Key takeaways
- YouTube's Trends tab is becoming Research, part of a Studio update that began rolling out gradually in July 2026 — and the headline addition is a number the paid tools have sold for years.
- An outlier multiplier is a ratio, not a score: it compares one video's views against what that channel normally gets. A 40x tells you the video escaped its channel. It tells you nothing about whether it would work on yours.
- Small channels manufacture enormous multipliers by accident. The denominator is the weak point — a 20-view average makes 300 views a 15x, and 300 views is not a signal.
- The genuinely new data is not the multiplier. It is the filter that isolates what your own viewers are watching elsewhere, which is the closest thing on YouTube to a transfer guarantee.
- A multiplier scores the whole package at once. Before you copy anything, separate the idea from the packaging — and the fastest tell is whether the outlier's thumbnail looks like that channel's other thumbnails.
- Research finds candidates. It cannot tell you which version of your own image wins, and nothing except a live test can.
For about five years, the most valuable manual technique in YouTube research was also the dullest: open fifteen channels roughly your size, sort their uploads by views, and look for the video that beat its own channel by a factor of ten. Everyone who did it found things. Almost nobody did it, because it takes an hour and produces a spreadsheet.
The paid tools automated it and charged for it. Now Studio is doing it in the product. YouTube's update notes describe the Trends tab as evolving into Research, a destination for planning your next video, with the updated Studio experience rolling out gradually from July 2026. In August 2026, Tubefilter reported that the tab had gained two things creators had been asking for: outlier multipliers on the videos it surfaces, and data on what a creator's own viewers are watching. Videos in the tool carried multipliers ranging from 4x up to 132x, with filters for view count, outlier score and video age. It is an experiment, currently with a small number of users.
Here is the problem with getting the number for free. A multiplier is a ratio, and ratios are the easiest statistic in the world to misread. The creators who lose money on this feature will not be the ones who never open it. They will be the ones who open it, sort by highest multiplier, and make the video at the top.
What actually changed in Studio
Three separate changes landed close enough together that they get confused with each other. They do different things, and only one of them is about finding ideas.
| Surface | Status | What it is for |
|---|---|---|
| Trends → Research | Rolling out gradually from July 2026 | Searches, topics and — in testing — breakout videos with outlier multipliers |
| Inspiration tab | Being deprecated from August 2026 | Idea cards, outlines, suggested titles and thumbnails |
| Ask Studio | Named as the replacement for brainstorming | Conversational questions answered against your own channel data |
| Analytics → Insights | Test, small group of creators | Regrouped metrics plus AI cards: Channel Summary, Content Patterns, Audience Loyalty, Video Summary |
The Inspiration tab going away matters more than it looks. It was the browsable half of YouTube's idea tooling — a feed of cards you could scan without forming a question first. What replaces it is a chat box, which is a different cognitive act: Ask Studio answers what you ask, and the quality of the answer is capped by the quality of the question. We covered that trade-off in detail in YouTube's AI tools for creators, and the short version still holds — these tools are decent analysts and mediocre strategists.
Research is the surface that survived, and it now holds the one piece of data that is genuinely hard to get anywhere else.
What an outlier multiplier actually measures
The definition is simple enough that most people skip it and then misuse it. A multiplier compares one video's views to what the rest of that channel's videos normally do — in the Research tab's case, based on the video's first seven days. If a channel averages 50,000 views and one video does 600,000, that is a 12x.
Third-party tools compute the same shape of number. vidIQ's outlier score compares a video's view count to the channel's average within a similar timeframe, and its own documentation uses the cleanest possible example: a channel that usually gets 1,000 views posting a video that gets 10,000 has scored a 10x, while a channel that usually gets 10 million views posting a video that gets 10 million has scored a 1x. The second video has a thousand times the views and a tenth of the signal.
That inversion is the entire point of the metric, and it is why raw view counts are a bad way to research. A million views from a channel that always gets a million views is a normal Tuesday. It tells you the channel has an audience. It does not tell you the video did anything.
The arithmetic, two channels at once
Put two channels side by side and the behaviour of the ratio becomes obvious.
| Channel | Normal views | This video | Multiplier | What it is worth |
|---|---|---|---|---|
| A | 20 | 300 | 15x | Nothing. 300 views is noise, not discovery. |
| B | 40,000 | 600,000 | 15x | A lot. Browse and suggested carried it past the subscriber base. |
| C | 5,000,000 | 2,000,000 | 0.4x | A flop with more views than most channels will ever see. |
Channels A and B have identical multipliers and nothing else in common. This is the first and most expensive misreading: sorting by multiplier descending puts A above B, because tiny denominators produce spectacular ratios for free. Any filter you set on the Research tab should be paired with a floor on absolute views — whatever counts as real distribution in your niche. Below that floor, a 15x is an accident of arithmetic.
Five ways a multiplier lies to you
None of these are reasons to ignore the number. They are the things you check before you act on it, and they take about ninety seconds each.
- The denominator problem. A new channel, a channel that posts rarely, or a channel whose last ten videos all underperformed will throw huge multipliers off modest videos. Always read the multiplier next to the raw view count, and treat a high ratio on low absolute views as a curiosity rather than evidence.
