October 14, 2026
Your CTR fell from 5.2% to 3.1% in ten days. You changed nothing. You reuploaded the thumbnail, tweaked the title, waited. Nothing. The mistake almost every faceless documentary channel makes is treating CTR as one number. It is not. A single video has four or five different CTRs running in parallel — one for mobile, one for desktop, one for TV, one for Home, one for Suggested — and they move for completely different reasons. A YouTube CTR drop by device diagnosis takes 20 minutes and tells you exactly which surface broke and what to fix.
This guide gives you the reference thresholds per placement, a comparison table you can copy into a spreadsheet, and a decision tree that turns the data into one specific action. If mobile holds but TV collapses, your thumbnail text is too small. If Home drops but Search rises, the video is fine — the packaging is the problem. Those are two different jobs, and mixing them up wastes weeks.
Everything below is built for 25–35 minute faceless documentaries, where the audience splits roughly 55–65% mobile, 15–25% TV, 10–20% desktop, with the rest on tablets and consoles. Numbers shift by niche, but the diagnostic logic does not.
YouTube Studio shows an aggregate CTR for the video. That aggregate is a weighted average of impressions across every surface: Home feed, Suggested, Search, Subscriptions, Shorts feed (if applicable), and the watch page. Each surface has a different baseline, a different audience intent, and a different thumbnail rendering size.
Mobile Home shows a thumbnail roughly 320–360px wide on a typical phone. Desktop Home shows 210–246px. TV shows a thumbnail that occupies a huge portion of a 55-inch screen, but the viewer is sitting 2.5–3.5 meters away. Search shows small thumbnails with the title directly beneath. Suggested shows medium thumbnails next to a video the viewer already trusts.
That means the same 1280x720 image performs differently in five places. A thumbnail with 40px text is readable on a TV and almost invisible on a phone. A thumbnail with a busy collage works on desktop where the eye lingers, and fails on mobile where the scroll is fast. Your CTR dropped because one of those contexts changed — either the surface mix shifted (YouTube pushed you more into Home, less into Suggested), or your audience mix shifted (more TV viewers after a viral push), or the packaging genuinely got worse for one context.
You cannot fix a weighted average. You can fix a surface.
Use these as sanity checks, not gospel. Documentary and long-form educational channels typically land in these bands. If you are more than 30% below the band, that placement is the problem.
| Placement | Device / context | Healthy CTR band | Warning zone | Common cause of drop |
|---|---|---|---|---|
| Home feed | Mobile | 4.5–7.5% | <3.5% | Thumbnail not readable at 320px; weak hook in first 3 words of title |
| Home feed | Desktop | 3.0–5.5% | <2.4% | Title too long, truncated; thumbnail lacks focal point |
| Home feed | TV | 5.5–9.0% | <4.0% | Text under 60px on the 1280x720 canvas; low contrast |
| Suggested | Mobile | 3.5–6.0% | <2.8% | Topic mismatch with the source video; thumbnail style clash |
| Suggested | Desktop | 2.5–4.5% | <2.0% | Weak topical adjacency; title does not signal payoff |
| Search | All devices | 6.0–11.0% | <4.5% | Title does not match query intent; thumbnail generic |
| Subscriptions feed | Mobile | 8.0–15.0% | <6.0% | Audience fatigue; posting cadence too high |
| Shorts feed (if used) | Mobile | 60–90% (swipe-adjusted) | <45% | First frame weak; no motion hook in first 0.8s |
Two things to internalize. First, Search CTR is always higher than Home CTR — the viewer typed a query, intent is high. Second, TV CTR is often higher than mobile CTR for documentary content because the thumbnail is physically enormous and the viewer is not scrolling. If your TV CTR is lower than your mobile CTR, something is structurally wrong with the thumbnail at distance.
Pull these numbers from YouTube Studio: Analytics → Content → select video → Advanced mode → Traffic source, then switch the dimension to Device type. You will not get CTR broken down by device AND traffic source simultaneously in the default UI. Workaround: export the CSV from the Traffic source card and the Device type card, then cross-reference by date. It takes 10 minutes and it is the only way to see mobile-Home vs TV-Home separately.
Run this in order. Stop at the first branch that matches.
Print that list. It is the whole article in six lines.
