How to Set Up Analytics and Understand Your Users

TLDR

• Measure what leads to a decision. Signup rate, activation, and retention beat raw pageviews. If a number can't change what you build next, stop staring at it.
• Run two tools, not five. PostHog free tier for product analytics (events, funnels, session replay), plus Plausible or Fathom for clean, privacy-friendly pageviews. Total cost around $0 to $15 a month.
• Pageviews tell you who arrived. Events tell you what they did. You want both, but events are where the money hides.
• Funnels show you the exact screen where users quit. Fix that one step before you spend a dollar on more traffic.
• GA4 is free and fine for traffic and Google Ads, but it needs a cookie banner in the EU and defaults to deleting your detailed data after 2 months.

Most founders set up Google Analytics on day one, glance at a visitor count, and never make a single decision from it. That's not an analytics problem, it's a measurement problem. You installed a tool before deciding what question you were trying to answer. This guide fixes that: what to track, what to ignore, which tool to pick, and how to turn the data into a change you actually ship.

What to measure (and what to ignore)

A metric is only worth tracking if it points to an action. Eric Ries, who popularized the term, put it plainly in an essay on Tim Ferriss's blog: the useful metrics are the ones that help you make decisions, while vanity metrics "might make you feel good, but they don't offer clear guidance for what to do." Total pageviews, total signups, and social followers are the classic offenders. They go up and to the right no matter what, so they never tell you to fix anything.

The metrics that matter track whether people actually use and return to your product. Josh Elman, an early growth lead at Twitter, LinkedIn, and Facebook, argues you should define one number for "how many people are really using your product?" At Twitter his team found that if someone visited at least seven times in a month, they almost always stuck around. So "7 visits a month" became the real usage metric, not the registered-user count that looked bigger in a pitch deck.

Here is a starter set for an early product. Each row maps a metric to what it tells you and the decision it should trigger.

Pageviews vs events: the difference that matters

This is the one concept that unlocks the rest. A pageview is a single load of a page. It answers "how many people visited, and where did they come from?" That's traffic analytics, the world of Google Analytics, Plausible, and Fathom.

An event is any specific action you choose to track inside your product: a button click, a completed signup, a feature used, an upgrade. It answers "what did people do, and did they stick?" That's product analytics, the world of PostHog, Mixpanel, and Amplitude. A pageview is really just one kind of event.

The practical takeaway: pageviews tell you who arrived and from where, events tell you what happened next. A marketing dashboard full of pageviews can look healthy while your product quietly leaks every new user on step two of onboarding. You need both lenses, which is why most teams end up running one traffic tool and one product tool.

Pick your tool: PostHog, Plausible, or GA4

You do not need an enterprise stack. For a non-technical founder in 2026, three tools cover the field, and they solve different jobs.

PostHog gives you 1 million events a month free, plus 5,000 session recordings, feature flags, and funnels on the free plan. That is generous enough that most early products never pay. It is open source, so you can self-host if you want full data control. This is your product-analytics workhorse.

Plausible starts at $9 a month for 10,000 pageviews. It is EU-hosted, open source, uses no cookies, and processes no personal data, which means no cookie-consent banner and a script around 1KB that won't slow your site. Fathom is a near-identical alternative starting at around $10 a month. Either gives you clean traffic numbers your competitors' ad-blocked GA4 misses.

GA4 is free and worth keeping if you run Google Ads or want to export to BigQuery. Two catches for founders: in the EU it needs a cookie banner and Google Consent Mode, and its detailed data retention defaults to 2 months (14 months maximum on the free plan, and you have to switch it on manually in Admin). Its deeper Exploration reports also start sampling once a query passes 10 million events, so big numbers get estimated rather than counted.

The realistic 2026 stack: PostHog free tier for product analytics, plus Plausible or Fathom for pageviews. GA4 only if you're spending on Google Ads. Here is the official 90-second tour of what PostHog actually does.

Funnels: find the exact step users quit

A funnel is an ordered sequence of steps you expect a user to complete, for example Signup, then Onboarding completed, then First core action, then Upgrade. Your analytics tool shows the conversion rate and drop-off between each step, so you can see precisely where people fall out.

This is the single most useful view in product analytics, because it turns a vague worry ("users aren't sticking") into a specific bug ("68% of signups never finish onboarding step two"). In PostHog you build one by picking the events in order; it draws the bar chart of who made it through. Fix the worst-performing step, watch that step's conversion, then move to the next leak. That loop alone will improve your product faster than any redesign.

A funnel converts a feeling into a number: not "people aren't converting" but "we lose 68% of users on step two." One is an anxiety. The other is a to-do.

Turning data into decisions

Data is not the goal, a decision is. Build the habit with a short weekly loop instead of a dashboard you doom-scroll.

• Pick one number that matters this month (activation or week-4 retention are good first choices). Write it down. Ignore the rest.
• Form a guess, then check it. "I think people drop off because onboarding is confusing." Open the funnel and see if the numbers agree before you build anything.
• Change one thing. Rewrite the confusing step, then watch that step's conversion for a week. One change at a time keeps cause and effect clear.
• Watch a few real sessions. PostHog session replays show you where a user hesitates or rage-clicks. Ten minutes of watching beats an hour of guessing.
• Kill vanity reports. If a chart hasn't changed a decision in a month, remove it from your view so it stops stealing attention.

Ries's test for a good metric is the three A's: actionable (it points to a clear cause and effect), accessible (your whole team can read it), and auditable (you trust the number is real). Run every metric you track through that filter and most dashboards get a lot shorter.

Your setup checklist

Do this in an afternoon, in order.

• Write your one question first. "Are new users reaching the core action?" beats "let me see the numbers."
• Install PostHog (free tier) and confirm pageviews are recording. This is your product-analytics base.
• Add 3 to 5 key events by hand: signup completed, onboarding done, first core action, upgrade. Name them clearly and consistently.
• Build one funnel from those events and note today's drop-off numbers as your baseline.
• Add Plausible or Fathom for clean, cookie-free pageviews, or set up GA4 if you run Google Ads (and change data retention to 14 months in Admin).
• Turn on session replay in PostHog and watch five real sessions this week.
• Book a 20-minute weekly review with yourself. One metric, one guess, one change.
• Delete the vanity charts that never lead to a decision.

Analytics done right is small: two tools, a handful of events, one funnel, and a weekly habit of changing one thing. Get that loop running and you'll stop guessing what your users want, because you'll be watching what they actually do.