TL;DR
- Google published a Help Center article on 1 September 2026 extending Lookalike segments to Video brand campaigns with three tiered reach controls.
- The tiers are Narrow at 2.5 percent of users in the target location, Balanced at 5 percent as the default, and Broad at 10 percent.
- In Video campaigns those percentages are hard caps, unlike Demand Gen, whose Lookalikes became directional audience suggestions in March 2026.
- Seed lists need more than 100 active matched people, while Google documentation elsewhere recommends at least 1,000 active users for quality.
- Hana Kobzova discovered it and PPC Land reported it on 6 September 2026, with no blog post, no advertiser email and no performance data.
Google has extended Lookalike segments to Video brand campaigns with three tiered reach controls, Narrow at 2.5 percent of users in the target location, Balanced at 5 percent as the default, and Broad at 10 percent. In Video campaigns those percentages work as hard caps on audience size, which is the same tiered structure Demand Gen lost in March 2026 when Google turned its Lookalikes into directional audience suggestions instead.
What Google published, and on what date
Google published a Help Center article on 1 September 2026 titled "Use Lookalike segments in Video campaigns to grow your audience", setting out how Lookalike segments work in Video brand campaigns and naming three reach settings with fixed percentages attached. The change was discovered by Hana Kobzova and reported by PPC Land on 6 September 2026. There was no blog post and no advertiser email, so a documentation page is the entire public announcement. That matters for how much weight to put on it: the percentages and the seed list numbers are stated, and everything an advertiser would want beyond that is not.
What a Lookalike segment is and how a seed list becomes one
A Lookalike segment starts with a seed list, which is a group of people an advertiser already has, and Google builds an audience of people who resemble that group. The seed list is the input that decides what the resulting audience looks like, so the quality of the segment is downstream of the quality of the list. Google states the seed list requirement for this as more than 100 active matched people. Two words in that phrase do the work. Matched means people Google could actually recognise, not everyone in the source file. Active means people Google still counts as present, not people who were once on the list. A list can be large and still fail both tests, which is why the stated minimum is a floor for eligibility rather than a target to aim at.
A hard cap and a directional signal are different instruments
This is the part that decides whether the feature is useful for planning. A hard cap is a boundary the system will not cross: set Narrow and the audience does not exceed 2.5 percent of users in the target location, so an upper bound on reach is known before the campaign runs. A directional signal is an input the algorithm weighs alongside everything else it uses, with no ceiling attached, so the eventual audience size is an outcome rather than a setting. Google converted Demand Gen Lookalikes from hard targeting constraints into directional audience suggestions fed to the algorithm in March 2026. In Video campaigns the percentages function as hard caps on audience size. So the same words, Narrow, Balanced and Broad, describe two different kinds of control depending on the campaign type, and only one of them lets a planner say in advance how far the campaign can reach.
The three reach tiers, and what each one caps
The table below lists exactly the three Lookalike segment reach tiers Google stated for Video campaigns, against the share of users each one allows, with the Demand Gen position since March 2026 for contrast.
| Setting | Share of users in the target location | How it behaves |
|---|---|---|
| Narrow | 2.5 percent | Hard cap on audience size in Video campaigns |
| Balanced | 5 percent | Hard cap, and the default setting |
| Broad | 10 percent | Hard cap, the widest of the three |
| Demand Gen since March 2026 | No stated percentage ceiling | Directional audience suggestion fed to the algorithm rather than a targeting constraint |
What 2.5, 5 and 10 percent mean against a target location
The percentages are stated against users in the target location, which means the same setting produces very different absolute numbers depending on where the campaign is pointed. Narrow in a single city is a different quantity of people from Narrow across a country, and the tier does not change to compensate. Two consequences follow. The first is that the setting and the geography have to be planned together, because widening a location target while holding the tier constant raises the ceiling in absolute terms. The second is that comparing a Narrow campaign in one market against a Narrow campaign in another is comparing two different audience sizes that happen to share a label. For a brand campaign where reach is the point rather than a side effect, that arithmetic is the planning work.
