Custom Intent Audiences Are Now Custom Segments

If you came here looking for custom intent audiences, start with the correction: that name is gone. Google merged custom intent and custom affinity into one audience type called custom segments, and that is the only version of the feature you will find in the interface today. The idea survived. The label did not.

The job is still the same. You describe your ideal customer with keywords, website URLs, and app names, and Google Ads reaches people whose behavior matches that description. What changed, beyond the name, is where you can run it and what the inputs actually do, and that second part is where most eCommerce advertisers still get it wrong. This guide covers the rename, the current capabilities per Google's own documentation, how to build a segment that sells, and where your first-party customer data belongs instead.

What happened to custom intent audiences?

The original split made sense at the time. Custom affinity described broad interests, the kind of person who follows a category. Custom intent described active shopping behavior, the kind of person researching a purchase right now. Two builders, two audience types, two places in the menu.

Google simplified that. Both builders became one, and Google's own naming settled on custom segments, which is the term used across Google Ads Help today. If you inherited an account with an audience named "custom intent," it still runs. It is simply a legacy label on a custom segment.

Why does this matter beyond housekeeping? Because articles that still teach the old split teach differences that no longer exist in the product, and because the merged feature behaves less literally than the old custom intent description suggested. Which brings us to the part worth reading carefully.

What are Google Ads custom segments?

Google describes it plainly: custom segments let you decide how you want to reach your ideal customer by entering keywords, URLs, and apps. You get three input types, and each one means something specific:

• Keywords. Interests, in the form of words or phrases, that represent your ideal customer. Your ads reach people likely to be interested in, or with purchase intent for, those keywords based on their behavior and activity.
• URLs. Website addresses your ideal customer might visit. Your ads reach people who browse websites similar to the ones you enter.
• Apps. App names your ideal customer might use. Your ads reach people who download and use apps similar to those.

The word doing the heavy lifting in the last two is similar. Google is not handing you a list of people who visited a specific site. It is reading your inputs as a description of an interest profile and finding people who fit it.

How custom segments work

Think of the inputs as a brief rather than a filter. You are not writing a query that returns exact matches. You are telling Google what kind of person you are trying to reach, and Google assembles an audience from the behavioral signals available to it.

That has a practical consequence. A custom segment built from tightly related inputs produces a coherent audience. A custom segment built from a grab bag of loosely related keywords produces a vague one, because you have described several different people at once and Google has to average them.

Say you sell running gear. "Trail running shoes," "trail running gear," and "ultramarathon training" describe one person. Add "yoga mats" and "protein powder" to the same segment and you have described three, and none of them precisely.

Where you can use custom segments

This is the second thing the old custom intent framing gets wrong. Custom intent was often described as a Display and YouTube feature with rules about which input types unlocked which channel. The current availability is simpler, and it is worth checking against your media plan before you build anything.

So custom segments are a top and middle-of-funnel prospecting tool. They put you in front of people who have not searched for you, on surfaces where you interrupt rather than answer. That is a real job, but it is not the same job as capturing demand in Search, and the two should not share a success metric.

How to create a custom segment

Google reorganizes its interface often enough that any screenshot-level walkthrough ages badly. The structure below is what holds:

1. Write your inputs before you open Google Ads

   Draft one list of keywords and one list of URLs per intent you want to reach. Doing this in a document first stops you from padding a segment with whatever comes to mind in the interface.
2. Open Audience manager

   Click the Tools icon in your Google Ads account, open Shared library, then Audience manager. This is where audiences are created and stored across campaigns.
3. Create a new custom segment

   Give it a name that states the intent, not the product, for example "researching trail shoes" rather than "shoes 3." Future you, reading a performance report six months from now, needs to know what the segment was supposed to mean.
4. Enter keywords, URLs, or apps

   Add the interests, sites, and apps that describe the person. Keep every input inside one intent. If two inputs describe two different shoppers, that is two segments.
5. Apply it to a supported campaign

   Add the segment to a Display, Demand Gen, Gmail, Standard Shopping, or Video campaign. Give it its own ad group so its performance is readable on its own.
6. Exclude your existing customers

   Apply a Customer Match exclusion so prospecting budget goes to prospects. Paying to reach people who bought last week is the most common quiet waste in these campaigns.

Choosing the keywords and URLs

Keywords

The instinct is to paste your Search keyword list. Resist it. Search keywords capture what someone types when they already know what they want. Custom segment keywords describe an ongoing interest, which is a wider and softer thing.

