Find keywords for YouTube Shorts by combining YouTube search predictions, the terms that already bring viewers to your channel, and questions your audience needs answered. Narrow each candidate to one thing a Short can show clearly.
These eleven methods cover channels with existing analytics, new creators with little data, and businesses that can draw on website or customer questions. Each source tells you something different, so keep its limits attached to the phrase. The coffee examples below are illustrative wording, not verified keyword suggestions, search-volume estimates or tested results.
Choose evidence before choosing a keyword
Use a source you can actually access, then record what it establishes.
| Your starting point | Useful sources | What the evidence means |
|---|---|---|
| Little or no channel history | YouTube predictions and relevant Shorts results | Candidate wording and visible answer formats |
| Shorts already receive search traffic | Your YouTube search-terms report | Queries that found your content |
| Studio has audience insights available | Top searches, Shorts gaps and Audience reports | Interest or viewing patterns within the report's scope |
| You have a website or customer conversations | Website queries and recurring questions | Relevant leads that still need checking on YouTube |
When a Studio card is missing, record the source as unavailable and choose an accessible method.
Keep search discovery separate from feed distribution. YouTube's Shorts discovery guidance explains that search considers metadata match and whether viewers click and watch, while Shorts recommendations depend on viewer personalization and performance. A well-chosen phrase helps describe a useful video; it cannot guarantee that the feed will distribute it.
1. Collect YouTube search predictions
Type a seed topic into YouTube's search box and record relevant predictions before submitting the search. Expand the seed with a genuine question or task modifier rather than collecting every variation.
For a coffee channel, you might begin with espresso sour, then investigate wording around why it happens or how to recognize it. Copy only predictions that actually appear in your research session; the seed here is an example.
Save the phrase, language, market, date and sign-in context. YouTube says search predictions can reflect other people's searches, entered terms and personalization. Signing out does not guarantee a completely unpersonalized list.
Choose a phrase you can answer with a specific demonstration. A suggestion is a wording lead, not a monthly search count or evidence that the keyword is easy to rank for.
2. Read Shorts search results for specific viewer tasks
Search a candidate phrase on YouTube and examine the relevant Shorts that appear. Record the question each one answers, its title wording and what the viewer can actually see demonstrated. Keep Shorts and longer videos separate in your notes.
Look for a useful distinction the existing results blur. For example, sour vs bitter espresso suggests a comparison, while espresso setup for beginners could require a much broader explanation. A Short might show one comparison clearly without promising an entire course.
Use competing titles to discover language, not to copy another creator's presentation. A high-view Short does not reveal which search terms supplied its views, and it does not establish that a phrase caused its success.
Save a candidate only when you can name the viewer's task and offer a concrete answer or demonstration.
3. Use your audience's Top searches in Studio
In YouTube Studio, open Analytics > Trends, following the current Trends documentation. Explore a relevant topic and save useful searches. The Top searches card draws on your audience and saved searches over the last 28 days.
Use it to select a topic worth investigating, then narrow that topic into a Short-sized question. If a broad espresso-preparation topic is relevant, your candidate might focus on one preparation step you can show clearly.
This is an audience-interest lead, not proof of Shorts-only demand. Check the candidate against actual Shorts results before filming.
YouTube is updating the Studio experience, and some insights depend on country, language, device and available relevant data. If the card is absent, record that limitation instead of assigning an invented interest level.
4. Investigate Content gaps for Shorts
Where Studio provides Content gaps for Shorts, review the suggested search terms and the underlying viewer problem. The same Trends guide describes gaps as searches without enough suitable results, including missing exact matches or results that are old or low quality.
This answers a different question from Top searches: what useful answer might be missing? Inspect the current results yourself before accepting the opportunity.
For example, a broad espresso guide may mention several adjustments while failing to show the specific comparison a beginner needs. A candidate Short could supply that missing visual explanation. Record the exact gap you observed and how your video would address it.
A gap label is not a low-competition score or a ranking guarantee. Discard suggestions you cannot answer accurately.
5. Reuse search terms that found your own Shorts
Open YouTube Studio > Analytics > Content, select the Shorts view, and inspect how viewers found your Shorts. The Shorts analytics guidance identifies a YouTube search-terms report containing wording viewers used to find your content.
Record the date range, the Short or group analyzed, and the relevant returned phrase. Look for a follow-up question the current video leaves unanswered. If viewers found a basic espresso explanation, a useful next candidate might isolate a specific comparison rather than repeat the same explanation with a new title.
Keep YouTube search traffic separate from Shorts-feed views. These terms describe discovery of your existing content, not total demand across the platform.
This method needs usable channel data. When a report is missing or contains too little evidence, leave the research status unresolved; total views alone cannot supply the missing keywords.
6. Explore what your audience watches elsewhere
In Studio, open Analytics > Audience and look for What your audience watches. YouTube's Audience documentation describes outside-channel videos, Shorts, live streams and podcasts your viewers watched over the past seven days.
Read the topics and title wording, then identify a related question your channel can answer. If viewers watch broad coffee-equipment comparisons, a candidate Short might explain one visible difference between two preparation tools.
