How an AI educator runs a month of research in one morning
Most people work through research a search at a time. This AI and machine learning educator sends a few hundred searches at once through the API, then spends the rest of the week writing. Monthly views on the channel went from an average of 282K to an average of 438K across the period.
AI and tech channels, analysed
Live analyses of other channels in this space, each showing subscribers, average views and the videos pulling above that channel's normal.
OutlierKit Channel Analysis
Cleo Abram
Optimistic tech explainers
- Subscribers
- 8.3M
- Avg views
- 5.0M
OutlierKit Channel Analysis
Matthew Berman
AI news & tutorials
- Subscribers
- 621.0K
- Avg views
- 81.4K
OutlierKit Channel Analysis
Matt Wolfe
AI news & tools
- Subscribers
- 977.0K
- Avg views
- 102.2K
OutlierKit Channel Analysis
Programming with Mosh
Software Engineering Education
- Subscribers
- 5.1M
- Avg views
- 1.1M
OutlierKit Channel Analysis
freeCodeCamp.org
Programming & software
- Subscribers
- 11.9M
- Avg views
- 432.7K
OutlierKit Channel Analysis
John Hammond
Cybersecurity
- Subscribers
- 2.1M
- Avg views
- 48.1K
Stats are from our most recent snapshot of each channel. Scroll sideways for more. For live numbers, outlier videos, and up-to-date revenue estimates, run a fresh analysis on OutlierKit →
The numbers
| Who | A real AI and machine learning educator, anonymised |
| Subscribers | Around 156K |
| Views before the API work | 282K a month, averaged over six months |
| Views after | 438K a month, with a best month of 491K |
| How they work | Everything through the API, in large batches |
| What they call | Outlier search and keyword research, roughly three to one |
Scroll the chart sideways to see the full timeline.
How they work
This account signed up, ran two searches in the browser, and then went quiet for six months. The return was not to the web app. It was a script.
The first real sitting was about 115 searches in a day. A later one was about 320. Between them there are small top-ups of five or ten. Everything is outlier search and keyword research, at roughly three searches to every one keyword lookup.
That shape suits how a lot of technical creators plan. You decide a slate in one sitting, pull everything you need for it at once, and then stop thinking about research until the slate runs out.
Two ways to do this yourself
- In plain language
- The OutlierKit connector puts the same data inside Claude. Ask what is overperforming in your niche and get a ranked answer with the videos attached. Pro and above, and open to everyone on the free trial.
- In a script
- The API gives you outlier search, channel stats and keyword volumes to call in bulk, which is what this educator does.
What they used
Two endpoints carry the whole workflow.
Three outlier searches for every keyword check. The search finds what is working, the keyword volume says how many people are asking for it, and nothing else is needed to fill a slate.
What moved, and what did not
Views moved. The six months before the API work averaged 282K a month. The three months after averaged 438K, and September 2026 was the best month for views in the whole window at 491K.
Subscribers did not. The subscriber growth rate was flat across the same period. We are not going to dress that up: this is a case about reach and output, not about a subscriber jump.
Depending on what a channel sells, views may well be the number that matters more. But they are different outcomes, and only one of them moved.
What you can copy
- Batch the research, not the writing. Pull everything for a whole slate in one sitting, then close the tab. Research is the part that benefits from being done in bulk.
- Pair every outlier search with a keyword check. This account runs about one keyword lookup for every three searches. The search tells you what is working, the keyword volume tells you how many people are asking for it.
- Use the API once the query stops changing. If you are running the same search every planning cycle, that is the signal to script it.
- Judge it on the number your business runs on. On this account that is views and reach rather than subscribers. Pick yours before you start, or you will read the result wrong.
Questions about this case study
Can I run OutlierKit research in bulk instead of clicking through it?+
Yes, that is what the API is for. This educator does almost nothing in the web app. Outlier searches and keyword research go straight from a script, a few hundred calls at a time, then works through the results offline. The API is on Pro and Max, and the docs list every endpoint.
How many searches does a batch like this involve?+
The two biggest sittings on this account were about 115 calls and about 320 calls. In between there are small top-ups of five or ten. The pattern is a planning session every few weeks rather than steady daily use, which suits anyone who writes a content slate in one go.
Did the subscriber count jump too?+
No, and that is worth stating plainly. The subscriber growth rate did not rise in this period. Views did, from an average of 282K a month to an average of 438K, including the best month in the window. Those are different outcomes and we are only claiming the one the data supports.
Do I need to write code to use the API?+
For bulk work, yes, though it is a handful of lines. If you want the same data without code, the MCP connector puts it inside Claude so you can ask in plain language and get ranked answers back. Many people start there and move to the API once they are running the same query every week.
Published with the channel owner's agreement, and kept anonymous. Figures come from their OutlierKit account and from public channel statistics, read 8 October 2026. Views rose over a period of heavy OutlierKit use, and we have not tried to measure how much of the rise came from the research.
More case studies
- How a fitness coach turned one viral run into 725K+ subscribers the same loop, run by hand each week.
- How agencies run YouTube research for a roster of clients the same batching, across a roster of clients.
Pull a month of research in one sitting
Outlier search and keyword research are both on the API, so you can run a whole content slate in one go. Start with the free trial.
Try OutlierKit Free