F1 YouTube Research: What Performs Best?

Stories, technical breakdowns and titles that promise an explanation perform best in this dataset. If I were planning videos for a Formula 1 channel, that is where I would start.

I analysed 560 videos from 14 Formula 1 channels to answer a few practical questions: Which formats perform best? Which drivers attract attention? What should a title promise? And do more cars make a thumbnail stronger?

To make the comparison fair, I measured every video against the normal performance of its own channel. That means a smaller channel can compete with a larger one in the analysis.

Here is a simple example: if a channel normally gets 100,000 views and a video gets 124,000, that video performs 24% above the channel baseline. Every chart below uses this comparison. One unusually large video cannot distort the result because I use the middle result for each group.

1. The Formula 1 video formats that perform best

Format is one of the biggest choices you make before creating a video. The same team upgrade could become a quick news update, a technical explanation or a story about the team trying to solve a problem.

How videos perform across different Formula 1 formats

How videos perform across different Formula 1 formats: typical views compared with channel baseline
Category	Videos	Versus baseline
Driver & Team Stories	165	+18%
Technical Breakdowns	87	+13%
Controversies & Drama	79	−4%
Formula 1 History	63	−6%
Formula 1 News	43	−10%
Race Previews & Recaps	50	−12%
Rankings & Comparisons	42	−21%
Other Formats	31	−49%

Stories and technical breakdowns were close, so I would not declare one the clear winner. The more useful takeaway is that both beat the baseline, while faster news and race coverage fell below it.

If your upload schedule mainly follows the race calendar, try adding videos that go deeper. Instead of only reporting a new upgrade, explain the problem it is meant to solve, why earlier attempts failed and what makes this version different.

Give viewers something useful to understand or a story they want to follow.

Rankings & Comparisons were 21% below baseline. That surprised me more than the news result. Lists feel like a natural fit for sports channels, but the ones in this collection didn’t lead the typical-view ranking.

I’d use the format results to plan a small series, then compare those uploads with my channel’s usual numbers. A few stories. A few technical breakdowns. Topics I can actually deliver well. The data gives me a direction for that experiment.

2. The Formula 1 drivers that attract the most clicks

After choosing a format, the next question is simple: who should the video be about?

How videos perform when drivers are named in the title

How videos perform when drivers are named in the title: typical views compared with channel baseline
Category	Videos	Versus baseline
Kimi Antonelli	15	+106%
Charles Leclerc	9	+37%
Max Verstappen	32	+30%
Oscar Piastri	8	+8%
George Russell	12	+6%
Lewis Hamilton	21	−23%

Antonelli is the biggest number here, with 15 videos behind it. Leclerc’s group is smaller still: nine. Verstappen’s result has 32 videos, which gives me more examples to work through before choosing an angle.

That’s how I’d use this chart. Antonelli goes onto my research shortlist. Verstappen does too. Then I’d read the titles and look for the actual story: a career turning point, a performance people want explained, a rivalry, a decision with consequences.

The name gets me to the right set of videos. The next step is finding what those videos offered.

Lewis Hamilton’s title group landed 23% below baseline. I’d take that as a reason to inspect the represented stories more closely. Fame by itself wasn’t enough to put this group above its channels’ usual numbers.

The useful finding is a shortlist of subjects with different results in this collection. It isn’t a universal ranking of driver popularity. Antonelli’s 15 titles, for example, may capture a particular period of interest as much as an enduring preference.

3. The title promise that performs best

A strong title does more than name a topic. It tells viewers what they will get from the video.

How videos perform with different title promises

How videos perform with different title promises: typical views compared with channel baseline
Category	Videos	Versus baseline
Success & Improvement	36	+39%
Explanations & Reasons	142	+24%
Duels & Comparisons	23	+14%
News & Changes	26	+5%
Results & Reveals	40	−2%
Problems & Failures	189	−4%
Straight Information	59	−10%
Predictions & Future Questions	45	−29%

Compare these two angles: “Will this upgrade save the season?” asks viewers to watch a prediction. “Why this upgrade fixes the car’s biggest weakness” promises a clear explanation.

These are examples, not titles that were tested against each other. But the difference is useful: one guesses what might happen, while the other offers something specific to understand.

Before writing a prediction, look for the explanation. What changed? Why did it happen? What did everyone miss? Most importantly, make sure the video can deliver the answer.

Success & Improvement led the raw title ranking at 39% above baseline, based on 36 videos. Problems & Failures were 4% below. That’s useful context when choosing a story, but I wouldn’t turn it into a rule that every title should sound positive. The closer comparison didn’t establish a separate advantage from positive wording.

The part I’d carry forward is the promise of understanding. A failure can make an excellent explanation. So can a success. I’d rather find the reason people want to know than force either tone onto the subject.

4. What works in Formula 1 thumbnails

A popular subject does not automatically make a strong thumbnail. Verstappen is the clearest example: videos naming him in the title perform 30% above baseline, but videos showing him in the thumbnail perform 10% below.

