Set up analytics

Why this step matters

Without data you are guessing which marketing works, where users get stuck and whether the product is improving. With too much data you drown in dashboards nobody reads. The right setup for a young startup tracks a handful of events that answer three questions: are people finding us, do they reach the moment of value, and do they come back? Put this in place before launch, because you cannot measure a launch retroactively.

Analytics also keeps the team honest. When opinions differ about which feature matters or which channel works, a shared definition of activation and a simple weekly chart settle the argument faster than debate. Start small and precise; you can always add events later, but cleaning up a messy tracking plan is slow and error-prone.

How to do it

1. Define your key moments. Write down your activation event (the first time a user gets real value, for example

"created and shared a first report") and your retention signal (for example "active in week four"). These two definitions guide everything else.

1. Choose your tool categories. Most startups need two things: web analytics for traffic and sources (where

visitors come from, which pages they read), and product analytics for what signed-in users do (funnels, cohorts, retention). Some tools do both. Privacy-focused web analytics tools avoid cookies and can reduce consent work; self-hosted open-source options give you full control of the data.

1. Install web analytics on every public page. Add the script or tag to your landing page, docs and app. Set up

UTM conventions (source, medium, campaign) so links from launches, newsletters and ads are attributed correctly.

1. Track five to ten product events. Start with sign-up, activation, the core action, invite or share, upgrade

and cancel. Use a consistent naming pattern such as report_created and attach useful properties like plan or source, never personal details like names or email addresses.

1. Build three views. A sign-up to activation funnel, a weekly retention cohort chart and a traffic-by-source

report. If a chart does not lead to a decision, delete it.

1. Connect revenue. Once you take payments, pull key figures (new customers, monthly recurring revenue, churn)

from your payment provider's dashboard or a revenue analytics tool rather than recalculating them by hand.

1. Handle consent and privacy. In the EU and UK, cookies and similar tracking that are not strictly necessary

generally need prior consent; other regions have their own rules, such as state privacy laws in the US. Update your privacy policy to list the tools you use, and configure IP anonymisation and data retention where available.

1. Add qualitative signals. A short in-app survey, a feedback button and, where lawful and disclosed, session

recordings with sensitive fields masked show why the numbers look the way they do.

1. Review weekly. Put 20 minutes in the calendar to look at the three views, note one insight and choose one

change to test.

What it costs and how long it takes

Most analytics tools have free tiers that cover an early product, so $0 is realistic at the start; paid plans usually scale with events or monthly visitors and can grow quickly with traffic. Installation takes a day or two, but you need two to four weeks of data before retention numbers mean much.

Free credits, perks and services that help

Analytics, session recording and survey vendors are common in startup perk bundles, often with a free year or larger event limits for young companies. Browse developer tools perks and the full perks directory. Many website builders on website builders include basic traffic analytics too, which can be enough for a landing page.

Common mistakes

• Tracking without a question. Hundreds of events with unclear names produce no answers.
• Vanity metrics. Page views and total sign-ups feel good but say little about whether people stay.
• Personal data leaks. Emails in event properties or URLs end up in third-party tools and logs.
• Broken attribution. Missing UTM tags make every launch look like "direct" traffic.
• Consent banners that do not work. Loading trackers before consent in the EU or UK can breach the rules.
• Never acting. Data only helps if someone looks at it and changes something each week.

Go deeper

Find AI coding agents, Agent Skills and MCP servers that connect to analytics tools and help you instrument your app on VibeCodes.