A website custom audience is a group of people Meta has matched to visits on your website, built from the data your pixel or server sends back, so you can reach them again with Facebook and Instagram ads. You set the rules: everyone who visited in the last 30 days, people who looked at one particular service page, or shoppers who started checkout and did not finish.
How a website custom audience works
When someone loads a page carrying the Meta Pixel, their browser sends Meta an event such as PageView, ViewContent, Lead or Purchase, along with the page address. If Meta can connect that browser to a Facebook or Instagram account, the person becomes eligible for any audience whose rules they meet. Events sent from your server through the Conversions API can add people the browser alone would miss, provided they carry enough customer information for Meta to make the match.
You create the audience in the Audiences section of Ads Manager by choosing the data source, a rule and a retention window. Rules can cover all visitors, visitors to URLs containing a phrase (for example “/kitchen-fitting/”), the most engaged visitors by time spent, or people who triggered a specific event. You can stack include and exclude conditions, such as “viewed a product, but did not purchase”.
Retention can run up to 180 days. Each person drops out once their last qualifying visit is older than the window you chose, so the list keeps itself current without any work from you. You never see who is in it; Meta shows only an estimated size.
Why it matters
People who have already been on your site know your name and have shown some interest, so they tend to respond to ads at a lower cost than strangers. Website custom audiences are the raw material of Facebook and Instagram retargeting: reminding someone who priced a boiler service last week, or showing a basket reminder to a shopper who left without paying.
They work just as well as exclusions. Leaving recent customers out of a “first order” offer stops you paying to show a discount to people who have just bought at full price. And a high-intent audience, such as purchasers or people who sent an enquiry, makes a far better seed for a lookalike audience than everyone who ever landed on the homepage.
UK sites have an extra constraint. Under PECR the pixel counts as a non-essential tracker, so it should only load after a visitor accepts it through your cookie banner. Anyone who declines never joins the audience. Add Apple’s tracking limits on iPhones and the spread of ad blockers, and a UK website audience is often much smaller than the visitor numbers in GA4. That gap is normal, but it should shape how finely you slice your audiences.
Common mistakes
- One giant “all visitors, 180 days” audience. Someone who skimmed a blog post five months ago gets the same ad as someone who reached checkout yesterday. Split by intent and by recency.
- No exclusions. Existing customers keep seeing introductory offers, which wastes budget and irritates them.
- URL rules that no longer match. A rebuild changes “/services/roofing” to “/roofing/” and the audience quietly stops filling. Recheck every rule after a URL change.
- Audiences too small to deliver. A rule for one niche page with a seven-day window may hold so few people that the ad set barely spends. Lengthen the window or group related pages.
- Loading the pixel before consent. It adds a few people to the audience and puts the business in breach of PECR. The trade is not worth making.
How to act on it
Start with the data. In Events Manager, confirm the pixel receives PageView and your key events from the pages that matter, and that it fires only after consent. Then build a short set of audiences that mirror how people buy from you. For a service business that might be service page visitors in the last 30 days, people who reached the contact page without submitting, and enquirers from the last 180 days to exclude. For a shop: product viewers, basket abandoners and purchasers.
Give each audience a message that picks up where that person stopped, and keep an eye on frequency, because a small warm audience tires of one ad quickly. Check sizes monthly. When an audience shrinks suddenly, the cause is usually tracking or a changed URL rather than a fall in demand.