SciTube Insights & Guides / Video performance

What does good look like on YouTube?

For research videos, success is not simply going viral. The useful question is whether the right audience discovers the work, understands the message and has a reason to take the next step.

Updated  ·  SciTube, a Science Diffusion brand
01 / Define success first

The three questions every research video should answer

A video can be doing a useful job with a modest audience, and can collect a large view count without delivering any meaningful engagement. Set the goal before reading the analytics.

Discovery

1. Did anyone discover it?

Review views, impressions, click-through rate and traffic sources. Ask whether the video is being found beyond your existing followers.

Engagement

2. Did they stay and watch?

Use watch time, average view duration and audience-retention patterns to see whether the explanation held attention.

Useful outcomes

3. Did it help the research?

Look for informed questions, classroom reuse, stakeholder enquiries, collaborator discussions and visits to the source paper.

Views are a starting point, not the final score. A 500-view animation used by a specialist clinical group could be more useful than a 10,000-view video watched by people with no connection to the intended audience.

02 / Compare responsibly

Why YouTube league tables can mislead

You may have seen comparisons of academic publisher and science communication channels ranked by average views. These can be interesting, but only if the data and sampling rules are available.

Our earlier version of this article presented a ranking and suggested that 3,000–5,000 views was a top-10% result. We cannot substantiate that percentile or reproduce a comparable cross-channel dataset from the information published there. We have removed those claims rather than treat an estimate as established fact.

QuestionWhy it matters
Same age of video?A video published yesterday has had fewer opportunities than one published two years ago.
Same format?Long-form videos, Shorts, livestreams and trailers attract different viewing behaviour.
Same subject and audience?A specialist pathology explainer should not automatically be benchmarked against general science entertainment.
Median or average?One breakout hit can push up an average. A median helps show what a typical upload achieves.
Paid or organic traffic?Promotion can influence counts and reach. Explain whether campaigns are included.
Comparable time window?Use a defined upload period and check performance at an agreed age, such as 30 or 90 days.

A credible benchmark should identify channels, selection criteria, upload dates, sample sizes, exclusions, the date metrics were collected and the treatment of promoted views. Without that, you cannot confidently claim one science channel is objectively the most successful.

03 / Independent evidence

What researchers have found about SciTube on YouTube

A 2026 study in Information Research examined SciTube's published YouTube content. It is a useful starting point because the authors explain what they studied rather than simply describing a few standout videos.

411 videos

SciTube uploads from 2019–2025 examined through YouTube metadata.

50 videos

A recent subset examined for narrative, visual, sound and accessibility features.

The researchers described SciTube as a professionalised model of science communication, but also identified weaknesses in metadata and accessibility. That is a useful reminder: well-made videos still need good titles, descriptions, captions and discoverability work.

Source: Font-Julián, Rodríguez-Rodríguez and Orduña-Malea (2026), Divulgación científica profesionalizada en YouTube: el caso de SciTube, Information Research, DOI 10.47989/ir31iConf64296 ↗. The study examines SciTube, not a representative sample of all research channels, and does not establish a universal view-count percentile.

04 / Practical improvements

Seven ways to make research videos more discoverable and useful

01 / Audience

Choose a real viewer

Write for students, clinicians, policymakers, collaborators or another defined group, not “everyone”.

02 / Opening

Lead with the question

Show the problem and why it matters before introducing a full list of methods.

03 / Title

Use searchable language

Translate a specialist paper title into terms an interested outsider might use.

04 / Thumbnail

Make the promise clear

Use one legible idea or visual. Don't promise more than the video delivers.

05 / Access

Add accurate captions

Review captions, names and technical terminology; provide accessible context where feasible.

06 / Research link

Connect to the paper

Include author attribution and a stable link or DOI to the underlying study.

07 / Distribution

Take it to the audience

Give partners, universities, teaching teams and project websites a reason to feature it.

Example: a research video after 30 days

Rather than writing “our video got 2,000 views”, a more informative report might say where viewers discovered it, what proportion watched meaningfully, which countries or professional groups were reached where identifiable, and whether the research team received relevant enquiries or requests to reuse the explanation.

Do not infer audience professions or impact from views alone. Combine analytics with actual responses, feedback and reported uses.

05 / Measuring over time

A simple 90-day research-video review

Choose a consistent measurement window and record what happened. YouTube's own analytics are more useful than a public view-count screenshot.

Day 0

At publication

Check title, description, thumbnail, captions, research DOI link and the first distribution plan.

Day 30

First month

Record views, traffic sources, retention and any referrals or genuine requests from the target audience.

Day 90

After three months

Look at continuing search traffic, institutional embeds, teaching reuse and relevant outcomes.

In YouTube Studio, the Reach report identifies traffic sources and search terms; the video analytics and content reports cover audience engagement and watch time. YouTube also cautions against treating CTR in isolation because it changes with audience and impression volume.

Platform documentation: YouTube Help — video reach ↗ · CTR and impressions ↗ · Content metrics ↗.

06 / Useful answers

Research video performance FAQs

How many YouTube views is good for a scientific animation?

There is no defensible universal number. Start with the intended audience and compare similar videos over the same measurement period. Views alone cannot establish understanding, quality or research impact.

Is 3,000–5,000 views in the top 10%?

We do not have a verified cross-platform dataset that supports this percentile. Avoid presenting it as fact.

Should I compare my channel with a major publisher?

Only cautiously. Different channel ages, audience sizes, promotion methods, Shorts mixes and topics make straightforward rankings misleading.

What is more useful than view count?

Watch time, retention, traffic sources and evidence that the target audience actually used or responded to the video. No single metric tells the whole story.

Do promoted views count?

Promotion can create genuine exposure, but a fair report should distinguish promoted and organic performance where the data allows.

How can a researcher help after publication?

Share the video with relevant departments, collaborators, educational programmes, professional groups and project partners. See our animation examples and publication guidance.

Your research, clearly explained

Make the video useful. Then help it be found.

SciTube combines scientific editing, professional animation, publication and dissemination, with researchers involved in the approval process.

Explore SciTube animation options →