Verified against ChatGPT · 2026-07-28
Plan which videos an end screen should actually point to, and why
Chooses end-screen elements based on session-time strategy — what keeps this specific viewer watching this channel next — instead of defaulting to "most recent upload" or a generic subscribe element in every video.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are planning the end-screen elements for a YouTube video — which specific videos, playlists, and subscribe prompt to place in the final 5-20 seconds, and why those specific choices over the available alternatives. This is a strategic placement decision, not a script; the goal is maximizing the chance a viewer who just finished this video keeps watching something else from this channel, which is a session-time decision, not a "what's newest" default. THIS VIDEO'S TOPIC AND LIKELY VIEWER INTENT A video comparing two mirrorless cameras for wildlife photography specifically — viewers likely arrived comparing autofocus tracking specifically CANDIDATE VIDEOS OR PLAYLISTS TO CONSIDER A comparison of two different camera brands for the same use case, a general "best cameras 2026" roundup, and a tutorial on wildlife photography settings unrelated to gear comparison CHANNEL'S TOP-PERFORMING CONTENT The other brand comparison video has 74% average retention; the "best cameras 2026" roundup has 41% retention; the settings tutorial has 58% END SCREEN LENGTH AVAILABLE 18 seconds STRATEGY RULES Prioritize a video that shares this video's actual viewer intent — the specific question or interest that brought someone to click this video in the first place — over a video that's merely topically adjacent or merely the channel's most recent upload; a viewer who watched a video comparing two products is a stronger candidate to watch another comparison than to watch an unrelated tutorial, even if the tutorial is newer or generally higher-performing. Where a candidate video is one of the channel's strongest performers by retention or watch time, weight it more heavily than a similarly relevant but weaker-performing option — an end screen is a recommendation, and recommending a video this channel's own data shows holds attention well compounds the chance this viewer keeps watching versus recommending one the data shows tends to lose people early. Match the number of end-screen elements to the actual available duration and to YouTube's own display limits — up to four elements can display, but a video with under roughly 25 seconds of usable end-screen time may only comfortably support one or two before elements overlap or get cut short; do not propose more elements than will actually render cleanly in the time available. If this video and the top candidate are part of a natural sequence — a follow-up, a "part 2," a direct comparison's counterpart — say so explicitly and prioritize that sequential relationship over a general popularity-based recommendation, since a viewer who was clearly left with an open question by this video is unusually likely to click through to the video that resolves it. Include the subscribe element only where it doesn't crowd out a genuinely higher-conversion video suggestion — a video with strong per-viewer conversion evidence toward watching a specific next video may reasonably use all its element slots on video suggestions rather than splitting attention with a subscribe prompt that has lower marginal value in that specific slot. OUTPUT FORMAT 1. The recommended end-screen elements in priority order, each with the specific video or playlist and a one-line reason tied to viewer intent or performance data. 2. Confirmation the total element count fits the stated available duration. 3. If a sequential/follow-up relationship exists, one line naming it and why it outranks a purely popularity-based pick.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Prioritizing shared viewer intent over topical adjacency or upload recency reflects what YouTube's own recommendation system is actually optimizing for at the channel level — session watch time, meaning how long a viewer stays engaged with content across multiple videos in one sitting, not just this single video's own metrics — and a viewer whose specific reason for clicking was a head-to-head comparison is a measurably better match for another comparison than for a generally newer or generally popular video that doesn't share that same intent, because intent match is what actually predicts whether the next click happens, not surface-level topic similarity. Weighting a candidate's own retention or watch-time performance into the decision, rather than treating all topically-relevant options as interchangeable, treats an end screen as a real recommendation with real data behind it instead of a slot to fill — a channel's own retention numbers are a direct, specific, first-party signal about which videos actually hold viewers, and ignoring that data in favor of a merely-relevant pick is leaving a channel's best available evidence on the table exactly where it would matter most. The element-count-versus-available-duration rule is a mechanical constraint many end-screen plans ignore: YouTube supports up to four end-screen elements displaying, but they need real screen time to render and be clickable, and a plan that proposes four elements onto a video with 15 seconds of usable end-screen space produces an outcome where elements overlap, get cut off before a viewer can read them, or simply blur together — a technically valid element count on paper that fails in the actual player. Recognizing an explicit sequential relationship — a natural part 2, a direct follow-up resolving a question this video raised — as a distinct, higher-priority category than general popularity matters because a viewer left with an unresolved question by design (per the earlier open-loop principle in scripting) is in an unusually high-intent state specifically for whatever resolves that exact question, a narrower and stronger signal than "viewers of this video also tend to enjoy that video" statistics can capture on their own.
Verified against
ChatGPT GPT-5.1 · 2026-07-28
Claude Sonnet 4.6 · 2026-08-04
Changelog
- 2026-07-28 — Initial publish, verified against ChatGPT (GPT-5.1) and Claude (Sonnet 4.6).
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
