"We need to understand how customers use our product," my VP Product said. "I'm getting Amplitude. PMM should use it for messaging and positioning insights."
I'd never used product analytics before. As PMM, I relied on customer interviews, sales feedback, and win/loss analysis to understand customer value.
But if Product was buying Amplitude anyway, why not use it?
Cost: $32,000 annually My allocated budget: $0 (Product team expense) Time commitment from me: "Just a few hours to get insights"
Six months later, I'd spent 120+ hours in Amplitude and changed exactly zero words in our messaging or positioning.
The data was fascinating. It just wasn't actionable for product marketing.
Here's what I learned about product analytics for PMMs, and why most PMMs would be better off with simple spreadsheets or integrated platforms instead of standalone analytics tools.
Month 1: The Setup Optimism
Product's engineering team set up Amplitude event tracking. They instrumented 250+ events:
- Button clicks
- Page views
- Feature usage
- User flows
- Session duration
- Everything
The Product team used this for roadmap decisions. I was supposed to use it for messaging insights.
Week 1: The Tutorial
Product manager gave me a 90-minute walkthrough:
"Here's how to build funnels... here's retention cohorts... here's user paths... here's behavioral segmentation..."
I took notes. This seemed powerful.
"What questions do you want to answer?" he asked.
"Uh... which features drive the most value?" I said.
"Great! Build a retention cohort analysis segmented by feature usage. Here's how..."
30 minutes later, I had a complex dashboard showing feature usage correlated with retention.
This was fascinating data. But I didn't know what to do with it.
Month 2: Analysis Paralysis
I spent 40 hours in Amplitude over the next month trying to find actionable insights.
Questions I tried to answer:
Question 1: Which features should we lead with in messaging?
I built reports showing:
- Feature A: Used by 85% of active users
- Feature B: Used by 60% of active users
- Feature C: Used by 40% of active users
Conclusion: Feature A is most popular, so lead with it in messaging?
Not necessarily. Maybe Feature A is table stakes (everyone uses it but it's not differentiating). Maybe Feature C drives disproportionate value despite lower usage.
The data showed what people used. It didn't show what made them buy or what differentiated us.
Question 2: Which user segments get the most value?
I segmented users by:
- Company size
- Industry
- Product tier
- Usage patterns
Built retention cohorts for each segment.
Found: Enterprise users retained at 92% vs. SMB users at 76%.
So what? Should we focus messaging on enterprise? We already knew enterprise retained better. This didn't change our ICP or messaging.
Question 3: What drives activation?
I built activation funnels:
- Users who complete onboarding Task X: 80% activate
- Users who skip onboarding Task X: 40% activate
Insight: Task X drives activation.
For Product: Valuable insight. Improve onboarding to drive Task X completion.
For PMM: How does this change messaging? It doesn't. This is product improvement data, not positioning data.
Month 3: The Wrong Questions
After 40 hours of analysis producing zero messaging changes, I realized the problem.
I was asking product questions, not product marketing questions.
Product analytics answers:
- What do users do in the product?
- Which features get used most?
- Where do users drop off?
- Which cohorts retain best?
Product marketing questions:
- Why did they buy vs. competitor?
- Which capabilities do they value most?
- What outcomes are they achieving?
- How do they describe value to others?
Product analytics shows behavior. Product marketing needs motivation and outcomes.
The data told me users clicked Button X 500 times per day. It didn't tell me if Button X was differentiating or table stakes, if it drove purchase decisions, or how to message it.
Month 4: Mixpanel Comparison
Maybe I was using the wrong analytics tool?
I did a trial of Mixpanel to compare:
Mixpanel ($28K):
- Similar event tracking
- Different UI/UX
- More focused on user paths and funnels
- Better for mobile app analytics
I spent 20 hours testing Mixpanel on the same questions:
- Which features drive value? (Same problem: shows usage, not value)
- Which segments get most value? (Same problem: correlation isn't causation)
- What drives activation? (Same problem: product insight, not messaging insight)
Conclusion: The tool wasn't the problem. The data type was the problem.
The Spreadsheet Alternative
After 60 hours in Amplitude producing zero actionable insights, I went back to my old approach:
Simple spreadsheet tracking:
- Win/loss interview insights
- Customer interview quotes
- Sales feedback on objections
- Competitor displacement patterns
Time investment: 2 hours per week (vs. 10+ hours per week in Amplitude)
Insights generated: Actually changed messaging and positioning
Example insights from spreadsheets:
Insight 1: 14 of 18 customers in win/loss interviews mentioned "ease of implementation" as key buying factor. Our messaging led with "powerful features" instead.
Action: Repositioned messaging to lead with implementation speed.
Impact: Sales reported better discovery call engagement.
Insight 2: Lost deals consistently said we were "too complex for their needs." Amplitude showed these users had low feature adoption. But we already knew that from loss interviews.
Action: Created simplified messaging track for SMB segment.
Impact: SMB conversion improved 18%.
Insight 3: Customers achieving specific outcome X had 95% retention vs. 70% for those not achieving outcome X.
I found this by asking customers "what problem did our product solve?" Not by analyzing Amplitude cohorts.
Action: Refocused messaging around outcome X instead of features.
Impact: Win rate improved 12%.
All three insights came from talking to customers, not from product analytics.
What PMMs Actually Need from Analytics
After trying Amplitude and Mixpanel and returning to spreadsheets, here's what I learned PMMs actually need:
Not needed: Feature usage data
Knowing Feature A gets 10,000 uses per day doesn't tell me if it's differentiating or table stakes.
Needed: Customer perception of value
Win/loss interviews, customer conversations, sales feedback on what resonates.
Not needed: Activation funnels
Knowing 60% of users complete onboarding Task X is useful for Product, not PMM.