Build one segment per funnel stage instead of one segment for the whole funnel. Someone reading "how to train for a marathon" is not the same buyer as someone comparing "carbon plate racing shoes," and they should not receive the same creative, the same offer, or the same bid. Split them, then let the numbers tell you which stage deserves the budget.

URLs

Given how URLs actually work, choose them as descriptions rather than as targets. A niche category publication, a specialist retailer, a well-known review site in your vertical: each of those tells Google something clear about the audience you want. A general news homepage tells it almost nothing, because the people who browse it have nothing in common.

You can be granular and point at a specific section rather than a homepage, which sharpens the description. What you cannot do is treat the URL as a guarantee that you are reaching that site's visitors. You are reaching people whose browsing looks like theirs, which is useful, and which is a different promise.

eCommerce plays that still work

Three uses hold up in practice.

Reaching a niche your category segments cannot describe. Google's in-market segments cover broad purchase categories well. They struggle with narrow ones. If you sell left-handed kitchen tools or gear for one specific sport, a custom segment built from the vocabulary your buyers actually use will describe them better than any prebuilt list.

Getting in front of research, not just search. Someone planning a kitchen renovation reads and watches for weeks before typing a product query. Display and YouTube custom segments let you appear during that stretch, which is exactly the window Search cannot reach.

Reaching the audience around your competitors, honestly framed. A competitive analysis that surfaces the keywords and sites your category's buyers gravitate toward gives you excellent custom segment inputs. What it gives you is proximity to that kind of shopper, not access to a competitor's customer list. Expect awareness and consideration lift, and set your measurement accordingly.

One claim from the older playbook deserves retiring alongside the old audience name: custom segments are not an SEO shortcut. Paid impressions can bring traffic to a page while its organic ranking is still weak, which is a fine reason to run them. They do not move you up the results page.

Worth remembering why any of this exists. When Nielsen studied more than 44,000 campaigns across 17 countries in 2016, it found that only 53% of ad impressions served in the UK were viewed by people of the age and gender advertisers intended. Better audience definition was the answer then and it is the answer now. The tools just changed names.

Where your own customer data belongs

This division of labor matters more than it used to. Google removed similar audiences from campaigns in 2023 and replaced them with automated alternatives: optimized targeting for Display, Demand Gen, and Video action campaigns, audience expansion for video reach and consideration, and Smart Bidding for Search and Shopping. Every one of them performs in proportion to the quality of the first-party data you feed it. The lookalike button went away; the value of a good seed list went up.

Which raises the question the old article asked and never really answered: whose behavior should you be mimicking? Not your average customer. Your best ones. Customer Lifetime Value varies enormously between segments, and a seed list built from everyone who ever checked out teaches Google to find more discount hunters.

RFM segmentation gives you the split. Score customers on Recency, Frequency, and Monetary value, then decide which segments deserve to become audiences.

Customer Intelligence in Nexus by Omniconvert builds those segments from your order data and pushes them to Google Ads, Meta, and Klaviyo, refreshing them so lists stay above the 100-member and 540-day eligibility thresholds. Nexus is built on 13 years of data across 7,000+ websites and 15+ industries. For a wider look at which audiences to build across platforms, see our guide to the best eCommerce custom audiences.

Best practices for custom segments

1. One intent per segment, always

   The most common mistake is packing every keyword and URL variation into one audience. It may perform acceptably, but you lose all visibility into which grouping brought the qualified prospects. Treat segments the way you treat Search: test one idea, measure it, keep or kill it.
2. Mine your existing campaigns for inputs

   If a keyword cluster is already producing in Search or Shopping, it is a proven description of a buyer. Reuse it as custom segment input to reach the same kind of person earlier, on Display and YouTube.
3. Seed automation with your best customers, not all of them

   Optimized targeting and Smart Bidding learn from what you give them. A Customer Match list built from high-CLV RFM segments teaches Google to find more of those. A list of everyone teaches it to find your average.
4. Exclude, do not just include

   Existing customers, recent purchasers of the same product, and known low-value cohorts should be excluded from prospecting campaigns. Exclusions often move cost per acquisition more than any new targeting idea.
5. Judge prospecting by the right metric

   These campaigns reach people before they search, so last-click ROAS will always flatter Search and punish Display. Look at new customers acquired, assisted conversions, and cost per new customer instead.
6. Test the page, not only the audience

   A perfectly defined audience landing on an untested page still converts badly. Run the landing page through A/B testing before you scale spend behind the segment.

For the wider paid picture, including campaign structure and feed setup, our guide to Google Ads for eCommerce covers the ground around this one.

Frequently Asked Questions