The seed list numbers do not agree with each other
Google states the requirement as more than 100 active matched people, while Google's own documentation elsewhere recommends at least 1,000 active users in the seed list for better audience quality. Those are not contradictory statements, since one is a minimum to be allowed to build the segment and the other is a recommendation for the segment to be worth building, but a ten times gap between the two is worth naming out loud. Treated as a threshold to clear, 100 matched people will produce a segment. Whether the resulting audience resembles anything an advertiser wants is a separate question that the stated minimum does not answer. The practical reading is that the 100 figure describes eligibility and the 1,000 figure describes quality, and a media plan built on a list that barely clears the first number is inheriting a risk that is documented rather than hidden.
How a media planner should approach a test
The structure of the feature suggests where the useful comparisons sit. A short list follows, all of it based only on what Google stated.
- Start with the seed list rather than the setting, since the segment is built from it and Google's own quality recommendation is ten times the stated minimum
- Test tiers against each other rather than against an unrelated audience, because Narrow, Balanced and Broad differ in exactly one stated variable, the percentage ceiling
- Hold the location target constant while comparing tiers, since the percentage is calculated against users in that location
- Note that Balanced at 5 percent is the default, so a campaign nobody configured is running the middle tier rather than no tier
- Keep the Demand Gen difference in mind when reading old internal notes, because a Lookalike test run there after March 2026 was measuring a directional signal, not a cap
For teams that stopped using Lookalike audiences when the Demand Gen behaviour changed, the relevant point is narrow and specific: the percentage ceilings still exist in this campaign type, so a plan that depended on them has somewhere to live again. Anyone already buying YouTube ads for brand reach has the seed lists and the campaigns needed to check it without new infrastructure.
What the source did not establish
Three gaps are worth stating plainly. No performance data was published, so there is nothing showing that a Lookalike segment in a Video campaign outperforms any other targeting choice, and the reach control being available says nothing about whether it works better. No market availability list was given, so which countries can use this is not established. And Google has not said whether the Video treatment will later change the way Demand Gen's did, which is the question anyone planning around the hard caps would most want answered, given that Demand Gen had this same structure until March 2026. An advertiser can plan with the caps as they are stated today, but the precedent for them being replaced with a directional signal exists inside this same story.
What this means for Thai marketers
For brand campaigns bought on reach in a single market, the tier arithmetic is the immediate practical point, because 2.5, 5 and 10 percent of users in Thailand are three specific and very different audience sizes, and the default setting is the middle one. Advertisers running a Bangkok-only or region-only location target will get correspondingly smaller ceilings from the same three labels, so the setting cannot be copied between campaigns without recalculating what it means. The seed list gap matters here too. First-party lists in Thai accounts are often smaller than the equivalent lists in larger markets, which puts more of them in the range between the stated 100 person minimum and the 1,000 active user quality recommendation. Market availability was not stated in the source, so whether this is live for a given account is a question to answer in the account, and the sensible sequence for anyone running Google Ads alongside video is to size the seed list first and pick a tier second.
FAQ about Lookalike segments in Video campaigns
Is this available in Thailand?
The source did not state this, because no market availability list was published with the Help Center article and there was no blog post or advertiser email. An advertiser can check whether the Lookalike segment reach tiers appear in their own Video campaign setup.
What is the difference between this and Demand Gen Lookalikes?
In Video campaigns the percentages function as hard caps on audience size, while Google converted Demand Gen Lookalikes into directional audience suggestions fed to the algorithm in March 2026. That is the whole difference in behaviour, and it is why a planner can bound reach in one and not the other.
How big does my seed list need to be?
Google states the requirement as more than 100 active matched people, while its documentation elsewhere recommends at least 1,000 active users in the seed list for better audience quality. The first number gets the segment built and the second describes what makes it worth building.
Which tier should I start with?
The source published no performance data, so there is no evidence-based answer to point at, only the stated structure: Narrow caps at 2.5 percent, Balanced at 5 percent and Broad at 10 percent of users in the target location, with Balanced as the default.
Will Google soften the Video caps like it did for Demand Gen?
Google has not said, and the source did not establish this either way. What is on the record is that Demand Gen had the same tiered structure until March 2026 and now does not.
Reach controls that can be planned against are worth using while they behave as documented, and the work sits mostly in seed list sizing and location arithmetic rather than in the campaign setup itself. If a second pair of eyes on a video reach plan would help, Relevant Audience is happy to look at it.