- Baseline contamination. The channel average that forms the denominator is assembled from whatever that channel has been uploading. If it mixes Shorts, streams, podcast episodes and produced videos, its "normal" is a blend of four different distribution systems. A 20x against that blend may just mean "this was the long-form one".
- The seven-day window. Scoring the first week favours videos that spike and penalises videos that compound. A tutorial that quietly accrues search traffic for three years can be the most valuable asset on a channel and will never post an impressive multiplier. If your channel's model is evergreen search, an outlier list is structurally biased against the kind of video you are trying to make.
- No counterfactual. You are looking at a list of things that worked. You cannot see the eleven channels that made the same video in the same fortnight and got nothing, because nothing is not surfaced. Every outlier feed is a survivorship machine, and survivorship inflates your confidence in whatever the winners have in common.
- Borrowed distribution. Some outliers are caused by something that is not repeatable: a collaboration with a larger channel, a news event, a link from a large subreddit, a mention in someone else's video, a platform promotion. The video is not the cause. Check the upload date against anything that was happening that week, read the top comments for referrals, and look at whether the channel's next three videos held any of the gain. If they did not, the format did not travel — an audience visited.
The one-question test before you copy anything
Ask: would this video have worked if a different channel made it? If the answer depends on the host's face, their access, their guest, or an event that happened once, the multiplier is describing a moment rather than a format. Moments cannot be reproduced. Formats can.
The filter that matters more than the multiplier
The more interesting half of what Tubefilter described is not the score at all. It is the ability to see what videos and creators your own viewers are watching, and to isolate outlier data to that group.
That is a different class of information. A niche-wide outlier tells you something worked for somebody's audience. An outlier among the people who already watch you tells you something worked for your audience, on a channel that is not yours. The gap between those two statements is most of the risk in content planning.
Used properly, that filter answers three questions no keyword tool can:
- Which channels are your real competitive set? Not the ones in your niche by topic — the ones your viewers actually spend time with. These are frequently adjacent rather than identical, and the adjacency is where your next format lives.
- What is your audience's appetite that you are not serving? If your viewers are reliably watching a format you have never made, that is a content gap with your own audience attached to it, which is far stronger evidence than a search-volume estimate.
- Where is your packaging losing to someone else's? When your viewers watch a competitor's video on a topic you have also covered, the topic is not the variable. Open both thumbnails side by side and the answer is usually visible in four seconds.
The older Research surfaces still do useful work here too. Viewer searches, searches across YouTube, and the content-gaps filter — which YouTube describes as topics where demand is not well served by the results that exist — remain the cheapest way to find phrasing that people actually type. The outlier data tells you what formats are travelling; the search data tells you what language to package them in.
From a 40x to a decision
Here is where most research routines stop and where the value actually starts. A multiplier scores an entire package: topic, format, title, thumbnail, timing, host and channel authority, all fused into one number. Your job is to work out which of those components did the lifting, because that is the only part you can transfer.
Separate the idea from the packaging
There is a quick diagnostic that costs nothing. Open the outlier video's thumbnail, then open the same channel's previous ten thumbnails.
- If the outlier's thumbnail looks like the rest of them, the packaging was held constant and the idea carried the video. The transferable asset is the topic or the format, and you should be studying the title and the premise.
- If the outlier's thumbnail is visibly different — a face where there are usually no faces, one word where there are usually six, a single dominant colour against an otherwise muted set — then the package changed and the result changed with it. The transferable asset is the design decision, not the subject.
That second case is far more common than creators expect, and it is the cheapest lesson available on the platform, because it is the one you can apply this week to a video you were making anyway. Our breakdown of thumbnail composition covers what to look for once you are in there: subject scale, the direction of the gaze, how much of the frame is doing nothing, and whether the text is a label or a hook.
The three questions that decide whether it transfers
Before an outlier becomes a production slot, it has to survive three questions. Most do not.
- Can I make this better than they did? Better evidence, better access, better demonstration, better production, or simply more specific knowledge. If your version is a worse copy published later, the algorithm has no reason to prefer it and neither does a viewer.
- Does my audience have a reason to accept it from me? Channels have permissions. A 30x cooking video on a channel your viewers trust for cooking does not transfer to your camera-gear channel because the multiplier was high.
- Can I package it differently and still honestly? If the only way to win is to reproduce their thumbnail, you are entering a fight where they own the recognisable version. Find the second angle on the same promise. That is the whole skill.
