Our overall activation rate was 52%. Not great, not terrible. The product team was focused on improving it through better onboarding.
Then I did something that changed everything: I segmented activation rates by company size, industry, and use case.
The results stopped me cold:
Activation rate by company size:
- 1-10 employees: 78%
- 11-50 employees: 61%
- 51-200 employees: 44%
- 200+ employees: 23%
We didn't have one activation problem. We had three completely different products for three completely different audiences, and we were treating them all the same.
This insight changed our entire product strategy.
The Discovery: One Product, Multiple Realities
I spent a week pulling cohort data. I segmented every signup from the past 6 months by:
Company characteristics:
- Employee count
- Industry
- Revenue (when available)
- Geography
User characteristics:
- Job title/role
- Team size
- Department
Behavioral characteristics:
- Referral source (how they found us)
- Signup trigger (demo request, free trial, product-led signup)
- Use case selected during onboarding
Then I calculated activation rates for each segment.
The patterns were stark:
Finding 1: Company Size Correlated Inversely with Activation
SMB (1-50 employees): 71% activation rate Mid-market (51-200): 44% activation Enterprise (200+): 23% activation
This was backwards from what we expected.
We'd been building features for enterprise customers because they paid more. But enterprise customers were activating at 1/3 the rate of SMB customers.
Why?
I interviewed 20 enterprise users who didn't activate. Pattern emerged:
"Your product looks great for small teams, but we need [compliance feature], [SSO], [admin controls], [data governance], [multi-region deployment]... we can't even start using it until those exist."
Enterprise customers weren't failing to activate because of onboarding problems. They were failing to activate because our product wasn't enterprise-ready.
Meanwhile, SMB customers activated easily because our product was perfect for their simpler needs.
Finding 2: Use Case Determined Activation Success
We let users select a use case during onboarding:
- Marketing analytics: 68% activation
- Product analytics: 59% activation
- Financial reporting: 31% activation
- Sales analytics: 29% activation
2x difference in activation between best and worst use case.
Why?
Our product had deep functionality for marketing and product analytics (the use cases our founders came from). Financial and sales analytics were half-baked—we'd added them because prospects asked for them, but they weren't truly ready.
Users selecting financial/sales analytics couldn't accomplish their goals because the features weren't complete. They churned during trial.
We were attracting users for use cases we couldn't serve well.
Finding 3: Referral Source Predicted Activation
Users from product-led signup: 64% activation Users from sales demo: 41% activation Users from paid ads: 37% activation
Why?
Product-led signups were self-selecting—they'd tried the free version, knew what to expect, and signed up when they hit limitations. High intent, aligned expectations.
Sales demos attracted aspirational users—people who liked what they saw in the demo but whose reality didn't match the demo scenario. Misaligned expectations.
Paid ads attracted the least qualified traffic—people clicked because the ad was compelling, not because they had an urgent need.
Finding 4: Industry Showed Extreme Variance
SaaS companies: 74% activation E-commerce: 61% activation Healthcare: 38% activation Financial services: 29% activation
Why?
Our product had built-in integrations for common SaaS and e-commerce tools (Stripe, Shopify, Google Analytics).
Healthcare and financial services users needed integrations we didn't support (EMR systems, banking APIs). They couldn't connect their data, so they couldn't activate.
We were accepting signups from industries we couldn't serve.
The Uncomfortable Truth
When I presented this analysis to the executive team, it sparked a 3-hour debate.
The data showed clearly:
- We activated SMB SaaS companies doing marketing/product analytics at 74%
- We activated enterprise healthcare companies doing financial reporting at 19%
These are fundamentally different products with different needs, different onboarding, different features, and different business models.
We were trying to be one product for everyone. We were succeeding with nobody.
The CEO asked the hard question: "Should we pick an ICP and focus on them, or try to serve everyone better?"
The data made the decision obvious.
The Strategic Pivot
We made three major decisions based on cohort analysis:
Decision 1: Define Our True ICP
New ICP: SMB SaaS and e-commerce companies (1-50 employees) using our product for marketing or product analytics
Why:
- 72% activation rate in this segment
- 84% retention at 90 days
- $40 average monthly revenue per user
- Low support burden
- Used integrations we already had
Not our ICP: Enterprise companies in healthcare/financial services doing financial/sales reporting
Why:
- 24% activation rate in this segment
- 41% retention at 90 days
- Required features we didn't have
- High support burden
- Would take 18+ months to serve properly
This was painful. We had to say no to large enterprise deals because we couldn't serve them well yet.
But the math was clear: 10 SMB customers activating at 72% was better than 1 enterprise customer activating at 24%.
Decision 2: Redesign Onboarding for Our ICP
We'd built "universal" onboarding that tried to serve every segment.
New approach: Onboarding optimized specifically for SMB SaaS/e-commerce doing marketing analytics.
