Retargeting Ads Explained: How Your Product Follows Customers Across Facebook, Instagram and Google

The shop window that follows you home

Picture a furniture shop on the high street. Someone walks in, looks at a green sofa for five minutes, checks the price tag, and leaves. In the physical world, that is the end of it. The shop never sees them again unless they come back on their own.

Online, it does not have to end there. If the shop's website is set up correctly, that visitor can see the same green sofa the next day while scrolling Facebook, again on Instagram over the weekend, in a short advert before a YouTube video, and on the side of a news site. Same person, same product, four different places.

That is retargeting. You have almost certainly been on the receiving end of it. Most people who see it assume it takes a big brand with a big budget. It does not. A dentist, a kitchen fitter, a language school or a small online shop can run it with the same tools the big brands use.

And it matters, because very few people buy on their first visit. They compare, they get distracted, they wait for payday, they ask their partner. Retargeting is how you stay in the picture while they make up their mind.

Retargeting or remarketing?

For you as a business owner, they mean the same thing. Google has always called it remarketing (in Google Ads it is now listed as "your data segments"). Meta, which runs Facebook and Instagram, calls the lists Custom Audiences. Most people in the industry just say retargeting. Do not let the vocabulary put you off: it is one idea with three names.

How it actually works, in four steps

Here is the mechanism without the jargon.

1. Your website takes a note

A small piece of code from each ad platform sits on your website. Meta's is called the Pixel. Google's is the Google tag. When someone visits, the code sends a short message to that platform: someone looked at the green sofa page, or someone put the sofa in the basket and left.

2. The platform recognises the person

Meta and Google then try to match that visit to one of their own users: the person logged into Facebook, Instagram, Gmail or YouTube on the same phone or computer. If they can match it, the person goes onto a list. You never see who they are. You only see that your list has, say, 2,400 people on it.

3. You decide who gets which ad

This is where it gets useful. You can make different lists for different behaviour and show each one a different message:

• Looked at a product: show them that product, with a review from a happy customer.
• Put something in the basket but did not pay: remind them, maybe mention free delivery.
• Already bought: stop showing them the same product (nothing annoys people more), and show them something that goes with it instead.

4. The ads appear wherever those people spend time

Meta shows them across Facebook, Instagram and Messenger. Google shows them on YouTube, in Google search results, in Gmail and across the websites in its display network. People stay on a list for as long as you choose, within limits: Meta allows up to 180 days for website visitors, and audiences built in Google Analytics can last up to 540 days.

Why it quietly stopped working as well as it used to

Everything above depends on step one: the message from your website actually reaching Meta or Google. Over the last decade, more and more of those messages have been getting lost on the way. Not because anyone at your business did something wrong, but because the browsers and phones in between changed the rules.

2017: Safari starts shortening memories

In June 2017, Apple announced Intelligent Tracking Prevention (ITP) for Safari, the browser on every iPhone, iPad and Mac. It has been tightened many times since. Today, according to WebKit's own documentation, Safari:

• blocks all third-party cookies, with no exceptions,
• deletes cookies created by website code after 7 days without the visitor interacting with your site,
• and when someone arrives from a click on a tracking-heavy site with a tracking code in the link, limits those cookies to 24 hours.

In plain English: someone taps your Facebook ad on their iPhone on Monday, browses, and does not come back until the following week. As far as Safari is concerned, the trail may be gone. They might drop off your retargeting list, and if they do buy later, the purchase may not be credited to your ad.

This matters more than most people think. In the UK, Safari accounted for roughly 45% of mobile browsing in mid-2026, according to Statcounter. Nearly half of your mobile visitors are on the browser that is hardest on tracking.

2021: iPhone apps have to ask permission

With iOS 14.5 in 2021, Apple introduced App Tracking Transparency: every app has to ask before tracking you across other companies' apps and websites. Many people tap "Ask App Not to Track". The effect on advertising was big enough that Meta told investors in February 2022 it expected the change to cost it around 10 billion dollars in revenue that year.

Every day: ad blockers and cookie banners

Ad blockers stop the tracking code from loading at all. And in the UK, you need people's consent before using advertising cookies and similar technologies, under the PECR rules the ICO enforces. People who click "Reject" should not be tracked, and they will not be. That part is not a problem to be fixed. It is the law, and it is fair.

One note on Chrome: Google planned for years to remove third-party cookies from Chrome, then dropped the plan in 2024 and confirmed in April 2025 that it would not roll out a separate prompt either. So Chrome is not the main problem today. Safari, iPhones, ad blockers and consent are.

What the losses cost you

When the messages get lost, three things go wrong at once:

• Your retargeting lists are smaller than they should be. People who were interested never get the reminder.
• Your reports look worse than reality. Sales happen, but the ad platform cannot see that your ad caused them.
• The algorithm learns the wrong lesson. Meta and Google decide who to show your ads to based on who bought before. If they only see part of your sales, they learn from part of your customers, and spend your money on the wrong people.

