YouTube today launched an experimental feature that allows signed-in viewers to build a personalised content feed on the platform's Home tab by entering a plain-text prompt - a structural departure from purely algorithmic content delivery that carries direct implications for how audiences are built, measured, and monetised on the platform.

What the custom feed does

The feature, announced today by Dave from TeamYouTube in the platform's official community forum, introduces a new chip labelled "Your custom feed" that appears alongside the standard Home option at the top of the YouTube Home page. Viewers tap or click the chip, type a description of the content they want, and YouTube generates a dedicated feed of matching videos that refreshes continuously.

According to YouTube's Help Center documentation published alongside the announcement, the feature is designed to give viewers "a dedicated space for a specific topic that remains anchored to your Home screen." The practical effect is that a viewer can describe a specific interest or routine - the documentation offers the example "Help me unwind after work with guided meditations under 10 minutes" - and receive a curated, persistent feed without having to re-enter search queries each time.

According to the announcement, viewers can also choose from a set of suggested prompts rather than writing their own. The feed can be updated at any time by editing the text in the prompt box at the top of the custom feed. Only one custom feed can be active per account at any given time.

There is one automatic expiry mechanism built into the system. According to the Help Center, "Custom feed prompts expire after 30 days of inactivity. Once a prompt expires, that custom feed will no longer be available, and you can enter a new prompt to create a new custom feed." This 30-day clock resets with each use of the feed, meaning active users will not lose their configuration unless they stop visiting the custom feed for a full month.

Technical requirements and availability

The rollout carries specific prerequisites. According to YouTube, viewers must be signed in to a YouTube account, must have both their YouTube watch history and search history turned on in account settings, and must have their YouTube device language set to English. The feature is currently available only in the United States, on mobile and desktop, and YouTube's own documentation notes that "availability may change in the future."

The watch history and search history requirement is technically significant. Those two data sources feed the recommendation engine that generates the custom feed's results. Without them, the system lacks the behavioural signals needed to contextualise and filter content against the viewer's prompt. Viewers who have paused or deleted their history for privacy reasons will not be able to access the feature in its current form.

The announcement labels this explicitly as an experimental feature. According to the Help Center, "This experimental feature is currently available in English in the United States." YouTube's track record with experimental rollouts suggests the geographic and language restrictions are standard for early-stage testing before broader deployment.

How the prompt input works

The prompt system accepts free-form natural language rather than a structured set of categories or toggles. According to YouTube's community post, examples include open-ended requests like "give me something different beyond my usual feed" as well as narrowly scoped requests such as "help me unwind after work with guided meditations under 10 minutes." The breadth of acceptable input is deliberately wide.

Prompts are subject to YouTube's Community Guidelines. According to the Help Center, "Certain prompts may not be allowed if they violate these policies." This means the content moderation layer applies at the input stage, not just at the level of individual videos surfaced by the feed.

If the feed does not accurately reflect what the viewer intended, a feedback mechanism is available. According to the documentation, viewers can tap the three-dot menu next to the prompt box and select "Something wrong?" to share feedback directly with YouTube. The platform has indicated that its teams use this information to make improvements.

Monetisation and watch time implications

The monetisation mechanics of the custom feed follow the same rules as the standard Home feed. According to YouTube's Help Center FAQ, "This feature behaves like the Home feed. If a viewer discovers and watches a video through their custom feed, it contributes to the creator's watch time and monetisation as usual."

That parity matters for creators. Watch time accumulated through the custom feed counts toward monetisation thresholds and toward the recommendation signals that determine how widely a video is distributed. A creator whose content appears prominently in custom feeds built around their niche benefits from the same revenue share as they would from a standard algorithmic recommendation.

YouTube has acknowledged uncertainty about the aggregate effect on creator performance. According to the Help Center, "We are continuing to closely monitor the feature's impact on views, watch time, and revenue." The platform has not provided projected figures for how the feature might alter traffic distribution between channels, and the documentation frames ongoing monitoring as an active commitment rather than a completed analysis.