What changed:
- Pre-built integrations for Stripe, Shopify, Google Analytics (most common for our ICP)
- Use-case templates specific to SaaS metrics (MRR, churn, CAC) and e-commerce metrics (LTV, AOV, conversion)
- Removed enterprise features from main flow (SSO, admin controls, compliance)—they became add-ons
- Industry-specific sample data (show SaaS companies SaaS examples, not generic examples)
Results after 8 weeks:
- Activation rate for ICP segment: 72% → 81%
- Activation rate overall: 52% → 59% (improved despite dropping enterprise features from main onboarding)
- Time-to-activation for ICP: 3.2 days → 1.8 days
Decision 3: Adjust Go-to-Market Strategy
Marketing changes:
- Stopped paid ads targeting broad "analytics" keywords (attracted wrong ICP)
- Started content marketing targeting SaaS/e-commerce operators (our ICP)
- Updated homepage messaging from "analytics for everyone" to "analytics for SaaS and e-commerce companies"
Sales changes:
- Sales qualified leads based on company size and industry (automatically disqualified enterprise healthcare/finance)
- Demo flow tailored to SaaS/e-commerce use cases instead of generic demo
- Stopped pursuing enterprise deals that didn't fit ICP
Product changes:
- Roadmap prioritized features for SMB SaaS/e-commerce (revenue forecasting, cohort analysis, churn prediction)
- Deprioritized enterprise features (SSO, complex permissioning) to later phase
- Built 8 new integrations for tools our ICP uses
The Results: Activation and Retention Both Improved
6 months after the strategic pivot:
Activation metrics:
- Overall activation rate: 52% → 67%
- ICP activation rate: 72% → 83%
- Non-ICP signups: Decreased from 40% of total to 18% (more qualified funnel)
Retention metrics:
- 90-day retention: 58% → 71%
- 12-month retention: 42% → 63%
Business metrics:
- MRR growth: +34% (despite turning away enterprise deals)
- CAC: -22% (marketing more efficient targeting ICP)
- Support tickets per customer: -31% (ICP needed less support)
- NPS: 38 → 56
Same product. More focused ICP. Better outcomes across every metric.
What Cohort Analysis Taught Me
This experience taught me several lessons I now apply to every product:
Lesson 1: Overall Metrics Hide Critical Insights
Looking at "activation rate: 52%" told us nothing actionable. That number was an average of:
- SMB SaaS marketing analytics: 78% activation
- Enterprise healthcare financial reporting: 19% activation
Averages hide the truth. Segmentation reveals it.
If we'd only looked at the overall number, we'd have kept building features for everyone and succeeding with no one.
Lesson 2: Product-Market Fit Is Segment-Specific
We thought we had "decent" product-market fit (52% activation, 58% retention).
We actually had:
- Excellent product-market fit with SMB SaaS/e-commerce (78% activation, 84% retention)
- Poor product-market fit with enterprise and non-SaaS industries (24% activation, 41% retention)
You can't have universal product-market fit. You have product-market fit with specific segments.
Lesson 3: Not All Revenue Is Good Revenue
We were excited when enterprise companies signed up because they paid 5-10x what SMB customers paid.
But the data showed:
- Enterprise activation: 23%
- Enterprise retention: 41%
- Enterprise support burden: 3.2x higher than SMB
Lifetime value calculation:
- Enterprise: $10,000/year contract × 41% retention × 23% activation × 50% likelihood they actually pay (many churned before first invoice) = $471 actual LTV
- SMB: $500/year contract × 84% retention × 78% activation × 95% payment rate = $312 actual LTV
Not as different as we thought. And enterprise required way more support resources and product development.
Focus on revenue from segments where you have strong product-market fit, not just highest ACV.
Lesson 4: Cohort Analysis Reveals Roadmap Priorities
Before cohort analysis, our roadmap was a mix of:
- Enterprise features (because big deals asked for them)
- SMB features (because that's where usage was)
- Random feature requests (because someone asked)
After cohort analysis, roadmap prioritization became obvious:
Build features that:
- Improve activation/retention for our ICP (SMB SaaS/e-commerce)
- Help more people become our ICP (expand integrations for more SaaS tools)
- Increase ARPU within our ICP (advanced features they'll pay for)
Everything else: deferred or killed.
Lesson 5: ICP Misalignment Looks Like Activation Problems
When we had 52% activation and couldn't figure out why it was low, we kept trying to fix onboarding.
The problem wasn't onboarding. It was ICP misalignment.
40% of our signups were never going to activate because we couldn't serve their needs. No amount of onboarding optimization would fix that.
When activation is low, check if you're attracting the wrong ICP before you rebuild onboarding.
How to Run Cohort Analysis on Activation
Here's my process:
Step 1: Pull Activation Data by Cohort
Pull 3-6 months of signup data. For each user, track:
- Did they activate? (Yes/No)
- Days to activation
- Company size (employees, revenue)
- Industry
- Use case / job-to-be-done
- Referral source
- Geographic region
- User role/title
Step 2: Calculate Activation Rate by Segment
For each dimension, calculate activation rate:
By company size:
- 1-10 employees: X% activation
- 11-50: X% activation
- 51-200: X% activation
- 200+: X% activation
By use case:
- Use case A: X% activation
- Use case B: X% activation
- etc.