The third one is the expensive one, and it is invisible unless you know to look for it.

The fix: server-side tracking, explained with a post room

Think of the normal setup like this. Every visitor's browser is a messenger that carries letters from your website straight to Meta, Google and TikTok. On the way, Safari, ad blockers and privacy settings intercept a lot of those letters, because they recognise the addresses as advertising companies.

Server-side tracking changes the route. The browser now takes one letter to your own post room: a small server that sits on your own website's address, for example data.yourshop.co.uk. Your post room then forwards the right information to Meta, Google and the others, server to server. The browser only ever talks to your own website, so much less gets lost along the way.

The platforms built official doors for exactly this:

• Meta Conversions API for Facebook and Instagram,
• Enhanced Conversions and server-side tagging for Google Ads,
• and equivalents for TikTok, LinkedIn, Microsoft and others.

Meta's own advice is to run its Pixel and the Conversions API together, not one instead of the other. The two report the same events, Meta removes the duplicates, and what one misses the other often catches.

There is a second benefit. Because the information passes through your own server first, you decide what leaves it. You can strip out anything you do not want to share before it reaches a third party. That is control you do not have when every visitor's browser talks to the platforms directly.

So does it make advertising cheaper?

This is the question every business owner asks, so here is the honest answer.

It does not lower the price of a click directly. Clicks are priced in an auction, and better tracking does not change the auction. What it changes is who your budget is spent on. When Meta and Google can see more of your real sales, their algorithms find more people like your real buyers, and you pay for fewer clicks that go nowhere. The cost of each sale or lead goes down, which is the number that actually matters.

The published numbers come from the platforms themselves, so read them as indications, not promises:

• Meta reports that advertisers using the Conversions API together with the Pixel saw on average a 13% improvement in cost per result, based on 28 tests it ran between May and August 2022.
• In Google's early studies of Enhanced Conversions in 2021, the median conversion rate rose by 5% on Search and 17% on YouTube, because more of the conversions that were already happening became visible.

Your own result depends on how much data you are losing today. A business whose customers are mostly on iPhones usually gains more than one whose customers are mostly on Android desktops.

The catch most setups miss

Apple knows about server-side tracking too. According to WebKit's documentation, if your "post room" is really someone else's server dressed up with your web address (the technical terms are CNAME cloaking and third-party IP cloaking), Safari spots it and limits the cookies it sets to 7 days. Plenty of quick, cheap setups fall into this trap and only recover part of what they could.

Getting the full benefit means the server has to genuinely belong to your website as far as Safari can tell. This is one of the details we check first when we review an existing setup.

What server-side tracking is not

It is not a way around cookie consent, and anyone who sells it that way is putting you at risk. The rules on consent apply no matter which route the data takes. A good setup reads your cookie banner's answer and respects it on the server too. If someone says no, nothing personal about them goes to the ad platforms. We wrote more about this in our piece on Google's upcoming IP address change.

How we set it up (without giving away every secret)

Every business is different, but the shape of the work is always the same:

1. Audit. We compare what your shop, booking system or CRM says actually happened with what Meta, Google and Analytics think happened. The gap tells us how much there is to recover, and whether the project is worth it.
2. Consent first. We make sure the cookie banner works properly and that its answers reach every tag, in the browser and on the server, including Google's Consent Mode.
3. Your own server, on your own address. A server container on your domain, set up so Safari treats it as part of your website, not as an outsider.
4. Connect the platforms. Meta Conversions API, Google Ads Enhanced Conversions, and whichever other channels you use, with duplicates removed so nothing is counted twice.
5. Build the audiences. Retargeting lists for each stage (looked, nearly bought, bought), with buyers excluded where they should be, and sensible limits on how often one person sees an ad.
6. Watch it. Alerts when the numbers drop, because tracking breaks silently when a website or a platform changes.

For businesses that do not want to run their own server, we operate a managed server-side tagging platform that does the infrastructure part for you.

When retargeting is not worth it yet

• Very low traffic. If only a few dozen people visit your website a month, your lists will be too small for the platforms to show ads reliably. Spend on getting visitors first.
• One-off, instant purchases. If people decide in seconds and rarely come back (an emergency locksmith, say), reminders help less than being found at the right moment.
• A website that does not convert. Retargeting brings people back. If the page they come back to is slow or confusing, you pay twice to lose them twice. Fix the page first; our conversion rate guide is a good place to start.

trendfingers sets up retargeting and the measurement underneath it: server-side tracking, Meta Conversions API and Google Ads Enhanced Conversions. If you are not sure how many of your visitors your ads can actually see, that is the first thing an audit answers. Request an audit.