Conversion tracking is how you know which visits, campaigns and product changes lead to the outcomes you care about: a sign-up, a purchase, a booked demo, a paid upgrade. Without it, marketing spend and product work are judged on traffic and opinions. With it, you can say which channel brought customers and which step lost them.
This guide covers what conversion tracking means, the kinds of conversion worth tracking, how to set it up step by step, client-side versus server-side tracking, how ad platform conversion tracking fits in, and the mistakes that make conversion numbers wrong.
What does conversion tracking mean?
A conversion is a visitor or user completing an action you've defined as valuable. Conversion tracking records each conversion with enough context, such as the source, campaign, page and person, to attribute it and count it correctly.
The conversion rate is conversions divided by the people who had the chance to convert:
Conversion rate = conversions ÷ eligible visitors or users × 100
If 4,000 people visited your pricing page last month and 120 started a trial, the pricing-to-trial conversion rate is 3%. The conversion rate calculator does the arithmetic, and the conversion rate glossary entry covers variants.
Which conversions should you track?
Track a small set at three levels:
| Level | Examples | Who uses it |
|---|---|---|
| Macro conversions | Paid subscription, placed order, signed contract | Leadership, finance |
| Primary conversions | Sign-up, trial started, demo booked | Marketing, growth |
| Micro conversions | Viewed pricing, started checkout, created first project | Product, growth |
Macro conversions are what the business needs; micro conversions show the path to them and move fast enough to learn from. Our article on micro conversions explains how to choose them.
How to set up conversion tracking: step by step
1. Define each conversion precisely
"Sign-up" could mean a form submitted, an account created or an email verified. Pick one, write it down, and fire the event only when that thing has actually happened.
2. Decide where each event is sent from
- Server-side for conversions that matter: account created, payment confirmed, order placed. Your backend knows for certain that they happened.
- Client-side for interactions only the browser sees: clicked a pricing button, opened a demo video.
3. Capture the source
Record where visitors came from: referrer, UTM parameters and landing page. Tag every campaign link consistently; the UTM builder helps. Link the anonymous visit to the user when they sign up, so a conversion days later still knows its source.
4. Build the funnel
Conversion is rarely one step. Build a funnel from first visit to the macro conversion, counted per person in order, within a window. The funnel shows both the overall rate and the step where people drop.
5. Verify before you trust it
Trigger each conversion yourself and confirm it arrives once, with the right properties. Then compare a week of conversion counts with your source of truth, such as orders in your store or accounts in your database.
Conversion windows: how long should you wait?
A conversion window is the time allowed between the first step and the conversion for it to count. It changes the number dramatically, so choose it on purpose:
- Within a session or a day suits impulse actions: newsletter sign-ups, adding to cart, starting a free tool.
- Within 7 days suits most self-serve sign-up and activation flows.
- Within 30 days or more suits considered purchases: B2B trials, demos and annual plans, where people research, leave and return.
Look at the distribution before deciding. If most people who convert do so within two days but a long tail takes three weeks, a 7-day window undercounts the tail and a 30-day window delays every reading. Many teams track both: a short window for fast feedback on changes, and a long one for reporting.
Whatever you choose, keep it stable and print it on the chart. A conversion rate without its window can't be compared with anything.
Client-side vs server-side conversion tracking
| Client-side | Server-side | |
|---|---|---|
| Sent from | The visitor's browser | Your backend |
| Blocked by ad blockers | Often | No |
| Lost when a tab closes | Sometimes | No |
| Knows the outcome succeeded | No (only that a button was pressed) | Yes |
| Knows the referrer and UTM | Yes | Only if you pass it along |
The robust setup uses both: the browser records the visit and its source, the server records the confirmed conversion, and an identify call joins them. Real-time event tracking covers server-side sending in more detail.
Conversion tracking in Google Ads and LinkedIn
Ad platforms have their own conversion tracking: a tag or API that reports conversions back so the platform can optimise bidding and report return on ad spend. It answers "which ads drove conversions?" from the platform's point of view.
Product analytics conversion tracking answers a broader question: "which of all our sources, including organic search, referrals and AI assistants, drove conversions, and what did those people do next?" The two complement each other. Keep the ad platform's tracking for bidding, and use your own events to compare channels on equal terms and to follow converted users into retention.
LogZuno doesn't send conversions back to ad platforms. It records conversions as events with their source, so you can compare every channel side by side.
The new blind spot: AI assistants
A growing share of visitors arrive from ChatGPT, Perplexity and other assistants, and many of those clicks carry no referrer. They land in Direct, so conversions they drive are credited to nobody. Separately, the assistants read your pages from their servers without running any tag, so your conversion tracking never sees the research that led to the visit. See ChatGPT traffic in Google Analytics and AI visibility.
Mistakes that make conversion numbers wrong
- Counting bots. Automated browsers fill the top of the funnel and never convert, so rates look worse than they are. Count people, not page loads; see bot traffic in GA4.
- Double-counting. A thank-you page reloaded twice is two conversions. Use an idempotency key such as the order ID.
- Firing on clicks. A click on "Buy" is intent; the confirmed order is the conversion.
- Losing the source at sign-up. If the visit isn't joined to the user, every conversion looks like Direct.
- Dividing totals. A conversion rate from two monthly totals includes people who converted from an older visit. Use a per-person funnel.
Conversion tracking with LogZuno
LogZuno records the visit and the conversion in one project:
- The script tag records pageviews with referrer, UTM campaign and landing page, and counts visitors as people only after a real interaction.
- Your backend sends conversion events, such as Signed up or Placed order, with one HTTP request.
- Funnels follow each person from a page or click to the conversion, within the window you set.
- AI assistants show as their own channel, including likely ChatGPT visits that arrived without a referrer.
- Alerts tell you when a conversion step breaks.
Start with web analytics and the quickstart. Marketing teams can see the full setup in LogZuno for marketing teams.
Key takeaways
- Define each conversion precisely and track macro, primary and micro levels.
- Send confirmed conversions from your server; record the visit and source in the browser; join them.
- Use per-person funnels, count people not bots, and deduplicate with IDs.
- Keep ad platform tracking for bidding; use your own events to compare every channel fairly.