By referral source:
- Organic search: X% activation
- Paid ads: X% activation
- Sales demo: X% activation
- Product-led: X% activation
Step 3: Identify Highest and Lowest Performing Segments
Sort segments by activation rate.
Top performers: 70%+ activation Bottom performers: <40% activation
Look for 2x+ differences between segments.
Step 4: Interview Users in Each Segment
Interview 10-15 users from:
- Highest performing segment (to understand what's working)
- Lowest performing segment (to understand what's failing)
Ask:
- "Why did you sign up?"
- "What were you hoping to accomplish?"
- "What went well / poorly in your first week?"
- "Did you get the outcome you wanted?"
Step 5: Identify Patterns and Root Causes
Look for themes:
High-activation segments: What do they have in common?
- Similar needs?
- Similar workflows?
- Similar maturity level?
- Similar tech stack?
Low-activation segments: Why are they failing?
- Missing features?
- Wrong expectations?
- Poor product fit?
- Need different onboarding?
Step 6: Make Strategic Decisions
Based on findings:
Option A: Double down on high-activation segments
- Define them as your ICP
- Build features for them
- Optimize onboarding for them
- Adjust marketing to attract more of them
Option B: Fix low-activation segments
- Build features they need
- Create separate onboarding flows
- Invest in serving them properly
Most companies should choose Option A (focus on where you're winning) unless low-activation segments represent massive strategic opportunity worth multi-year investment.
Step 7: Measure Impact
After making changes:
- Track activation rate for target ICP
- Track % of signups that are target ICP
- Track retention for target ICP vs. others
- Compare to pre-change baseline
Goal: Higher activation and retention for focused ICP, even if overall numbers initially drop as you filter out bad-fit customers.
Common Patterns in Cohort Analysis
After running this analysis for multiple products, I see recurring patterns:
Pattern 1: Founder-Market Fit Drives Early Success
The segments with highest activation are usually the ones the founders came from.
SaaS founders build great SaaS tools. E-commerce founders build great e-commerce tools.
When you expand beyond founder expertise, activation drops.
Solution: Acknowledge where your expertise lies and focus there first.
Pattern 2: Free Trial Activation Differs from Sales-Led Activation
Product-led signups activate at higher rates than sales-led signups because they're self-selecting.
But: Sales-led deals often have higher ACV if they do activate.
Common mistake: Treating both segments with same onboarding.
Solution: Separate onboarding flows for self-serve vs. sales-assisted customers.
Pattern 3: Use Case Matters More Than Industry
You might think "healthcare" or "finance" are meaningful segments.
They're not. The use case matters more.
Healthcare company doing marketing analytics activates similarly to SaaS company doing marketing analytics.
Segment by job-to-be-done, not by industry.
Pattern 4: Company Size Correlates with Complexity Needs
Smaller companies activate faster because they have simpler needs.
Larger companies need more features (SSO, permissioning, compliance) before they can activate.
Don't try to serve 1-person companies and 1,000-person companies with the same product.
Pattern 5: Geographic Differences Often Reflect Product-Market Fit
If you activate North American customers at 70% but European customers at 35%, investigate why.
Often it's:
- Language/localization issues
- Missing integrations for local tools
- Different buying behaviors
- Pricing misalignment
Geographic differences reveal product-market fit issues.
The Uncomfortable Truth About Cohort Analysis
Most product teams avoid deep cohort analysis because they're afraid of what they'll find.
They're right to be afraid.
Cohort analysis often reveals:
- Your highest-paying customers have the worst activation/retention
- Your best product-market fit is with smaller customers than you want to serve
- Features you spent 6 months building only matter to 10% of users
- Your ICP definition is wrong
This is uncomfortable. It often requires strategic pivots, saying no to revenue, or admitting you built the wrong features.
But avoiding the analysis doesn't make the problems go away. It just means you'll keep:
- Building features for segments with poor product-market fit
- Marketing to audiences who won't activate
- Celebrating revenue from customers who will churn
- Wondering why your overall metrics are mediocre
The best product teams:
- Run cohort analysis on activation quarterly
- Segment by company size, industry, use case, and source
- Make hard decisions to focus on highest-performing segments
- Build separate onboarding for different cohorts when needed
- Track activation by segment, not just overall
The teams with low activation:
- Look at overall activation rate without segmentation
- Treat all customers the same in onboarding
- Chase any revenue without checking for product-market fit
- Build features for whoever asks loudest
- Wonder why improvements to onboarding don't move the needle
I was on the second team until cohort analysis forced me to confront uncomfortable truths.
Now I segment everything. And every time I do, I find insights that change strategy.
Your activation problem probably isn't universal. It's segment-specific.
Run the cohort analysis. Find where you're winning. Double down there.
Stop trying to be everything to everyone. Be exceptional for someone.