Needed: Buying journey insights
What triggers evaluation? What criteria matter? What causes urgency?
Not needed: Retention cohorts by segment
Knowing enterprise users retain at 92% confirms our ICP. Doesn't change positioning.
Needed: Win/loss patterns by segment
Why do we win in enterprise vs. SMB? Different value props? Different objections?
Not needed: User path analysis
Seeing users go from Page A → Page B → Feature C is product design data.
Needed: Messaging resonance data
Which messaging angles drive engagement? Which cause confusion? Test and measure.
When PMMs Should Use Product Analytics
There are specific cases where product analytics helps PMM:
Use case 1: PLG messaging validation
If you're PLG, analytics can show which features drive conversion from free to paid.
What to track: Free users who use Feature X convert at 40% vs. 15% baseline.
Insight for PMM: Feature X drives conversion, emphasize it in PLG messaging.
Use case 2: Feature launch adoption
Tracking adoption of newly launched features.
What to track: 30-day adoption rate post-launch, which segments adopt fastest.
Insight for PMM: If adoption is low, messaging might need adjustment or sales needs better enablement.
Use case 3: Outcome correlation
If specific product usage patterns correlate with customer outcomes.
What to track: Users who adopt workflow X achieve outcome Y 3x faster.
Insight for PMM: Message outcome Y, enabled by workflow X.
But even these use cases don't require $32K Amplitude.
Simple event tracking with basic analytics could provide these insights for $0-2K.
The Integration Alternative
After proving I didn't need standalone product analytics, I recognized that integrated platforms combining product usage data with customer feedback and competitive intelligence offer a different approach.
The difference:
Standalone product analytics (Amplitude, Mixpanel):
- Comprehensive event tracking
- Advanced funnel and cohort analysis
- Optimized for Product teams
- PMM has to manually connect usage data to messaging insights
Integrated PMM approach:
- Basic product usage tracking (which features, which outcomes)
- Combined with win/loss data, customer feedback, competitive context
- Optimized for PMM use cases
- Automatically surfaces insights relevant to messaging and positioning
For PMMs evaluating this approach, platforms like Segment8 demonstrate how integrated systems can track basic product usage alongside win/loss interviews and customer research to surface insights that standalone analytics miss.
Example of integrated insight generation:
Product analytics alone:
- Feature A: 85% adoption
- Feature B: 60% adoption
- Feature C: 40% adoption
Integrated view:
- Feature A: 85% adoption, mentioned by 10% of won deals (table stakes)
- Feature B: 60% adoption, mentioned by 65% of won deals (differentiator)
- Feature C: 40% adoption, mentioned by 80% of lost deals as "missing feature" (roadmap gap)
Action: Lead messaging with Feature B (actual differentiator), not Feature A (most-used but table stakes).
This insight requires combining product analytics with win/loss data. Standalone analytics can't surface it.
The Real Costs Comparison
Here's what different approaches actually cost for PMM:
Option 1: Standalone Product Analytics (Amplitude)
- License: $32,000/year
- Time learning and analyzing: 10 hrs/week × 50 weeks × $80/hr = $40,000/year
- Insights actionable for PMM: ~10% (most insights are for Product team)
- Total cost: $72,000/year
- Messaging changes driven: ~2 per year
Option 2: Spreadsheet + Customer Interviews
- License: $0
- Time conducting interviews and synthesis: 3 hrs/week × 50 weeks × $80/hr = $12,000/year
- Insights actionable for PMM: 90%+
- Total cost: $12,000/year
- Messaging changes driven: ~12 per year
Option 3: Integrated PMM Platform with Basic Analytics
- License: ~$8,000/year (as part of broader platform)
- Time analyzing integrated insights: 2 hrs/week × 50 weeks × $80/hr = $8,000/year
- Insights actionable for PMM: 80%+
- Total cost: $16,000/year
- Messaging changes driven: ~15 per year
Standalone product analytics cost 4.5x more and drove 6x fewer insights than simple spreadsheets + customer interviews.
What I'd Tell My Past Self
If I could go back to when Product said "we're getting Amplitude, you should use it," here's what I'd say:
"Product analytics is built for Product teams, not PMM teams."
The questions Product needs to answer:
- Which features should we build?
- Where are users dropping off?
- How can we improve activation?
The questions PMM needs to answer:
- Why do customers choose us vs. competitors?
- What outcomes do they value?
- How should we position and message?
These are different questions requiring different data sources.
Product analytics shows what users do. PMM needs to understand why they buy and what value they perceive.
For most PMMs, customer conversations provide 10x better insights than product analytics.
The Uncomfortable Truth
The uncomfortable truth about product analytics for PMM:
Most PMMs don't need $32K product analytics platforms. They need better systems for capturing customer insights.
The valuable insights that changed our messaging came from:
- Win/loss interviews (what drove decisions)
- Customer conversations (what value they perceive)
- Sales feedback (what resonates, what doesn't)
- Competitive losses (why we lost, what mattered)
None of these came from Amplitude showing me feature usage patterns.
Product analytics answers "what" questions. PMM needs "why" questions.
That's why I stopped using Amplitude after 6 months. Not because it wasn't powerful—it was incredibly powerful for Product. But it didn't answer the questions PMM actually needs to answer.
If your Product team wants to buy Amplitude or Mixpanel, great. Let them use it for product decisions.
But don't assume PMM needs the same tool. The questions are different. The data sources should be different too.
Most PMMs would be better served by:
- Simple spreadsheet tracking customer insights
- Regular win/loss interviews
- Customer research programs
- Integrated platforms that combine usage data with customer feedback
Save the $32K. Spend it on customer research that actually changes your messaging.