Guides · 2026-09-05 · By VSNARY | Emmanuel Orta · 0 views

YouTube SEO case study: a dormant channel with 81 videos rewritten, measured before and after

Terrarium Station’s channel had not uploaded in 19 months and was still growing 42% a year in search. Every title and description was rewritten in two dated batches. The baseline is published now so the after cannot be moved; the result is written on 3 October, flat included.

Terrarium Station is a reptile and invertebrate shop we operate. Its YouTube channel, @terrariumstation, has 1,259 subscribers and 86 uploads. Its last upload was on 19 January 2025. In the nineteen months since, nobody touched it, and its search views still grew 42% year on year. That combination, a compounding asset nobody was maintaining, is why it was chosen as the case study for the YouTube check: the channel is real, the traffic is real, and every number on this page was read from YouTube Studio or from the check itself on the date given beside it.

Where the views come from

The traffic-source export (28 May 2022 to 14 August 2026) puts lifetime views at 322,528, from 1,122,725 impressions at a 5.35% click-through rate. Half of those views came from the Shorts feed, and that half is history: the Shorts feed sent 128,495 views in 2022 and 3,258 in the first seven months of 2026. YouTube search went the other way, from 10,649 in 2022 to 41,474 in the same seven months of 2026, at a 5.56% click-through rate on 775,896 impressions. Search is now about 85% of current views. This is a search channel that still looked like a Shorts channel.

What the channel tally found

The channel tally was run on 2 September 2026, between the two batches of edits. It read 86 uploads through the Data API and 85 of them through a browser, which is the only vantage that can tell a manual caption from an auto-generated one or read a transcript. The first batch had rewritten titles and descriptions; it had not touched captions or chapters, and that is where the tally is weakest.

The defect that had nothing to do with ranking

On 14 August, before the tally existed, five descriptions were read by hand from the public watch pages. Three of the five linked to a marketplace storefront that returns HTTP 403, a page the shop left years ago. None of the five linked to the shop’s own site. Each carried the dead auto-appended “Please watch” block YouTube used to insert. And the strongest query on the channel, millipede terrarium setup at 3,397 search views, sat as a tag on the video that ranks for it while that video’s title read “Building A Bioactive Enclosure For Breeding Rare Flame Leg Millipedes”. The tags knew what people typed; the titles did not. Roughly 7,500 views a month were arriving with no path to anything the shop sells.

Search demand on the channel is also lopsided against its own catalogue. Of the top 100 queries, isopods account for 35.2% of search views, millipedes 21.6%, springtails 17.4%, crested geckos 9.7% and mantises 9.6%. The single largest cluster, rubber-ducky isopods, had no product page when the audit was run. It has one now.

What was changed, and when

1. 2026-08-14Audit. Traffic-source export read; descriptions sampled; the outbound layer found pointing at a 403 page and at nothing else.measured
2. 2026-09-01Batch 1. 15 long-form videos and 20 Shorts rewritten: query-first titles, dead links removed, every description given a shop link that returned 200 when checked. Channel description and channel links rebuilt, shop first.edited
3. 2026-09-02Tally. The channel check run for the first time (the figure above). The site half found three VideoObject nodes on the shop without uploadDate; fixed the same day with verified dates.measured
4. 2026-09-05Batch 2. The remaining 46 Shorts rewritten, 45 of them in 178 seconds by replaying the Studio’s own metadata call rather than clicking through the editor. The baseline below was read from Studio Analytics in the same hour.edited
5. 2026-09-06Post-window opens. 28 days.window
6. 2026-10-03Post-window closes. The export is re-run and the result is written into this page.window

A rewrite, precisely: the title leads with the query the video already ranks for, with no channel-name suffix and no adjective in front of the animal. The description opens with one sentence that states the subject, then a link to the shop category that returned 200 when checked, a care-guide link where a real guide exists, the site root, the breeding-records registry and the community group. No claim the footage cannot support is made anywhere in it.

Correction, 6 September 2026

This page first gave the date of the first batch as 29 August and treated 8 August to 4 September as a clean pre-edit baseline. Both were wrong: the first batch landed on 1 September, which puts four of that window’s days after the edit rather than before it. The dates above and the measurement design below are corrected, and the original window is left visible rather than replaced, because a baseline that moves after the fact is not a baseline. The result on 3 October will be read against 4 to 31 August.

