Custom audiences: building audiences from intent signals

Advertising without knowing who is genuinely ready to convert means wasting a significant portion of your budget on people with no active purchase intent. Custom audiences solve this by directing your campaigns towards people who have already shown measurable interest through their search behaviour, website visits, video interactions or actions within your own channels. In 2026, intent-based targeting is no longer an advanced tactic reserved for large advertisers. It has become the standard approach for any advertiser serious about achieving high ROAS while keeping CPA under control. This article explains what custom audiences are, how to build them from intent signals, and how AdBrains AI technology systematically elevates this process above what manual management can achieve.

What are custom audiences and why does intent matter so much?

A custom audience is a self-built segment that you deploy in Google Ads or Meta Ads. Rather than relying on standard demographic segments such as age or gender, you construct a custom audience from behavioural signals: what has someone searched for, which pages have they visited, which products were viewed, and how far did they progress through the purchase process? All of these behaviours are expressions of intent, the intention to buy, request, or act on something specific.

Intent is the most valuable targeting variable available to advertisers today. Someone searching for "electric folding bike lightweight" or "air conditioning installation quote" sends a clear signal: there is a concrete, active need. Translating these signals into a well-defined custom audience ensures your ads reach people with the highest purchase readiness. That not only improves CTR and conversion rate, but also provides the Smart Bidding algorithm with richer signals to calculate optimal bids. The difference compared to generic targeting is substantial, and as advertising costs continue to rise in 2026, the quality of reach matters as much as the volume of it.

Types of intent signals you can use

Custom audiences can be built from multiple sources of intent data. Understanding which signals are available allows you to choose the right combination for your campaign objective.

• Search behaviour: In Google Ads, Custom Intent Audiences let you build segments based on specific search terms people have recently used. Someone searching for "heat pump installer near me" or "foldable e-bike buy" is clearly in an active consideration phase.
• Website visits: Through remarketing audiences, you target people who have already visited your website, clicked a specific product page, or started a request form without completing it.
• Customer lists: Upload your CRM data as a customer match audience. People who have already been a customer or submitted a lead are valuable signals for both retargeting and lookalike audience construction.
• Video interactions: People who have watched a YouTube video about your product demonstrate engagement, which is a strong intent indicator for Display and YouTube campaigns.
• Lookalike audiences: Based on your best-converting audience, the platform automatically builds similar profiles. On Meta in particular, lookalike audiences remain a powerful prospecting instrument when seeded with high-quality intent data.

The real power lies in combining these signals. By stacking multiple intent indicators, you refine your audience to a group that shows both demonstrable interest and the characteristics of your most valuable customers.

How to build an effective custom audience step by step

1. Define your conversion point: For Clima-Active.nl, this is a completed quote request for air conditioning or heat pump installation. For ToetsJeKennis.nl, it is a completed purchase of an online exam or course. Every audience you build should be oriented towards reaching people most likely to hit this specific conversion point.
2. Collect intent data: Activate conversion tracking and enhanced conversions to capture all relevant signals. Implement a properly configured server-side tracking setup to minimise data loss caused by browser restrictions.
3. Segment by behavioural stage: Distinguish between visitors in the orientation phase (product page viewed, no purchase), the consideration phase (cart added, request started), and high-intent visitors (multiple visits in a short time, long session duration). Each phase deserves its own message and bid.
4. Link audiences to campaigns: Add created audiences to campaigns, both as observation (to collect data without restricting targeting) and as targeting (to reach exclusively that group).
5. Apply Smart Bidding: With Target ROAS or Target CPA as a bidding strategy, Google Ads automatically incorporates audience signals into bid optimisation. The richer your audience data, the better the algorithm can sharpen its bids.
6. Evaluate and refine: Analyse performance per audience segment monthly. Which group converts best? Which segment has too high a CPA? Adjust segment boundaries and bids based on what the data shows.

Custom audiences in practice: e-commerce and lead generation

To understand how custom audiences work in practice, it helps to look at two different business types side by side. For an e-commerce platform like ToetsJeKennis.nl, an online exam and course provider, custom audiences are particularly effective in the final stage of the purchase process. Visitors who viewed a course page but did not complete the purchase form a valuable remarketing audience. Showing them a personalised ad with a direct link to the product they viewed significantly increases the probability of conversion. At the same time, a customer match audience of existing buyers can drive upsell campaigns targeted at people who already purchased one exam and may now be interested in a follow-up course.

For a lead generation advertiser like Clima-Active.nl, priorities differ. Here the goal is not a quick transaction but a high-quality quote request. A custom intent audience built on search terms such as "heat pump installer" or "air conditioning installation cost" attracts precisely those people actively searching for an installation company. Combined with an RLSA audience of people who have already visited the website, this creates a layered targeting structure that reaches both new prospects and re-engages warm leads simultaneously.

The same principle applies to E-4motion.com, the online store for new electric folding bikes. A strong custom intent audience built on searches like "lightweight electric folding bike" or "foldable e-bike buy" combines well with a remarketing audience of people who visited a product page. For test ride requests, a valuable lead generation conversion for E-4motion, a separate audience targeting people who visited multiple product pages without making a purchase performs exceptionally well.