The log that makes this compound
Research that is not written down gets relearned every quarter. One row per outlier, and after about twenty rows the pattern in your niche stops being anecdotal.
| Column | Why it earns its place |
|---|---|
| Channel, video, upload date | Lets you check what happened to the channel after the outlier |
| Views and multiplier | Both, never one — the ratio without the absolute number is unreadable |
| Watched by my viewers? | The strongest single predictor of transfer you have access to |
| Idea or package? | Your verdict from the ten-thumbnail comparison above |
| Promise in one line | Forces you to state what the viewer was actually being offered |
| Thumbnail note | Subject, text word count, dominant colour, what the frame withholds |
| Borrowed distribution? | Collab, news, external link — the disqualifier that saves you a week |
Where this leaves the paid outlier tools
Honestly: less differentiated than they were, and still useful for a specific job. The multiplier itself was never proprietary — it is division. What third-party suites sell is coverage and monitoring across channels you do not own.
| Studio Research | Third-party outlier tools | |
|---|---|---|
| Your own viewers' watching | Yes — nobody else has this | No |
| Arbitrary competitor tracking | Limited | Yes, with alerts and saved lists |
| Historical databases and exports | No | Usually |
| Availability | Rolling out; outlier data in testing | Immediate, on a subscription |
| Cost | Free | Monthly, and the outlier features sit in the upper tiers |
If you are paying purely for a multiplier, that subscription is getting harder to justify as this rolls out. If you are paying for alerting across forty tracked channels, it is not.
What Research cannot tell you
It cannot tell you which of your own thumbnails wins. That is not a limitation of this feature; it is a category difference. Research is descriptive data about videos that already exist, gathered from channels that are not yours, judged by audiences that are only partly yours. The only instrument that answers the packaging question for your video is a live test against your own impressions.
YouTube's Test & Compare runs up to three thumbnails or titles against real traffic and settles it on watch time per impression rather than raw clicks — the mechanics, the sample sizes and the failure modes are in our guide to winning the three-thumbnail test. It is slow, and it is the only source of truth you have. An outlier feed proposes; a test disposes.
The second thing Research cannot do is tell you whether your click-through rate problem is a packaging problem at all. A low CTR on a video whose topic nobody wanted is a demand problem wearing a design costume, and the diagnosis is different in every case — what counts as a good CTR depends almost entirely on where the impressions came from, which is why the same number means two different things on two surfaces.
If you do not have the new tab yet
Most creators will not see outlier multipliers for a while — it is an experiment on a small number of users, and YouTube has not said whether it will expand. That is not a blocker, because the arithmetic is public and takes an hour.
Pick fifteen to twenty channels within roughly a factor of three of your size. For each, note the median views across the last ten uploads, then look for any video above three times that median in the last ninety days. Divide, sort, and you have reproduced the feature by hand, with one advantage: you chose the channel set, so you know exactly what your denominators are made of. The full manual routine, alongside eight other free trend signals, is in our guide to finding trending video ideas.
The weekly routine, timed
Thirty minutes, once a week, with a hard stop. Research expands to fill whatever container you give it, and it is more pleasant than filming, which is exactly why it needs a timer.
- Five minutes. Set filters: recent window, an outlier floor around 3x to 5x, and an absolute view floor that means something in your niche.
- Ten minutes. Scan the "watched by my viewers" cut first. Everything there is pre-qualified by the only audience whose behaviour predicts yours.
- Ten minutes. Log three to five rows. Run the ten-thumbnail comparison on each. Mark idea or package.
- Five minutes. Kill anything with borrowed distribution, then pick one candidate for the following month's schedule. One. The constraint is the point.
Twenty rows in and the log starts doing something no individual idea does: it tells you what your niche rewards structurally, which formats keep reappearing at the top, and which thumbnail patterns show up on outliers but not on the same channels' ordinary uploads.
Frequently asked questions
What is a good outlier multiplier?
Above 3x on a recent upload usually means the video travelled past its own subscriber base into browse or suggested. Above 10x is worth studying frame by frame. But the threshold is meaningless without an absolute view floor, because a channel with a tiny average produces double-digit multipliers from videos that barely got distributed at all.
Does the Research tab show outlier scores for everyone?
Not yet. The Trends-to-Research change has been rolling out gradually since July 2026, and the outlier multiplier and viewer-comparison data were reported in August 2026 as an experiment with a small number of users. If you do not see it, the manual method above computes the identical number.
Is the Inspiration tab gone?
It is being deprecated from August 2026, with Ask Studio named as the place to brainstorm ideas instead. The practical difference is that you now have to arrive with a question rather than browse a feed of cards.
Should I copy a video with a high outlier multiplier?
Copy the mechanism, never the artefact. Work out which component earned the result — topic, format, title or thumbnail — then ask whether you can do that component better for an audience that has a reason to accept it from you. A worse copy published two weeks later is the most common wasted upload on the platform.
What the number is actually for
The outlier multiplier is a filter, not an answer. It takes the thousands of videos published in your niche this month and narrows them to the handful that did something their own channel could not normally do. That is a real service, it used to cost money, and it is worth the thirty minutes a week.
What it does not do — what nothing in Studio does — is decide what your version looks like in a feed next to theirs. By the time a viewer sees your video, all of this research has collapsed into one image and one line of text, competing against four other images on a phone screen, in about a second and a half. The research tells you which fight to pick. The packaging decides whether you win it.
That is the part worth spending your remaining hours on. Generate several genuinely different directions for the image rather than one you defend out of fatigue — Thumblore exists for exactly that step — then let a real test settle which one travels, using the process in our thumbnail A/B testing system. The outlier told you the idea can win. Only the image gets it clicked.