Context: a long period of algorithmic friction

The launch of the custom feed arrives after a sustained period during which YouTube's recommendation systems have been a source of conflict between the platform and its creator community.

YouTube's Home feed was documented reducing long-form video discovery in late December 2025, with one researcher reporting that browse recommendations dropped from six long-form videos per row to just two, redirecting the majority of available slots toward Shorts content. The structural shift intensified debate about algorithmic transparency and its impact on creators who depend on discovery through the Home tab.

In January 2026, YouTube removed the sort-by-upload-date function from search, replacing chronological sorting with a "Popularity" filter that incorporates watch time alongside view counts. Creators producing time-sensitive content reported structural disadvantages as a result. The removal eliminated a viewer-controlled pathway to recent content that had existed for years.

In March 2026, TeamYouTube published a FAQ consolidating guidance on the tools viewers already had available to influence their recommendations - including watch history deletion, the "Not Interested" flag, and channel suppression. That document arrived during what the platform itself acknowledged was "heightened scrutiny over how the platform's recommendation systems operate."

The custom feed represents a different approach to the same underlying problem. Rather than providing controls that modify the existing algorithm, it creates a parallel feed running on a viewer-supplied brief. The two systems coexist - returning to the standard Home tab restores the default algorithmic feed instantly. According to YouTube's documentation, "You can return to your standard YouTube experience at any time by selecting the Home option from the top of your Home feed."

Privacy and data architecture

The custom feed is account-bound and private. According to YouTube's Help Center, "Your custom feed is private and unique to your account. Other viewers cannot find or access your custom feed." This design choice means the prompt input does not become a public signal or shared taxonomy. It functions as a personal configuration layer sitting on top of the recommendation infrastructure.

The videos surfaced by a custom feed are added to the viewer's standard watch history in the same way as any other Home feed watch. According to the documentation, "Videos you watch from this feed are added to your standard watch history. These watches may influence your main Home recommendations." The two systems are therefore not fully isolated - activity in the custom feed feeds back into the standard algorithmic profile over time.

This creates a feedback loop worth noting. A viewer who builds a custom feed around a specific topic and watches extensively from it will see their main Home recommendations shift toward that topic. The custom feed thus functions both as a direct content delivery mechanism and as a steering input for the broader recommendation engine.

What refresh behaviour looks like

When a viewer refreshes the custom feed, the system generates a new set of recommendations based on the current prompt. According to YouTube's documentation, "Some videos may stay the same, while others will change. This process is similar to refreshing your standard Home feed." This behaviour distinguishes the custom feed from a static playlist or saved queue. The content is dynamic. The curation logic remains constant - defined by the prompt - but the specific videos that populate the feed change as new content is published and as the recommendation engine updates its outputs.

The prompt itself persists across refreshes. Changing the content of the feed requires editing the prompt text, not simply refreshing. This distinction is architecturally important: the feed is prompt-persistent rather than session-based, which allows it to function as a durable, named content space rather than a one-time query result.

Implications for media buyers and platform advertisers

For the advertising community, the custom feed introduces a new surface with direct implications for how inventory is served and how audiences are classified. The feature behaves like the Home feed from a monetisation perspective, which means ads delivered against custom feed views are handled through the standard Home feed ad infrastructure. The targeting signals available to advertisers - based on viewer identity and history - remain in place.

The more significant question is whether the custom feed will alter the composition of Home feed inventory. If viewers actively using the feature spend a material share of their Home tab session time inside the custom feed rather than the default feed, the aggregate audience exposed to Home feed ad slots changes in character. A viewer who has anchored their Home experience around a specific category prompt is, by definition, expressing a content preference that the custom feed then validates. That preference signal is not new - it was implicit in watch history before - but the custom feed makes it explicit and durable in a way that standard algorithmic browsing does not.

