Event tracking

Definition
Recording each specific action someone takes on a site or in a product, a page view, a click, a form fill, a feature used, as a discrete event.

Why it matters

Without events, every decision about a page or feature rests on assumption. With them, a vague complaint like "pricing is not converting" becomes specific: where people stop, how long they stayed and what they did next. Behaviour also predicts better than a survey. A buyer who revisits a security page three times is telling the sales team something. An account whose logins fall below its own normal level is a churn risk weeks before it cancels.

How to apply it

  • Write a tracking plan first: a list of event names, when each fires and which properties it carries.
  • Start with events tied to revenue, such as a trial signup, a demo request or a purchase, and make sure they fire reliably.
  • Add the steps above them in the funnel, then in-product events such as invitations sent or a first report created.
  • Name events consistently, for example an object followed by a past-tense verb.
  • Connect events to a known person or account in the CRM, so a named account visiting pricing twice this week becomes a trigger for sales.
  • Check the consent rules for the tool and region in use, because tracking may need permission.

What it is

An event is one thing that happened, stored as a small record. It has a name such as demo_requested, a time, who did it if known, and properties like the page, the plan or the campaign. Add up the records and the real path people take through a site or product becomes visible. It is rarely the tidy sequence a team assumes.

Event tracking also covers what is sometimes called activity tracking: following behaviour over time, such as emails opened, logins and features used, to see who is warming up and who is drifting away.

Common mistakes

  • Tracking every click without a question to answer, then drowning in data.
  • Changing event names without a record, which breaks comparisons over time.
  • Reading counts without watching a few real sessions to see why.
Worked example

Suppose a B2B software team assumes its pricing page is the main route to a demo request. It starts recording named events, such as pricing_viewed and demo_requested, each with a page and a campaign property. Within a month the events show that most demo requests follow a visit to the security page, not the pricing page. Those visitors then get a different follow-up email.

The events live in Mixpanel, which the team uses to see the path through the funnel. Every event name follows one pattern, an object followed by a past-tense verb, so the reports stay readable. The first tracking plan lists only the revenue events, and the rest are added later.

Tools in the example

Some links are affiliate links: we may earn a commission at no cost to you. It never decides a ranking. How we work with partners

  1. Article

    Conversion tracking

    The events tied directly to revenue.

  2. Article

    Funnel analysis

    Reading events as steps and finding where people drop out.

  3. Article

    Engagement

    A score built from the underlying behaviour.

  4. Article

    Multi-touch attribution

    A model that needs clean event data.

Where it shows up

  • Measuring what works and following data to make better decisions. It tells you which changes are worth keeping and which to drop.
    22 chapters