What a conversion is

A conversion is any outcome you decide is worth counting - a purchase, a sign-up, a demo request, a download. It is the point where a visit stops being traffic and becomes value. Everything else in marketing analytics exists to serve one question this makes possible: not "how many people came?" but "how many did the thing, and what sent them?"

This sits one level past the UTM and analytics events articles. Those capture that a visit happened and what the visitor did; conversion tracking is the discipline of marking specific events as goals and connecting them back to their origin.

How the tracking works

Mechanically, a conversion is a special analytics event - one you have flagged as a goal. When the visitor reaches the moment that matters (the order-confirmation page, the "thanks for signing up" screen), a piece of tracking code fires and records it:

// fire when the goal is reached (e.g. order confirmed)
gtag('event', 'purchase', {
  transaction_id: 'T-4821',
  value: 49.00,
  currency: 'USD'
});

Historically this was a "conversion pixel" - a tiny invisible image whose request to an ad platform signalled the goal. Today it is a tag or an event call, but the idea is unchanged: a signal fires at the moment of success, carrying enough detail (value, an order id) to be counted once and tied to money.

Tying it back to the source

A recorded purchase is useless for marketing unless you know what drove it. The link is made by carrying the visit's origin forward from the first click to the conversion. Two mechanisms do it:

  • The session's source, captured from the UTM parameters on the entry link and held through the visit, so the analytics tool already knows this session came from newsletter when the purchase fires.
  • A click identifier - ad platforms append their own IDs (Google's gclid, Meta's fbclid) that get stored and sent back with the conversion, so the platform can match the sale to the exact ad click.

Either way, the conversion arrives labeled with where the visitor came from. And that is where it gets genuinely hard - because a real customer rarely arrives just once.

Attribution models

A buyer might click an ad on Monday, find you via search on Wednesday, and convert from a newsletter link on Friday. Three touches, one sale - who gets credit? An attribution model is the rule that decides:

  • Last-click - all credit to the final touch (the newsletter). Simple, and the long-time default, but it flatters closers and ignores what created the demand.
  • First-click - all credit to the ad that started it. The mirror-image bias.
  • Linear - split credit evenly across all touches.
  • Time-decay - more credit to touches closer to the conversion.
  • Data-driven - a model distributes credit based on the observed contribution of each touch. Now the default in GA4.

The tradeoff to name: the model is not a detail, it changes the answer. The same campaign looks great under first-click and mediocre under last-click. There is no neutral choice - pick the model that matches how you actually believe demand is created, and know that switching models re-writes history.

The privacy shift

Conversion tracking was built on third-party cookies and near-perfect visibility, and that foundation has eroded - by browser restrictions (Safari's ITP, the long decline of the third-party cookie), by regulation (GDPR consent, so tracking only fires if the user agrees), and by ad blockers. The result is that the clean click-to-conversion chain now has gaps.

The industry's responses are worth knowing: server-side tracking (the conversion is sent from your server, not the browser, which is more robust and privacy-controllable), consent-gated tags that only fire with permission, and modeled conversions where platforms statistically estimate the conversions they can no longer directly observe. The trend line is clear: less deterministic tracking, more modeling.

What the numbers are worth

The honest, intermediate stance: conversion data is directionally invaluable and precisely wrong. It will tell you reliably that the newsletter outperforms display ads; it will not give you a to-the-dollar ledger, because attribution is a choice, cross-device journeys break the chain, and privacy gaps mean you never see everything.

So treat conversion tracking as a compass, not an audit. It is the piece that turns the marketing analytics story from "we got traffic" into "we got results, and here is roughly what caused them" - which is the whole point of measuring at all, as long as you hold the numbers with the right amount of humility.

References