohort, and real-time segmentation. A Mixpanel funnel shows you exactly what percentage of users moves from signup to first meaningful event, and within what time window.
The three analyses you want to set up the moment events start flowing in:
- Activation funnel: From signup through email verification to first meaningful action. Each step shows the drop-off number.
- Retention curve: What percentage of users who activated in week 0 returns in week 1, week 4, and week 8?
- Feature correlation with paid conversion: Which actions by free users correlate most strongly with conversion to a paid subscription?
Amplitude: the enterprise choice
Amplitude offers similar functionality to Mixpanel but has a stronger feature set for larger organizations: better collaboration, stronger governance options, and deeper integrations with data warehouses (Snowflake, BigQuery). For PLG companies already showing signs of moving fast toward enterprise accounts, Amplitude is the more logical choice over Mixpanel.
Amplitude's Compass functionality is particularly useful: it automatically identifies which feature combinations predict the highest retention, which dramatically accelerates the activation-moment investigation.
Pendo: when in-app guidance is the priority
Pendo is positioned differently from Mixpanel and Amplitude. On top of analytics, Pendo provides a full in-app guidance system: tooltips, walkthroughs, NPS surveys, and feature announcements, all without code changes. For PLG companies where the product team owns onboarding optimization and iteration speed is critical, Pendo's combination of analytics + in-app guidance is attractive.
The downside: Pendo is significantly more expensive than Mixpanel. It's a tool you justify when the product team wants to run onboarding experiments independently without needing engineering time every cycle.
Layer 3: CRM
The CRM is the layer where product data becomes commercial. The best product analytics in the world is worthless to sales if that data isn't available at the moment a rep calls a prospect or a CSM reviews an account.
HubSpot: the recommended choice for PLG companies
HubSpot is the best-fit CRM for the majority of mid-market PLG companies, and there are concrete reasons for that.
First: the integrations. HubSpot has native Segment integration, native Amplitude integration, and a well-documented Contacts API. You can load product data three ways: via Segment as a destination, via direct API calls from your backend, or via an integration partner like Zapier or Make.
Second: the workflow engine. HubSpot's workflow tool lets you turn PQL triggers into automated actions — create a deal, create a task for an AE, send an internal Slack notification, send an engagement email. All of it without code, which makes it accessible to RevOps teams without dedicated development.
Third: the lifecycle stage architecture. HubSpot's built-in lifecycle stages (Subscriber, Lead, MQL, SQL, Opportunity, Customer) are extensible with custom stages like Free User, Activated User, and PQL. For a detailed walkthrough of how to integrate product data into HubSpot, see the article on product data in HubSpot.
HubSpot's limitation: above a certain level of complexity (hundreds of custom properties, intricate scoring logic, high event volumes) HubSpot becomes slower and less flexible. Companies moving toward enterprise scale that evaluate Salesforce typically do so at ARR above €5-10 million and with a dedicated RevOps team that can absorb the Salesforce overhead.
Attio: for technical early-stage teams
Attio is a modern CRM option popular with technical founders and early-stage PLG companies. Its programmable data model — where you define your own objects, relations, and views — makes it more flexible than HubSpot for teams that treat their CRM as a database. The workflow automation is less mature than HubSpot's, but the developer experience is superior. If you have a technical team that wants to drive CRM automation via code instead of via a no-code workflow editor, Attio is worth considering.
Layer 4: Automation and orchestration
The automation layer connects the three previous layers and ensures signals are automatically turned into actions — without a RevOps person having to process each trigger manually.
Clay: for outbound-oriented PLG teams
Clay is the most powerful GTM data orchestration tool for PLG companies actively doing outbound on PQL signals. When a user crosses a PQL threshold — paywall hit, enterprise feature used, team expanded — Clay can enrich that signal with company data (funding, headcount, technographics), generate a personalized outreach message, and push it into your email sequencer.
Clay is especially useful for the use case where product data is combined with external signals: a user who used your enterprise feature AND whose company just closed a funding round gets a different message than a user who triggered the same feature at a company with no recent funding activity.