Audience types overview

Choosing the right audience type for each business type and funnel stage is essential. A combination of multiple layers, from prospecting via custom intent to conversion recovery via abandoned cart audiences, consistently delivers the strongest results. This layered approach is precisely what AdBrains automates and continuously optimises through its AI-driven audience management system.

How AdBrains AI takes audience management to the next level

Manual audience management is time-consuming and error-prone. Audiences become stale, segments overlap, bids are adjusted too late, and new intent signals are missed entirely. The AI technology developed by AdBrains is specifically designed to solve these problems structurally, not as a supplement to manual work, but as a fully automated system that operates continuously in the background.

The audience management automation system at AdBrains manages three types of audiences automatically: PROD audiences (production, actively deployed in campaigns), Incubator audiences (new segments being tested before promotion to production), and RLSA audiences (for Search campaigns). Every week, these audiences are automatically created, updated, and linked to the correct campaigns. This ensures an up-to-date audience structure is always active, without requiring manual intervention from an account manager.

AdBrains' Keyword Incubator also plays a direct role in audience construction. New keywords are first tested in a separate incubator campaign. When they generate sufficient intent signals, they are automatically promoted to the production campaign and simultaneously used to enrich custom intent audiences. This creates a self-reinforcing system where better keyword data leads to better audiences, which leads to better Smart Bidding guidance.

The multi-agent verification system at AdBrains monitors every change to the audience configuration. Four independent AI agents check each adjustment before execution. This means errors, such as an incorrect audience link, an overlapping segment causing budget waste, or targeting that is too restrictive and unnecessarily limits reach, are intercepted before they cause damage to the campaign.

Additionally, AdBrains' server-side signal enrichment strengthens audience signals with first-party data. Through a proprietary sGTM infrastructure, conversion signals are enhanced so that the Smart Bidding algorithm can more accurately predict which users have the highest conversion probability. This is particularly valuable for clients like Clima-Active.nl, where the value of a conversion varies significantly by request type and region. The result is a bidding strategy that optimises not just on volume, but on the true value per lead.

Common mistakes when building custom audiences

• Segments that are too broad without qualification: An audience of all website visitors over the past 540 days is rarely useful. Add a minimum session duration, a specific page condition, or an additional behavioural qualifier to improve quality.
• Not excluding existing customers: When running prospecting campaigns, always exclude people who have already converted to avoid wasting budget on existing customers.
• Never refreshing audiences: Intent signals are time-bound. An audience built six months ago but never updated contains many people who have already made a decision. Refresh audience definitions at least monthly.
• Overlapping audiences without prioritisation: If a user falls into multiple audience segments, this can lead to duplicate impressions or inconsistent bids. Define a clear priority order per campaign.
• No distinction between remarketing and prospecting: Remarketing requires a different message than prospecting. Someone who has already visited a quote page does not need an introductory ad, but a concrete reason to convert now.

Frequently asked questions about custom audiences

What is the difference between a custom audience and a standard audience in Google Ads?

A standard audience in Google Ads is a pre-assembled segment created by Google, such as "people interested in fitness" or "homeowners". These segments are broad and based on algorithmically inferred interests. A custom audience is a self-built segment based on specific intent signals, such as particular search terms, visited URLs, or CRM data. Custom audiences are therefore more precise, more current, and better aligned with the specific funnel of your product or service. They also provide richer signals to Smart Bidding, directly contributing to better Target ROAS or Target CPA performance.

How many people does a custom audience need to be effective?

For Search campaigns and remarketing in Google Ads, a minimum of 1,000 active users in an audience is required to activate it. For Display and YouTube the threshold is lower. In practice, we recommend a minimum of 3,000 to 5,000 users per segment for reliable Smart Bidding guidance, giving the algorithm enough patterns to learn from. For niche advertisers or smaller markets, combining segments or using a longer lookback window can help reach sufficient volume.

Can I use custom audiences in both Google Ads and Meta Ads?

Yes, and this is one of the strongest strategies in a full-funnel approach. In Google Ads you use custom intent audiences and RLSA to reach people who are actively searching. In Meta Ads you use custom audiences based on website activity via the Meta Pixel or server-side events, and customer lists for retargeting and lookalike construction. By leveraging the same intent signals across both platforms, you create a consistent brand experience at every touchpoint. Make sure to adapt audience definitions per platform to align with the capabilities and behavioural patterns of each platform's users.

How often should I update my custom audiences?

The update frequency depends on the type of audience and the speed of your purchase cycle. Audiences based on recent website activity with a lookback window of 7 to 30 days update themselves automatically through the audience definition in Google Ads. However, the audience definition itself, meaning which signals you include and which segment boundaries you apply, should be evaluated at least monthly. Customer match audiences based on CRM data deserve an update every time there are significant changes to your customer base, and at minimum every quarter. At AdBrains, audience definitions are automatically checked and updated weekly through our audience management automation system.