Keep the original format in your notes. Interest in a long comparison does not establish that the same audience wants a shortened version with missing context.
This report reveals viewing interests, not the search queries that generated those views.
7. Mine viewer comments for unresolved questions
Review comments on your relevant Shorts and longer videos. Look for requests to show a step, explain a term, compare two options or correct a misunderstanding. Group comments by the task they describe rather than counting every wording variation as a separate topic.
An illustrative comment such as “Can you show the difference instead of just naming it?” points to a visual explanation the existing video did not provide. Turn that into a specific candidate phrase, such as sour vs bitter espresso, then check the YouTube results and plan the demonstration.
Keep the original video's context alongside the idea. A question about a particular machine may require that model name rather than a generic coffee title.
Comments establish that someone asked a question. They do not supply YouTube search volume or prove that every viewer shares the problem.
8. Check Google Trends with YouTube Search selected
In Google Trends, enter a relevant term, choose your target country and time range, and change Web Search to YouTube Search. Google's advanced Trends lesson explains this search-type filter.
Compare candidate phrases under the same settings. Keep a search term distinct from a broader topic when interpreting the comparison. Then inspect the related queries for useful wording and check whether interest looks seasonal, persistent or tied to a short-lived event.
For example, you could investigate whether a coffee question is relevant throughout the year or mainly around a particular season. Record what the actual graph shows rather than assigning an imagined score to the example.
Google Trends data is sampled and normalized on a 0–100 scale. A score of 100 does not mean 100 searches. This filter covers YouTube searches, not Shorts-only searches, and insufficient data does not establish zero demand.
9. Turn website search queries into video questions
If you have an authorized website property in Google Search Console, inspect queries associated with a relevant help article or product guide. Look for questions whose answers would become clearer through a visible demonstration.
A coffee-accessory business might find a troubleshooting question connected to an existing guide. That phrase is a candidate for a Short only if the video can answer a specific part of the task accurately.
If you already use Screpy's Search Console dashboard, you can review the connected website's query and page evidence there. Preserve the property, date range and market context when copying selected phrases into your research notes.
These are Google searches involving your website, not searches inside YouTube. Any accompanying website keyword estimates also belong to their own scope. Recheck the wording and available Shorts results on YouTube before treating the idea as a video opportunity.
10. Use Google question suggestions as wording leads
Search a relevant question stem on Google and examine the predictions that actually appear. This can uncover how people outside your existing YouTube audience describe the problem. Google's autocomplete explanation describes factors including language, location and trending interest.
Keep only questions that benefit from a visual answer. For example, a candidate about recognizing a preparation mistake may fit a demonstration better than a question requiring a detailed purchasing guide.
Then run the candidate on YouTube and inspect the relevant Shorts results. Record its origin as Google wording until YouTube evidence supports a stronger conclusion.
Google predictions are not YouTube search-volume data. Adding “Shorts” to a web query does not validate demand; the useful output is a clearer viewer question you can investigate.
11. Collect recurring customer and support questions
Review questions your business is authorized to use from support conversations, sales calls, onboarding sessions or customer FAQs. Extract the task and remove personal details before turning it into a public content idea.
This source can reveal product-use questions that never appeared under a video. For a coffee-accessory business, an illustrative candidate might be how to check whether an espresso basket fits. The Short would need a clear, accurate demonstration using the relevant product context.
Group repeated requests by the underlying job. If you count occurrences, record the period and the records reviewed so the figure remains a customer-research observation.
Customer frequency can establish practical relevance to your business. It does not establish YouTube search volume. Check the phrase on YouTube and prioritize questions you can answer usefully without exposing private customer information.
Turn the shortlist into a Short you can evaluate
Keep the phrase, its source and the intended answer together. Copy this CSV into a spreadsheet; both rows are fictional examples, and the empty date is for your actual research date.
candidate_phrase,source,evidence_status,market,checked_on,short_promise,next_check
"sour vs bitter espresso","example viewer comment","illustrative; not researched","US / English","","Show one side-by-side comparison","Inspect relevant YouTube results"
"how to check whether an espresso basket fits","example customer question","illustrative; not researched","US / English","","Demonstrate the relevant fit check","Verify product context and YouTube wording"
Before selecting a phrase, confirm three things: it fits your audience, you can deliver the promised answer, and you know what the evidence does and does not establish. Several phrases describing the same task can belong to one planned Short. If you also maintain related website content, an editorial keyword map can help preserve those relationships.
Use a clear title that describes what the viewer will get, and write the description naturally. Avoid turning the research list into a block of tags. YouTube says tags play a minimal discovery role, except when they help with commonly misspelled content.
After publishing, compare the relevant Shorts over consistent reporting periods. Use the traffic-source breakdown to separate YouTube search from Shorts-feed discovery, then examine viewing behavior and feedback. A feed-driven view increase alone does not show that a keyword caused it.
The next research decision should follow that evidence: keep a useful question, improve an unclear answer, or investigate a genuinely different viewer task.