Custom audiences: building audiences from intent signals

Custom audiences are audiences you build yourself, based on active search behaviour, visited websites or specific keywords, rather than relying on ready-made segments from Google or Meta. A custom intent audience is one of the most precise targeting tools available in Google Ads: you reach people at the moment they are actively searching for exactly what you offer.

Key takeaways

• Custom intent audiences are built on keywords and URLs you provide, targeting high-intent users who are actively comparing or researching.
• Pre-built audiences like in-market and affinity segments are broad by definition; custom audiences give you control over intent depth.
• Choosing the right intent signals determines your CPA and ROAS: not all signals carry equal weight.
• AdBrains automates audience creation, refinement and rotation with a proprietary AI system that continuously learns from conversion data.
• For both e-commerce (like ToetsJeKennis.nl) and lead generation (like Clima-Active.nl), a well-built intent audience delivers structurally better returns.

What are custom intent audiences?

Custom intent audiences are audiences Google assembles based on input you provide as an advertiser: search terms, competitor or industry website URLs, or a combination of both. Google then identifies which users have recently searched those terms or visited those URLs, grouping them into an audience you can activate in Display, YouTube or Performance Max (PMax) campaigns.

The difference from pre-built audiences is fundamental. Google's in-market audiences are defined by Google itself and include everyone who is "likely" interested in a category. That sounds useful, but the definition of "interest" is deliberately broad. A custom intent audience is far sharper: you define the keywords, you define the URLs. The audience becomes smaller but significantly more relevant.

In Meta Ads, the principle is different but comparable. There, audiences are called "Custom Audiences" and you can upload your own customer list, target website visitors via the Meta Pixel, or create lookalike audiences based on that first-party data. Google Ads offers a similar mechanism through Customer Match, which lets you upload your own customer lists to build audiences directly, complementing the search-term and URL-based custom intent audiences described above. Intent as a signal on Meta is inferred from behavioural patterns and interests rather than active search queries, but the core idea of "your own data as the foundation" remains the same.

Pre-built audiences versus custom intent audiences

Pre-built audiences are not useless, but they have a clear limitation: you have no influence over who is in them. Google decides the composition. For large brand-awareness campaigns, or when introducing a new product without your own conversion data, in-market audiences can be a reasonable starting point. Once you have your own conversion data, however, building custom intent audiences is almost always the better move.

ToetsJeKennis.nl, an online exam and course platform, uses pre-built in-market audiences during the early phase of a new course category to gather data quickly. Once a campaign has enough conversion data, the system switches to a custom intent audience built from terms like "practice exam online" and "buy course Netherlands", supplemented with URLs from comparable e-learning platforms.

Intent signals: Google Ads vs. Meta Ads

Google and Meta use fundamentally different models for intent-based targeting. Google Ads is by nature an intent-first platform: search queries are the strongest intent signals that exist. Someone typing "buy electric folding bike" is actively looking to purchase. E-4motion.com, the webshop for new electric folding bikes, benefits directly: Search campaigns capture the explicit demand, while custom intent audiences in Display re-reach that warm audience across subsequent websites they visit.

Meta Ads works differently. Intent is inferred from behaviour, interests and social signals. You cannot enter search terms as audience seeds, but you can:

• Upload a customer list (email addresses from your CRM) as a Custom Audience, then build a Lookalike Audience from it.
• Segment website visitors via the Meta Pixel: people who visited a product page without purchasing, abandoned their cart, or spent more than 60 seconds on a page.
• Create engagement audiences based on video views, lead form completions or page followers.
• Target specific URL visitors who browsed multiple subpages, which is a strong intent signal.

For Clima-Active.nl in a Meta context, this means: everyone who visited the quote landing page but did not submit the form receives a separate ad with a different angle, such as a customer testimonial or a message about heat pump subsidies. That audience is small but has demonstrated interest, which structurally lowers CPL.

How to select powerful audiences for your campaigns

The question of how to easily select powerful audiences for your campaigns starts with intent depth. Not every signal carries equal weight. A user searching "electric folding bike buy compare" is far further down the funnel than someone who once read an article about sustainable transport.

For Google Ads, these are the steps for building a well-structured custom intent audience:

1. Collect your highest-converting search terms from your Search campaigns via the search term report. These are terms that have already proven to work.
2. Add competitor URLs: include the domains of direct competitors or industry comparison sites. Users who visit those sites show active purchase intent.
3. Exclude irrelevant segments: add negative keywords to your campaign and exclude audiences (e.g., existing customers or recent converters) so budget is not wasted on people who already converted.
4. Apply audience layering: combine a custom intent audience with demographic filters or location targeting for additional precision.
5. Test and refine: use A/B segments to identify which keyword combinations produce the lowest CPA, then update the audience based on conversion data.

How AdBrains automates this with AI

Building, refining and maintaining audiences manually takes time that most advertisers simply do not have. The composition of an intent audience changes constantly: terms that convert today may not perform three months from now. AdBrains has built a proprietary AI system that manages the full audience lifecycle, from initial build to continuous optimisation.

