YouTube Studio replaces Trends with Research for creators
Find the new Research tab under Insights on desktop. A sample of four AI SEO videos shows how to interpret views, age and outlier scores.
YouTube Studio’s Research tab is under Insights on desktop. TeamYouTube announced on November 13 that it is replacing the Trends tab with a personalized Research destination. The new interface was available when SearchEngineAnswer checked on November 21. It can surface videos that performed strongly for their channel and let you filter ideas before committing to production. TeamYouTube’s announcement
For most creators, the first useful filter is format. A beginner’s course and a quick demonstration may answer related questions, but they ask for different amounts of viewers’ time. Compare long videos with long videos or Shorts with Shorts, then use channel size, publication age, and outlier scores to bring the examples into focus.
Find Research and run one useful query
Open YouTube Studio on a desktop browser. Select Insights in the left navigation and Research at the top. Enter a topic you can plausibly cover. Choose Videos or Shorts based on the format you would use for the next upload. YouTube’s Research instructions
Start with the problem your viewers bring to the channel. “AI SEO” is a useful discovery query for a publication that covers search, but it can return clips about writing, audits, and AI citations. Refine it to one task when you know what to make. For example, a query about tracking AI Overviews is more likely to yield videos where you can compare the test and evidence directly.
Use the upload-date filter when the subject is changing. For a new interface, a recent screen recording solves a much different problem from an old video with a high lifetime view count. Then try the outlier filter to find examples that exceeded their channel’s previous results. Change one filter at a time and keep a short note about what changed. That makes it easier to see whether a candidate appeared because the topic fit or because you widened the sample.
Read the view count and outlier score together
YouTube describes an outlier score as a comparison with a channel’s previous uploads. Its help page illustrates the idea with a long-form video that gets five times as many views as earlier long-form videos. Treat the baseline as channel-relative. The badge gives you a reason to inspect the video and the channel. It cannot isolate which title, timing, recommendation, or promotion produced the difference.
A high-view video with no badge in the results can still be relevant. A very large badge on a small channel can be a useful lead if the video solves the same viewer problem. Check whether the channel pivoted subjects, had an unusual launch, or has a history of comparable uploads. Public views also accumulate unevenly. Comparing a month-old upload with one published seven months ago tells you about their visible scale at the moment of capture. You need age-matched histories to compare early performance.
| Example from the AI SEO search | Views displayed | Age displayed | Outlier badge displayed |
|---|---|---|---|
| Surfer Academy and Matt Kenyon’s beginner course | 380K | Seven months | More than 100× |
| Sabrina Ramonov’s beginner AEO tutorial | 48K | Two months | 2× |
| Neil Patel’s AI SEO rules for 2027 | 43K | One month | No badge displayed |
| Matt Diggity’s SEO agent experiment | 10K | Three months | No badge displayed |
SearchEngineAnswer recorded this small sample from the Videos results on November 21. The examples came from four creator entries. The view figures are rounded as displayed, and no verified U.S. audience filter was applied. This sample shows that beginner tutorials and worked examples have reached viewers. It cannot show whether any of these videos were gaining views faster on the day of the check.
Separate search interest from video performance
Research puts relevant videos and channel comparisons in front of you. If your task is to compare the wording people search for, use the YouTube Search setting in Google Trends as a separate check. In SearchEngineAnswer’s five-term U.S. comparison, “YouTube SEO” had the highest displayed YouTube Search average at 13, ahead of “ChatGPT Search” at 4 and “Google AI Mode” at 3. These relative scores use a single scale for the selected phrases and time period. They are sampled and do not represent monthly search counts. Google Trends YouTube Search comparison Google Trends data FAQ
Changing the terms can shift the scale. Changing the property from YouTube Search to Web Search starts a different comparison. Keep the selected phrases, country, source, and observation date in a note next to the chart. Then return to YouTube Research to see which formats actually answer the query. A high relative index alone gives you little guidance on what to record.
Build a video around the evidence you can show
The sampled beginner courses cover a large subject. A narrower video can be easier for a small team to produce and easier for a viewer to finish. In this example, film one AI SEO audit. Run a check manually, give the agent the same page evidence, and show the first finding that can be verified. Then show one unsupported claim and explain what file, tag, or export would resolve it. The AI SEO audit checklist supplies a baseline readers can follow.
If you choose a Short, each example should fit in one visual proof. A canonical target or a screenshot of the actual interface can carry the whole explanation. If you choose a longer video, show how the finding was reproduced, which issues remained unknown, and the decision you made for the page. For search-led topics on TikTok, the creator discovery analysis uses a separate set of signals to decide what belongs in a short clip.
Before recording, write down the exact viewer question and name the piece of evidence you can capture yourself. Pick one upload-age window for review and compare against recent videos in the same format. If the result is weak, inspect whether viewers found the video through Search and whether the opening actually answered the question promised by the title. That is a practical use of Research even when the sampled outliers cannot predict how the next upload will perform.
Keep learning
Continue this topic
Earlier in this topic
Use ChatGPT to prepare personal-information removal requests safely
Tools & Workflows
Ask a question or join the discussion