17 event types, zero tags written

One tag and trigger per conversion

In Google Tag Manager, a Google Ads conversion is a tag of its own. You then choose the triggers that decide when it fires.

Server-side Tag Manager works the same way. Google's setup guide for Google Ads there has you select a trigger for the tag.

In both, someone first decides which forms and links count, then sets up each one.

An automatic tracking audit starts from the other end. It reads the site first and proposes what to measure. You decide what stays.

A draft plan from your pages

Sightwick's conversion tracking starts with a scan. It opens your site in a real browser and, by default, reads up to 25 pages.

You enter your website when you sign up, and the scan runs in the background while you add the script.

From your home page it goes to pages like contact, quote, demo and pricing first. On each page, 14 detectors look for:

• forms, and phone, email and WhatsApp links
• booking widgets and chat tools
• quizzes and downloads
• shop events, from a product view to the order

Evidence and a score per finding

Each action names the page it was found on. Under Evidence you see the HTML of that form or link, the detector that found it and how often it appeared. One finding from our own site:

• Contact form on /call
• Form detector, score 0.5, found once
• <form class="form" method="post" action="/aanvraag" novalidate="">, with a field labelled Domain

The score runs from 0 to 1, and the screen shows it as a percentage. Below 0.75, the action gets a May break note. Below 0.5, it stays in the list switched off, for you to tick.

The score is a rule of thumb that each detector sets for itself. Read May break as a warning, not a prediction.

Three conversion types on our site

We scanned our own site, sightwick.com. On 21 September 2026, the scan read 9 pages in 29 seconds. Our pages have changed since, so this is that day's scan.

It found three conversion types: a contact form, a trial sign-up form and email links. It added five supporting signals and set six types of noise apart.

We tagged none of them by hand. Our site loads the Sightwick script and no other tracking tag.

The scores show where it was unsure. Our call page asks for a call, yet the scan filed its form as a contact form, at 0.5. A form error it saw scored 0.45, below the line for switching on, so it stayed off.

Bookings and chats, four tools each

The scan recognises 8 booking widgets and 13 chat tools. Most of those it can find but not count.

Since 25 September 2026, the script counts these by itself, once the scan has found them:

• a booking in Calendly, Cal.com, HubSpot or SavvyCal
• a started chat in HubSpot, Drift, Tawk.to or Trengo

The scan does not find a widget that only loads after a click, like the calendar on our own call page.

Typeform, Intercom and video players are found by the scan, but not counted by themselves. Integrations shows which tools count on their own.

17 event types on one shop

We also scanned a Dutch web shop we work closely with. It runs on Shopify. It is not an independent case study.

On 23 September 2026, the scan read 25 pages in eight minutes. It made 300 findings, set 150 apart as noise and sorted the rest into 17 event types:

• 5 proposed as conversions: Contact form, Email click, Trial signup, Purchase and Subscription started.
• 3 steps before the order: Add to cart, Checkout started and Quiz completed.
• 9 signals: product views, cart views, removals from the cart, site search, opened FAQ items, form starts, form errors, quiz starts and clicks on call-to-action buttons.

Nobody wrote a tag or a trigger for any of them.

When a shop connects Meta, page views, product views and the three steps can go there as events, not as conversions, when the visitor said yes.

Scanned before the shop rule

After this scan, Sightwick got a rule for shops. Only Purchase and Subscription started begin as conversions. Contact form and Email click become signals, and Trial signup starts switched off.

The rule applies to new findings. This shop kept the list from its scan.

Discount forms became trial sign-ups

That Trial signup was a mistake. The scan took two discount forms, in the footer and in a pop-up, for trial sign-ups.

Since 24 September 2026, the scan files both under Newsletter signup. But a rescan keeps an action it no longer finds.

When we checked on 5 October 2026, Trial signup was still switched on there, on the footer form, and marked to go to ads. Email click was marked to go to ads too. On that shop, both have to be changed by hand.

On the same shop, 79 of the 150 findings not set apart as noise scored below 0.75, the May break line.

That is why the list is a proposal. Check it before you connect an ad platform.

Where the scan stops

• The scan runs when you sign up and again every week. A form you add in between is not in the list until then.
• By default it reads up to 25 pages, and a page it did not read is not in the list.
• It skips pages your robots.txt closes to bots.
• It does not test whether the tags you already have fire.
• Counting needs the script on your site. Installing is one line in the head; Shopify shops also add one custom pixel under Customer events. Whoever can edit your site's head or GTM can do it.
• Automatic events are not new: GA4's enhanced measurement records a fixed set without code changes.

Checks before you trust the list

1. Start with everything marked May break.
2. Open the Evidence on each conversion. Check that the HTML is the form or link you expect.
3. Untick anything that is not a real conversion for you.
4. Look for forms on pages the scan did not read. If one has a thank-you page, add it with Add a page; if not, book a call.
5. After you install, submit one test form and find it in Live events.
6. Only then connect your ad platforms. Sending can start as soon as you do.

Nothing goes to an ad platform before you connect one, and then only when the visitor said yes.