Most creators test thumbnails at full size on a 27-inch monitor. That is the least useful view. Here is the 4-step test that maps to real feeds.
Step 4 is the one almost nobody does, and it is the one that correlates most with real CTR movement. A thumbnail that wins in isolation often loses in a feed of 6 competing images with similar color palettes.
On a 1280x720 canvas, text smaller than 90px tall is a coin flip on TV. Below 60px, it is invisible at 3 meters. For mobile, the practical minimum is 110px because the thumbnail is rendered at roughly 25% of its native size. The sweet spot for a 3-word title overlay is 130–170px with a heavy sans-serif font (Impact, Bebas Neue, or a bold grotesque) and a 4–6px dark stroke or drop shadow.
Contrast matters more than color choice. A white text on a mid-grey background fails everywhere. White on near-black, or yellow on deep blue, wins on all four surfaces. Test with a grayscale conversion — if the text disappears in grayscale, it will disappear on a dim phone screen in daylight.
Home and Suggested are not the same audience. Home is browse — the viewer is idle, scrolling, low intent, high volume. Suggested is contextual — the viewer just watched something and is deciding whether to continue. That difference changes what wins.
On Home, the thumbnail carries 70–80% of the click decision. The title is secondary because the viewer scans images first. On Suggested, the title matters more because the viewer is already in a watching mindset and reads the title to judge relevance to what they just watched.
Practical implications:
If your thumbnail CTR feed placement pattern shows Home strong and Suggested weak, you have a packaging that hooks browsers but not continuers. The fix is usually a title that explicitly connects to the source video's topic — "The Other Side of [Topic From Source Video]" outperforms a generic curiosity title in Suggested.
Honest section. Most CTR recovery advice circulating in 2026 is either outdated or actively harmful.
Realistic recovery timeline: if you correctly diagnose the surface and fix the packaging, expect CTR to recover 60–80% of the drop within 10–14 days. Full recovery is not guaranteed because the surface mix may have permanently shifted. Budget 2 hours of work per video for a proper diagnosis and rebuild.
Use this every time a video underperforms. Fill it in before you change anything.
Nine fields. Ten minutes. It replaces every guess you would otherwise make.
Almost always a surface mix shift. YouTube pushed your video into a new feed (often Home or TV) where your packaging was never optimized. Check the Traffic source and Device type reports for the last 14 days — you will see the mix change before the CTR change. The fix is to optimize for the new dominant surface, not to revert a thumbnail that was working elsewhere.
Aggregate CTR between 4% and 7% is healthy for 25–35 minute documentaries. Below 3% across all placements indicates a packaging problem. Above 8% aggregate usually means you are being shown mostly in Search or Subscriptions, which is a different (and smaller) game. Always judge by placement, not by the aggregate.
Open the video in Analytics, go to Advanced mode, and add the Device type dimension. You will see impressions, views, and CTR split by mobile, desktop, TV, tablet, and game console. To combine device with traffic source, export both CSVs and cross-reference by date — the UI does not show the intersection directly.
Yes, but only the text layer. Keep the composition and subject that work on mobile, and enlarge or simplify the text overlay so it reads at 3 meters. This preserves mobile performance while recovering TV. Do not rebuild the whole image — you risk breaking the surface that was working.
Seven days minimum, and ideally 10–14 for long-form documentary. YouTube needs enough impressions per placement to produce a stable CTR. Judging at 48 hours is the single most common mistake and leads to endless thumbnail churn with no learning.
One. Never diagnose CTR from the aggregate. Split it by device (mobile, desktop, TV) and by placement (Home, Suggested, Search). The drop lives in one cell of that grid, and that cell tells you the fix.
Two. Use the decision tree. Mobile strong + TV weak means text legibility. Home weak + Search strong means packaging. Suggested weak across devices means topical mismatch. Each has a different, specific fix.
Three. Give every change 7–14 days before judging. Most CTR recoveries happen in that window, and most self-inflicted CTR disasters come from changing too fast. If you want a faster path, an AI-powered documentary production system like the one behind VAATIK can handle packaging variants and surface testing at scale — but the diagnostic logic above still applies to every video you publish.
See How VAATIK Can Run Your Channel → Partner Program → €200 + 10% Recurring