YouTube's viewer analytics introduced in July 2025 already segment audiences into new, casual, and regular viewer categories, providing creators with a framework for understanding engagement depth. The custom feed adds another dimension: viewers who have deliberately self-selected into a topic-specific content environment. Whether YouTube will eventually expose that signal to advertisers or creators in any structured form is not addressed in the current documentation.

The feature also has implications for how YouTube's broader discovery infrastructure is understood by the marketing community. Recommendation systems on the platform have historically been opaque, with limited viewer control beyond indirect feedback signals. The custom feed introduces a direct, explicit input layer that, if scaled beyond the current U.S. English experiment, could reshape how content surfaces to audiences in ways that algorithmic adjustments alone cannot predict or replicate.

Early user response

Response in the YouTube community thread has been mixed. Several commenters expressed straightforward appreciation for the update. One early reply read: "Thank you Team YouTube for this new update. This custom feed feature looks very helpful and interesting for discovering new content easily. Hoping it will also help small creators grow faster."

Not all initial feedback was positive. One commenter raised a functional concern directly relevant to the feature's core promise: "Except that the suggestions do not connect with your history and give you a bunch of irrelevant content to what you are looking for. There is no integration." This observation - that the prompt-driven feed may not yet accurately reflect viewer preferences - points to a gap between the feature's stated purpose and its current performance. YouTube's own documentation acknowledges this possibility, noting the feedback mechanism available for cases where "the custom feed does not accurately interpret your request."

Timeline

  • July 2025 - YouTube introduces refined viewer analytics distinguishing casual and regular viewers, adding a new dimension to engagement measurement. (PPC Land)
  • July 11, 2025 - YouTube closes its Trending page after a decade, accelerating its shift toward personalised algorithmic discovery. (PPC Land)
  • August 7, 2025 - YouTube launches a collaboration feature allowing multiple creators to share credit and audience distribution. (PPC Land)
  • September 14-15, 2025 - Creators report significant view drops following undisclosed algorithm changes, with YouTube clarifying that systems were functioning as intended. (PPC Land)
  • October 23, 2025 - YouTube extends Communities to desktop, adding AI-powered Live Q&A stickers and a creator reinstatement pilot. (PPC Land)
  • December 28, 2025 - Research documents YouTube's Home feed reducing long-form video recommendations by up to 80% in favour of Shorts. (PPC Land)
  • January 11, 2026 - YouTube removes sort-by-upload-date from search, replacing chronological sorting with a popularity-weighted filter, drawing creator backlash. (PPC Land)
  • March 3, 2026 - TeamYouTube publishes a viewer controls FAQ consolidating tools for managing recommendations and search results. (PPC Land)
  • May 27, 2026 - YouTube today launches the custom feed feature on Home, allowing U.S. English users to generate personalised content feeds via free-form text prompts.

Summary

Who: YouTube (Google), announcing via Dave from TeamYouTube in the platform's official community forum, targeting signed-in viewers in the United States with YouTube language set to English.

What: A new experimental feature called "Your custom feed" that allows viewers to enter a plain-text prompt on the YouTube Home tab to generate a personalised, continuously refreshing content feed. Only one custom feed can be active per account at a time. Prompts expire after 30 days of inactivity. Videos watched through the feed count toward creator watch time and monetisation at the same rate as the standard Home feed.

When: Announced today, May 27, 2026. The feature is currently in experimental status with no confirmed timeline for broader availability.

Where: Available exclusively in the United States, on YouTube mobile and desktop, in English. Requires the viewer to be signed in with watch history and search history both enabled.

Why: The feature addresses a long-standing friction between YouTube's algorithmic recommendation system and viewer intent - offering a direct, prompt-driven input layer alongside the existing behavioural inference engine. It arrives after months of documented tension between the platform and creators over algorithmic changes that reduced Home feed discovery for long-form content and removed chronological search sorting. For the advertising community, the feature introduces a new behavioural signal - explicit viewer topic preference - within the Home feed inventory environment.

Share this article
The link has been copied!