How to Set Up Analytics for Your Startup

Analytics setup gets over-complicated. Founders who are intimidated by it delay it entirely (and fly blind). Founders who over-invest in it build elaborate dashboards they never look at. The right answer is a minimal stack that answers the three questions that actually matter for a pre-PMF startup:

1. Where are visitors coming from?
2. What do they do when they get there?
3. Which behaviors predict whether a user stays or churns?

Here's how to answer all three in a day's setup.

Quick answer: You need at minimum two tools: Plausible or Fathom for website analytics (cookieless, GDPR-friendly, fast setup) and PostHog for product analytics (generous free tier, handles events, funnels, and session recordings). Add Sentry for error tracking from day one. The weekly review habit — checking the same six metrics every week — matters more than which specific tools you choose.

Layer 1: Website analytics (free, immediate)

Install this first, before you do anything else.

Your website analytics tool tells you: how many people visit your site, where they come from (search, social, referrals), which pages they view, and what they click. This data is foundational — you cannot measure any marketing effort without it.

The right tool for early-stage:

• Plausible Analytics or Fathom Analytics: Cookieless, privacy-friendly, no GDPR consent banner needed, fast, and correct. From €9/mo (Plausible) or $14/mo (Fathom). Both give you all the data you actually need in one clean dashboard.
• Google Analytics 4 (free): Free, extremely powerful, but complex to configure correctly and requires a cookie consent banner in most jurisdictions. Worth using if you need Google Ads attribution or deep funnel analysis.

Install: Add a script tag to your <head>. Done.

What to check weekly:

• Top traffic sources (where is traffic coming from?)
• Top pages (what do people actually read?)
• Bounce rate or engagement rate (are visitors engaging or leaving?)
• Goal completions (are people signing up or contacting you?)

Layer 2: Product analytics (when you have 20+ active users)

Website analytics tells you what happens on your marketing site. Product analytics tells you what happens inside your product — who's using what, who's getting value, and who's about to churn.

The right tool:

• PostHog: Open-source, generous free tier (1M events/mo), self-hostable. Handles event tracking, funnels, session recordings, feature flags, and A/B tests. The best option for most indie SaaS at this stage.
• Mixpanel: More powerful cohort analysis and retention visualization. Free up to 20M events/mo. Better for products where retention and long-term behavior analysis is critical.
• Amplitude: Similar to Mixpanel. Free tier available.

Install: Add the JavaScript SDK to your product. Define 5-10 events that matter (listed below). Done.

The 10 events every SaaS product should track:

1. user_signed_up — with source (how they found you)
2. user_activated — completed your activation event (product-specific)
3. feature_used [feature_name] — for every core feature
4. upgrade_clicked — intent to upgrade
5. subscription_started — became a paying customer
6. subscription_cancelled — churned
7. invite_sent — viral / referral signal
8. file_imported or data_connected — setup completion signal
9. export_created — value delivery signal
10. support_ticket_opened — friction signal

Start with these and add more as you learn which events matter.

Layer 3: Error tracking and performance monitoring (day 1 in production)

You cannot debug production issues you can't see. Error tracking should go live the moment you deploy.

Tools:

• Sentry (free tier covers most early-stage needs): Catches JavaScript errors, backend exceptions, and performance problems. Shows you the exact line of code that caused the error and the user context when it happened.
• Posthog (free): Also handles basic session replay, which helps debug UI issues.

Install: Add the Sentry SDK to your frontend and backend. Configure the source map upload so stack traces show your code, not minified output. Done.

Layer 4: Revenue analytics (when you have paying customers)

Your payment processor (Stripe, Paddle, Lemon Squeezy) gives you basic revenue data. But understanding MRR, churn, LTV, and cohort analysis requires a dedicated tool or some calculation work.

Options:

• Stripe Dashboard (built-in, free): Shows MRR, churn rate, and revenue trends. Adequate for the first 50-100 customers.
• Baremetrics or ChartMogul: Revenue intelligence tools that sit on top of Stripe. Better visualization, cohort analysis, and benchmark data. From $129-199/mo. Worth it above ~$10K MRR.
• ProfitWell (now Paddle's Metrics): Free if you're on Stripe. Good basic MRR dashboard.
• Build it yourself in Metabase or a spreadsheet: For very early stage, calculating MRR from a Stripe export in a spreadsheet is entirely adequate.

Layer 5: Conversion tracking for paid acquisition (if running ads)

If you run Google Ads or Meta Ads, you need conversion tracking configured — otherwise you're spending money with no way to measure which ads produce paying customers.

Google Ads: Install Google Ads conversion tracking or import conversions from GA4. Tag the event that matters (trial started, subscription purchased).

Meta Ads: Install Meta Pixel on your site; configure standard events for the same conversion actions.

This step is often done wrong: many founders configure conversion tracking to fire on the signup page, not the payment completion page. Track the action that makes money, not the one before it.

The analytics dashboard: what to review weekly

Set aside 30 minutes weekly to review:

1. Website traffic: Visitors this week vs. last week. Top sources. Top landing pages.
2. Signup rate: Visitors → signups. Is the conversion rate stable or changing?
3. Activation rate: Signups → activated users this week. Are people reaching your core value?
4. Trial-to-paid rate: Trials that started this month → conversions.
5. Churn: Any cancellations this week? Who? Why?
6. Errors: Any new error spikes in Sentry?

These six numbers tell you the health of your business. Everything else is context.

Common analytics mistakes

Tracking everything instead of the right things. More events does not mean more insight. Define 10 events that directly connect to your business outcomes and track those perfectly before adding more.

Not checking the data. A dashboard nobody looks at is waste. Block 30 minutes weekly and look at the same 6 metrics every time. Pattern recognition requires consistency.

Not setting up analytics before launch. Every day you run without analytics is data you can never recover. Install the script the day you launch.

Using GA4 without configuring it properly. Default GA4 is a privacy consent banner nightmare and a configuration mess. Set up the conversion events, connect to Search Console, configure the attribution model, or use Plausible/Fathom instead.

Conflating correlation with causation in analytics. "Users who use Feature X have 40% lower churn" does not mean Feature X reduces churn — it may mean that retained users naturally use more features. Run experiments to establish causality before investing resources.

---

Frequently Asked Questions

Do I need multiple analytics tools or can I use one?

You need at least two: website analytics (Plausible/Fathom/GA4) and product analytics (PostHog/Mixpanel). These answer different questions. Error tracking (Sentry) is a third tool that's also essential. Revenue analytics is often handled by your payment processor at first.

When should I add product analytics — at launch or later?

At launch, or even before. You can instrument your product before you have users and the events will fire as soon as users arrive. Setting up analytics retroactively means missing early-stage data that would have been valuable.

Is PostHog free tier enough for early stage?

Yes. 1M events/mo free is enough for most products until $50K+ MRR. The session recording and feature flag features are also available on the free tier.

Should I build a custom analytics dashboard?

No — not until the off-the-shelf tools genuinely can't answer your questions. Building a custom analytics system is a significant engineering investment. Plausible + PostHog + Sentry answers 90% of early-stage analytical questions without custom development.

---

Find analytics and monitoring tools at tools.cafe.