The original version of this article from 2014 showed how to use Audience Insights and external tools to find fans of specific pages and turn them into an advertising audience. One of the recommended programs had already stopped working at the time. Today, a similar guide would not merely be outdated. It would lead a company towards a tactic on which neither data work nor marketing should be built.
The principle of precise targeting has not disappeared. It has shifted. It is less about a marketer finding a secret competitor-interest field, and more about the company providing the advertising system with a quality signal, a clear offer, and measurement of the actual result.
The most precise targeting does not begin with a list of interests
Meta now uses substantially more automation. With Advantage+ audience, age, interests, or custom audiences can serve as suggestions, and the system can look beyond them if it expects a better result. Some business boundaries can remain fixed, such as location or the exclusion of custom audiences. The current principle is described directly by Meta in its Advantage+ audience documentation.
That does not mean targeting has stopped being important. It means manually assembling dozens of interests is no longer automatically a sign of work well done. More important is telling the system what result it should seek and from which quality data it can learn.
I divide audiences by their relationship to the company
People who already know the company
Website visitors, people who engage with content, and customers have an existing relationship with the company. Advertising can remind them of an unfinished decision, build on specific content, or present a related offer.
That is not permission to pursue someone with the same banner for whole weeks. The audience needs a reasonable time window, an appropriate message, and the exclusion of people who have already taken the desired action.
First-party customer data
A company can create Customer List Custom Audiences from its own details, such as email addresses or telephone numbers. In its custom audience terms, Meta requires the advertiser to have the necessary rights, permissions, and legal basis to provide and use the data. Uploading a hashed list does not solve that responsibility.
First-party data makes sense when it is current, was collected fairly, and differentiates customer value. A list of every contact a company has collected somewhere over ten years is not a quality signal. It is more likely a future problem.
New audiences
For people with no prior relationship, I test a broader audience against meaningful inputs—such as a customer audience, a quality conversion event, or several genuinely relevant interests. I do not judge which ad set looks smarter. I judge the cost and quality of the result.
A very narrow audience can be expensive and tire quickly. An overly broad audience without enough data, in turn, may take a long time to find its way. Market size, budget, frequency of results, and signal quality are decisive.
Without the right event, you optimise the wrong behaviour
If you optimise a campaign for clicks, the system looks for people who click. If you optimise for video plays, it looks for viewers. But a company usually needs an order, a qualified enquiry, or another step with business value.
I therefore first verify that the chosen event really occurs only when the desired action takes place. A thank-you page available without form submission, or an event triggered on every load, will create attractive but unusable data.
Meta recommends combining Pixel and Conversions API for website events. Server-side delivery can be more resilient to some technical failures and, together with the Pixel, can help measurement. It is not a way to bypass privacy rules or a user's choices; this is explicitly stated in the Meta Conversions API documentation.
On my own website, I also use consistent UTM parameters. The platform, web analytics, and CRM will show different figures because of differing attribution. I need to understand the relationship between them, not force them to match to the last decimal place.
Creative is part of targeting
The advertising system can find a suitable person. But if the advertisement does not say who the offer is for and what problem it solves, it will bring a mix of curious people and random clicks.
Good creative filters out part of the audience by itself. It shows a situation the customer recognises, a concrete benefit, and a reason to believe. For a B2B offer, a calm analysis of the problem can be more effective than a banner with a large discount. For an e-shop, the product, its use, price, and delivery conditions are decisive.
I always test one understandable hypothesis: a different problem, proof, format, or offer. If I change the audience, optimisation, copy, image, and landing page at the same time, I may get a better result. But I will not discover why.
I judge Facebook advertising by economics
A cheap lead is not a win if people do not respond or fail to meet basic conditions. A high return in the advertising account also need not mean profit when the system attributes purchases that would have come even without advertising.
Alongside the cost of the result, I monitor:
- the share of qualified enquiries and closed deals;
- margin or contribution after variable costs;
- the time from first contact to a deal;
- the team's capacity to process new results;
- the difference between acquiring a new customer and the return of an existing one.
Automatic rules can watch budgets and obvious deviations; I describe the principle in the article about automatic rules in PPC campaigns. But a rule cannot replace judgement about the quality of the business.
When it is better not to launch advertising yet
Facebook advertising makes no sense when a company does not know what a conversion is, lacks authorisation to work with uploaded data, has a landing page that does not correspond to the advertisement, or cannot distinguish a qualified enquiry. At that point, more budget only accelerates the collection of bad data.
First review measurement, the website, and business follow-up. If you need to decide how to connect the offer, advertising, and data without further historical tricks, you can see how companies solve similar situations with me.