Make: for workflow automation without code
Make (formerly Integromat) is the most-used automation tool for PLG teams that don't want to dedicate a developer to internal tool plumbing. Make's visual interface makes it possible to build complex multi-step workflows: Segment event → HubSpot property update → Slack notification → create task in HubSpot → send email via HubSpot.
Make is less powerful than n8n for complex data transformations, but the learning curve is low and the integration library is broad. For the majority of PLG automation use cases — PQL notifications, data synchronization, onboarding triggers — Make is more than enough.
n8n: for teams with technical bandwidth
n8n is the self-hostable, open-source variant of Make. For PLG companies with higher event volumes, strict data sovereignty requirements, or complex transformation logic, n8n is the better choice. n8n's code nodes (JavaScript, Python) give you full flexibility for logic you can't express in a no-code tool. The downside: n8n has a higher technical bar and requires active maintenance if you self-host.
Minimum PLG stack per growth stage
The right stack differs per stage. Here are three concrete starting points:
Stage 1: Seed (0-500 active users, small technical team)
- Event tracking: PostHog (free tier, all-in-one)
- Product analytics: PostHog (built-in)
- CRM: HubSpot Starter or CRM Free
- Automation: HubSpot workflows + Zapier for simple connections
Total monthly cost: €0-100. Focus: define events, nail down the activation moment, draft a first PQL definition.
Stage 2: Early growth (500-10,000 active users, first RevOps hire)
- Event tracking: Segment (or RudderStack if EU data residency is required)
- Product analytics: Mixpanel Professional
- CRM: HubSpot Sales Hub Professional
- Automation: Make for workflow orchestration, HubSpot workflows for CRM actions
Total monthly cost: €800-2,000. Focus: PQL scoring live, product-data integration with HubSpot operational, first automated PQL workflow shipped.
Stage 3: Scale (10,000+ active users, dedicated GTM team)
- Event tracking: Segment (or RudderStack) + data warehouse (Snowflake or BigQuery)
- Product analytics: Amplitude or Mixpanel Enterprise
- CRM: HubSpot Enterprise, with Salesforce on the table depending on the enterprise sales motion
- Automation: Clay for PQL enrichment and outbound orchestration, n8n or Make for internal workflows
Total monthly cost: €3,000-10,000+. Focus: predictive PQL scoring, multi-touch attribution, expansion MRR tracking, full PLG RevOps architecture.
Common mistakes in PLG tool selection
Buying too early. The most common mistake is acquiring enterprise tooling before the fundamentals are in place. An Amplitude Enterprise license is worthless if event tracking is still messy. Buy tools based on current needs, not future ambition.
Skipping the event taxonomy. Which events you track and how you name them is more important than which tool you use. A poorly defined event taxonomy — inconsistent naming, missing properties, events that don't match the activation moment — guarantees that every tool layered on top produces unreliable output. Invest in the taxonomy first, then in the tools.
Building parallel stacks. It happens that marketing implements its own tracking (Google Analytics, Facebook Pixel), the product team buys a separate analytics tool, and sales works in the CRM on manually maintained data. Three separate "stacks" trying to measure the same thing. The fix: one source of truth via Segment or RudderStack as the central event bus that feeds all downstream tools.
Postponing the CRM integration. Teams that delay the product-data-to-CRM integration until "the product is stable" or "the stack is complete" miss months of PQL data. The integration doesn't need to be perfect to be useful. Start with three or four core properties — activation_date, last_active_date, sessions_last_30_days, paywall_hit — and build from there.
What do you actually need to get started?
The minimum working PLG stack has four elements: a way to capture events (PostHog or Segment), a way to analyze those events (PostHog or Mixpanel), a CRM where product data is visible to sales (HubSpot), and one automated workflow that turns a PQL threshold into a concrete action.
More tools add value the moment the base layer works and you have concrete questions current tooling can't answer. The order of implementation is the biggest lever — not the choice between Mixpanel and Amplitude.
If you want help mapping out the right stack for your growth stage, the GTM Scan gives you a concrete analysis of your current architecture and a prioritized recommendation for the next steps.