Keep bounces out of your bidding

A bounce goal buys more bounces

An ad platform can optimise a campaign on an event you send it. The campaign then goes looking for more people who trigger that event, so the event you pick decides who it finds.

Micro-conversions tempt because they happen often. Here is what each one asks a campaign to find:

• Scroll depth: people who scroll, whether or not they buy.
• Time on page: people who leave a page open on screen.
• Form starts: people who click into a field, whether or not they send the form.
• Bounces: more visitors who leave straight away.

None of these four is a sale or a lead. They show how a page works. Measure them, and bid on something closer to the sale.

Measured for your own reports

Sightwick can measure these signals on your own site:

• Scroll depth, at a quarter, half and three quarters of a page, and near the end.
• Time on page, at 30, 60 and 180 seconds with the page in view.
• Form starts, when someone first clicks into a form.
• Bounces, visits of one page, under ten seconds, with no conversion or step.

By default none of the four goes to Meta, Google Ads, TikTok or ChatGPT Ads as a conversion, and none carries an amount. Each one starts like this:

• Scroll depth and time on page: off on a new site. A tick on the Conversion actions screen turns them on, and they count from that day.
• Form starts: off on a new site, and listed under Not sent to ads on the Conversion actions screen.
• Bounces: on from the start, for visitors who accept analytics cookies. A visit has at most one, and it is listed under Not sent to ads too.

Once measured, they stay in your Sightwick reports: bounce rate under Live events, scroll depth per page under Reading depth. They can also go to GA4. Ask us to set that up.

Counted in reports, not in ads

On the Conversion actions screen, a lead site can set a form start to Counted as conversion. It then counts in your reports and stays Not sent to ads. On a web shop, a form start can't be set that way.

One conversion per visit

When the visitor said yes and an ad platform gets conversions from Sightwick, by default it gets at most one per visit. Over ten days on a Dutch web shop we work closely with, 26 conversions went to Meta and 10 repeat quiz completions were held back, so Sightwick itself never sent Meta more than one per visit. If Sightwick restarts for an update or after a crash, a second one from the same visit can still go out.

When a more frequent event helps

Meta's help pages make one exception: low volume.

An ad set usually leaves the learning phase "after about 50 results" in the week after its last significant edit, Meta says. When an ad set can't get there, Meta advises: "Consider choosing an optimization event that occurs more frequently." Its example: "move from purchases to add to cart."

So a more frequent goal can be right. Bid on the sale or the lead itself while you have the volume. When you don't, pick the step closest to the result:

• In a shop: add to cart or checkout, not scroll depth or time on page.
• On a lead site: a phone click. A form start is one click in a form, and by default Sightwick keeps it out of your ads.

Through Sightwick, these steps can reach Meta as Meta's standard events, when the visitor said yes: AddToCart, InitiateCheckout, and Contact for a phone click.

Shop steps can reach Meta

Sightwick's Whole funnel setting for Meta adds page views, product views, add to cart, checkout and quiz completions, when the visitor said yes. Bounces, scroll depth, time on page and form starts stay out.

• For now, we connect Meta with you on a call.
• On a web shop, a new Meta link starts on Whole funnel with sending on when Sightwick's scan recognised your shop platform, such as Shopify, or read your site and found no Meta pixel. Otherwise sending stays off.
• If your shop's own Meta pixel or app also reports to Meta, Sightwick can't see or limit what that sends, and Meta may count a step twice.
• Want conversions only, or sending on next to your own pixel? Ask us.

What this does not fix

• We have not measured what this does to your cost per result or your return on ad spend.
• Your own visits and test orders count like anyone else's.
• Bounces count only visits with analytics consent, so your bounce rate covers part of your traffic.
• Time on page is not reported per page, and exit intent and rage clicks are not measured.
• Sightwick doesn't touch micro-conversions your own pixel or tag manager already sends.
• The scan reads a limited number of pages and can miss a Meta pixel.

Before you send a micro-conversion

1. Ask what the campaign will look for. It finds more people who do the event, so the event should look like a sale or a lead.
2. Check your volume first. If an ad set rarely leaves the learning phase, Meta advises a more frequent goal.
3. Pick the step closest to the result.
4. If micro-conversions go to GA4, keep them out of Google Ads as conversions.
5. On Meta, check which events your own pixel or tag manager already sends.
6. Keep diagnostics where you read them, in your own reports, and change one thing at a time so you can see what it did.