ssh.autos

Recipes

Recipe · Apify

Find the videos that beat their channel, then read them

Pick a niche, find the recent videos that pulled far more views than their channel usually gets, and pull their transcripts to see how they open and what they promise.

COSTAbout $0.03 per run, billed by Apify
RUNS2 runs, 2 sources
TIMEUnder a minute
YOU NEEDAn Apify account and the Apify CLI, or the Apify Console.

What you get

We searched for “home espresso” videos from the last 30 days. These seven beat their own channel's typical views by the widest margin.

ScoreVideo and channelViewsChannel typical
9.84×Best Coffee Machine? What 4 Years of Testing Taught Me
Cafelista_Coffee · 2026-11-02		58,6235,959
8.7×The Ultimate Home Espresso Machines Under £2,000 ☕ 4 Machines Compared!
Bella Barista · 2026-10-25		25,9842,986
5.16×The Best Espresso Machines 2026 Review: Results After Months of Testing
Men's Journal · 2026-10-29		19,7473,825
4.94×$4,000 vs $700 vs $1,800 Espresso Machine: Which Should You Buy?
More George · 2026-10-15		74,06415,000
4.43×i finally got my dream espresso machine
tatsuki · 2026-10-31		241,63654,500
3.16×Best of 2026: Entry Level Espresso Machines
Whole Latte Love · 2026-10-14		108,04834,166.5
3.03×ZURIGA T2 Review — The Best Espresso Machine I’ve Tested?
Luke Makes Coffee · 2026-10-24		51,46216,958

Then the transcripts of the top two, read by an agent:

how the top two open · 2026-11-10
9.84×
Cafelista_Coffee opens on the buyer's problem, “more options than ever”, promises a pick in every category, and teases that “the most expensive option isn't always the winner”. She says the footage is her own testing.
8.7×
Bella Barista ties the video to the run-up to Christmas, sets a budget of under £2,000, and names all four machines with their prices in the opening two minutes: 1799, 1499, 1899 and 1599.
Pattern
Seven of the ten results are buying guides: “best”, “compared” or “which should you buy” is in the title.

Two real runs on 2026-11-10. Claude wrote the notes from the two transcripts, using the prompt below. Download the results: all ten rows, and the start of each transcript.

Run it

Two of our actors on the Apify Store, run on your own Apify account. First find the outliers:

terminal · step 1
apify call automatic/youtube-outlier-finder -i '{"searchQuery": "home espresso", "lookbackDays": 30, "maxResults": 10}' -s -o > outliers.json

Then pass the links of the videos you want to read to the transcripts actor:

terminal · step 2
apify call automatic/youtube-transcripts-scraper -i '{"sources": ["https://www.youtube.com/watch?v=vy-IH48TE6s", "https://www.youtube.com/watch?v=GGvrd35fXbI"], "maxVideos": 2}' -s -o > transcripts.json

We ran both commands on 2026-11-10: 31 seconds and 11 seconds. No CLI? Paste the same JSON into each actor's input on Apify.

What it costs

Both actors charge per result on Apify. At the list prices we read on 2026-11-10, ten ranked videos are $0.015 and two transcripts are $0.009, so this run is about $0.03. Apify bills it to your account; check each actor's page for today's price.

Hand the results to your agent

prompt.txt
Read outliers.json. Drop every row whose channelMedianViews is under 1,000, then list the rest by outlierScore with title, channel, viewCount and channelMedianViews.

Then read transcripts.json. For each video, describe in two sentences how it opens: the problem it names, what it promises, and any price or number it gives in the first 300 words. Quote at most ten words per video.

Finish with one line on what the titles in outliers.json have in common. Report what the videos do. Do not claim to know why YouTube showed them to more people.

Good to know

  • The score is a video's views divided by its channel's typical views. Tiny channels score high on almost nothing: in our run a video with 152 views scored 60.8 because its channel's typical video gets 2.5. That is why the prompt drops channels under 1,000.
  • You can give the outlier finder a list of channels instead of a search term, to watch competitors rather than a topic.
  • Both transcripts in our run were YouTube's automatic captions, so expect misheard brand names.
  • A transcript tells you what was said. It does not tell you why a video spread.
  • Long videos only by default. Shorts are a separate mode of the outlier finder.

Sources in this recipe

Machine-readable: openapi.json · llms.txt

Some links to Apify are affiliate links. We may earn a commission if you sign up and become a paying customer.