The baseline, as it was read on 5 September

Studio Analytics on 5 September, 28-day window 8 August to 4 September: 9,451 views (2.9K above what Studio calls typical), 26.6 hours watched, 13 subscribers gained (a 333% change against the previous 28 days), 1,259 in total. The realtime 48-hour figure was 859 views; the three videos carrying it were the rubber-ducky isopod enclosure (215), the praying-mantis care guide (185) and the orange-springtail master colony (77). Lifetime views per video were recorded for all 66 Shorts the same hour; the top four hold 36,132, 16,482, 10,931 and 10,384.

That window is not a clean baseline, and this page originally said it was. It runs 8 August to 4 September, and the first batch of edits landed on 1 September — so 24 of its 28 days precede the first edit and four follow it. It is published here as it was read, dated, rather than quietly swapped, but the comparison below uses the clean pre-edit window instead. The channel was also already growing 42% a year in search before anyone touched it, so reading the “above typical” line as an effect would be measuring a trend the edits did not cause.

The channel, measured on a schedule

The result below is promised for a date, and a number promised for a date is one somebody has to remember to fetch. So the channel is now sampled once a day and this section reads from that series rather than from the day the page was written.

The subscriber figure is rounded by YouTube, not by us. Above 1,000 the public count is published to three significant figures, so it moves in steps of ten and will read lower than the exact number in the owner’s Studio. The view counters are exact. A subscriber change smaller than ten is invisible in this series by construction.

Read from the channel’s public counters on 2026-09-11. The series opens on 2026-09-06 and holds 6 daily points across 5 days. It updates itself; nothing on this page is retyped.

What this series does not contain: impressions, click-through rate, watch time and traffic source. Those live in YouTube Analytics behind the owner’s login, and no public surface exposes them — which is why the 3 October read still comes from a Studio export. The series is public as JSON.

How the result will be read on 3 October

• A fresh traffic-source export, per video: search impressions, click-through rate and views.
• 28 days pre (4 August to 31 August, the last full day before any edit) against 28 days post (6 September to 3 October). 1 to 5 September contains both batches and belongs to neither.
• Two cohorts reported separately: the 35 videos edited on 1 September and the 46 Shorts edited on 5 September.
• The control is the channel’s own pre-edit trend. Every video was edited, so there is no untouched cohort, and the case says so rather than inventing one.
• The channel tally is re-run the same day; rows that moved are listed beside the ones that did not.

What this case can and cannot show

It can show whether per-video search impressions, click-through and views moved in the 28 days after each batch, against the channel’s own trend. It cannot show why. Click and watch-time data live inside Google, and the internal ranking scores are exposed on no public surface; the check lists them as not measurable on every result. One channel is one channel, not a rate. And an autumn window against a summer window carries seasonality the design cannot remove, so where the export allows it, the same weeks of the previous year will be printed beside the result.

Run it on your own channel

The single-video check is free and reads chapters, captions, description and language. The channel tally reads every upload and is included from Watch upward, one channel on Watch, three on Studio and ten on Agency. The site half answers whether your own pages carry the videos at all, and whether the VideoObject schema describing them is complete. None of it changes what YouTube does; it tells you what YouTube can read.

Questions this post answers

Why publish a case study with no result yet?

Because a baseline published after the result is not a baseline. The clean 28-day pre-window (4 to 31 August) and the exact edit dates are on record before the post-window has produced a single day. The post-window closes on 3 October and the number is written then, whatever it is.

Was the 9,451-view month caused by the edits?

No. It straddles the first edit — 24 of its 28 days precede 1 September and four follow it — so it is not a clean baseline, which the page now states. The clean pre-window is 4 to 31 August. The channel was also growing 42% year on year before any edit.

What is the control?

The channel’s own pre-edit trend. Every video was edited, so there is no untouched cohort; the two batches (1 September and 5 September) are reported separately instead.

Which signals cannot be measured?

Click and watch-time data, and the internal ranking scores themselves. The check reads what the public surfaces expose and lists the rest as not measurable rather than guessing.

Can I run the same check on my channel?

Yes. The single-video check is free. The channel tally reads every upload through the Data API and is included from Watch upward; captions and transcripts count only for videos a browser has observed.

Related findings

Writing about this? Facts, live figures and marks — every number on that page is dated and traceable to a scan.

All guides · The dataset · How the dataset works