It starts with automated search term mining. Every day, the AdBrains AI analyses all incoming search terms from Search campaigns. Terms that convert are automatically tagged as potential intent signals for Display and PMax audiences. Terms that are irrelevant or have produced no conversions after a statistically meaningful number of clicks are added as negative keywords. This runs continuously, without manual intervention.

The audience management automation system then takes over. AdBrains automatically creates new audience segments every week: a PROD audience based on proven converting signals, an Incubator audience for new, promising terms still being tested, and an RLSA audience for remarketing. These three layers work together: the Incubator feeds the PROD audience once a signal has proven its performance.

For bidding strategy, AdBrains uses its automatic tCPA/tROAS optimisation system. Audiences with a historically lower CPA or higher ROAS automatically receive higher bid modifiers, while underperforming audiences are scaled back or paused. This operates at audience level, not just campaign level, which makes a meaningful difference in steering precision.

Server-side signal enrichment also plays a role. Via AdBrains' own sGTM infrastructure, conversion signals are enriched with first-party data before being sent to Google's Smart Bidding. That means the Smart Bidding algorithm receives higher-quality signals to optimise on. For custom intent audiences, this is particularly valuable, because the quality of the conversion signal directly determines how sharp the audience match becomes.

The multi-agent verification system monitors every step: four independent AI agents check every audience change before it goes live. This prevents a poorly configured audience from running without review. In our practice, this verification process adds the most value during larger account changes, where an error in audience setup can quickly lead to wasted budget. For clients like ToetsJeKennis.nl and E-4motion.com, this means audiences are always current, always based on the most recent conversion data, and always set to the correct bid level.

From intent signal to conversion

The funnel above shows how a custom intent audience performs in a Display campaign. Relevance is high, so CTR is already meaningfully better than with generic Display targeting. Where things really diverge, though, is after the click.

Users who arrive through a high-intent audience tend to browse more pages and bounce less. They know what they want and recognise the offer when they land on it. That recognition translates directly into a higher conversion rate. To put some numbers around it: take a budget of €1,000 per month and a CPM of €3. That gets you roughly 333,000 impressions. At a CTR of 0.8%, you are looking at 2,667 clicks. Apply a 12% conversion rate on engaged sessions and you end up with approximately 320 conversions. Run that same €1,000 through a generic in-market audience at a 3-4% conversion rate and the result is somewhere between 80 and 107 conversions. Even allowing for small fluctuations in CPM or CTR, the gap is hard to ignore.

Practical tips for stronger intent audiences

• Start with your own search terms: download the search term report for the past 90 days and filter for terms with the lowest CPA. These are your strongest seeds.
• Add industry URLs: not just competitors, but also comparison sites, trade publications or review platforms where your audience is active.
• Exclude converters: build a separate audience of recent buyers or enquirers and exclude them from prospecting campaigns. Otherwise you pay to reach people who are already customers.
• Use observation mode first: add audiences in observation mode to collect data without restricting your targeting. Switch to targeting mode once you know which audience performs best.
• Combine signals: a custom intent audience layered with a remarketing segment (people who already visited your site) gives the highest relevance and typically the lowest CPA.

FAQ on custom audiences and intent targeting

What is the difference between a custom intent audience and an in-market audience?

A custom intent audience is built by you, using search terms and URLs you supply. An in-market audience, by contrast, is a ready-made segment that Google assembles around broad category interest. Custom intent audiences are more specific and give you considerably more control, which is why they tend to perform better on conversion-focused campaigns. In-market audiences can get you off the ground quickly when you have no conversion data of your own, but you have almost no say over the precise intent level of the users Google puts in that segment.

How do I easily select powerful audiences for my campaigns?

Pull up the search term report from your active Search campaigns. Filter for terms that drove the most conversions at the lowest CPA over the past 60 to 90 days, then use those as seeds for a custom intent audience in Display or PMax. Add URLs from competitors or industry-relevant sites while you're at it. In roughly an hour you have an audience grounded in actual, proven interest rather than guesswork. Start in observation mode first so you can validate performance before you restrict targeting.

Do custom intent audiences work for small budgets?

Yes, and on smaller budgets they can matter even more, precisely because they cut waste. You reach fewer people overall, but those people are far more relevant. For a lead generation campaign like LeroyBrouwer.nl, a tight, well-segmented audience will often lower CPL even when absolute reach shrinks. One thing to watch: your audience still needs to be large enough for Smart Bidding to learn from. Google recommends at least a few thousand users for dependable algorithmic steering (Google Ads Help, 2026).

Can I use custom intent audiences in Performance Max?

In Performance Max (PMax) the mechanics are different from traditional Display or YouTube campaigns. You don't add audiences as a direct targeting criterion. Instead you provide them as "audience signals," which are essentially hints you give Google about which users are likely relevant, after which the algorithm decides who it actually reaches. Custom intent audiences work well as signals in PMax because they hand Google's AI strong, first-party-driven intent information. That tends to shorten the learning phase and noticeably improves early targeting quality.