These groups overlap, so this does not mean that adding his face reduces views. It shows something simpler: choosing the subject and choosing the thumbnail image are two separate decisions.

How videos perform when drivers appear in the thumbnail

How videos perform when drivers appear in the thumbnail: typical views compared with channel baseline
Category	Videos	Versus baseline
Charles Leclerc	54	+24%
Kimi Antonelli	45	+23%
Lance Stroll	21	+11%
George Russell	46	−1%
Oscar Piastri	29	−9%
Max Verstappen	109	−10%
Fernando Alonso	50	−11%
Lando Norris	55	−11%
Lewis Hamilton	88	−13%
Sergio Pérez	23	−41%

I’d use those numbers to choose examples to inspect, then ask what the image is doing. Is it showing the person the story is about? Is there a visible tension? Does it communicate something the title leaves open?

I’d also resist relying on a team badge to do that job. Ferrari thumbnails sat at baseline in this collection. Mercedes was 6% above, Red Bull 2% below and McLaren 9% below. The brand alone wasn’t a strong shortcut to the top of the ranking.

Then there was the question about cars. More cars can mean more context, a comparison or a busier scene. Did that translate into stronger typical views?

How videos perform by the number of cars in the thumbnail

How videos perform by the number of cars in the thumbnail: typical views compared with channel baseline
Category	Videos	Versus baseline
One Car	152	+14%
Multiple Cars	70	+5%
No Cars	338	−3%

Both car groups were above baseline, but one car led this comparison. If a thumbnail currently has three cars competing for attention, I’d try a version built around one clear car and one clear idea. That’s an experiment suggested by the pattern, rather than a guaranteed increase in views.

After all four questions, the next video I’d research is quite concrete: a driver or team story, or a technical breakdown, with a specific explanation in the title. Antonelli and Verstappen are subjects I’d investigate first. The thumbnail would have a clear job within that story.

That’s what I wanted from the analysis. Enough evidence to choose where to spend the next few hours of research. Then a better brief, a better video, and a result I can compare with my own channel.

Data, methods and statistical checks

The charts are descriptive rankings of this collection. The adjusted tests ask whether selected differences remain after accounting for measured channel and video characteristics.

Videos: 560
Channels: 14
Snapshot: Oct 9, 2026
Eligible tests: 9 of 12

Download chart data (CSV)

Dataset and scope

The collection contains 560 unique videos from 14 channels, with no duplicate video IDs or zero-view records. Video ages ranged from approximately 0.8 to 1,013 days; 69 videos were under 30 days old, 15 were under seven days old and 35 were longer than 30 minutes. Views and classifications matched between the supplied CSV and database.

This is a selected collection, not a random sample of all Formula 1 YouTube. All 560 videos also appear in the earlier 610-video collection, so this is not an independent replication. One 50-video channel from that collection is absent.

How “typical views” is calculated

Relative views = video views ÷ median views for that channel

Each chart shows the median relative-view ratio for a group, converted to a percentage: (group median − 1) × 100. This is not an arithmetic average of raw views, a historical channel-wide average or an estimated causal lift.

The descriptive charts do not adjust for video age. Normalising by channel makes channel sizes more comparable, but it does not make video ages, topics or category coverage equivalent.

Classification

Labels were assigned automatically from titles and thumbnails; the videos were not watched to validate format. Driver and team stories are combined, as are previews and recaps. No manually labelled reference set was available to estimate classification accuracy.

People and team aliases were normalised before counting. A video can belong to several entity groups. Four derived team-presence flags disagreed with their underlying entity lists, so the underlying lists were used. Driver-title groups require at least eight videos; thumbnail-driver groups require 20. These are reporting cutoffs, not significance thresholds.

Adjusted comparisons

Models used log relative views, channel fixed effects, a cubic B-spline for log video age with three interior quartile knots, and log duration. Non-format comparisons also controlled for format; the global Verstappen-thumbnail comparison additionally controlled for his presence in the title.

Nine of 12 research questions met the eligibility rule: at least 20 videos on each side and at least five channels containing both sides. The questions were not prospectively preregistered. Results use approximate 95% channel-clustered CRV1 intervals and Benjamini–Hochberg corrected p-values from restricted wild-cluster bootstrap tests (9,999 Rademacher draws, seed 20261009). The intervals are not simultaneous or bootstrap-inverted.

Adjusted comparisons against the named comparison group, not the chart baseline.
Comparison	Videos	Ratio	95% interval	Corrected p
Technical Breakdowns vs. Formula 1 News	130	2.09×	1.18–3.70×	0.114
Stories vs. Formula 1 News	208	1.40×	0.79–2.49×	0.399
Technical Breakdowns vs. Stories	252	1.19×	0.75–1.88×	0.526
Success vs. Problem Promises	225	1.01×	0.62–1.66×	0.961
Explanation vs. Problem Promises	331	1.37×	0.94–2.00×	0.247
Verstappen Titles vs. Other Named People	144	1.72×	1.30–2.27×	0.114
Verstappen Shown vs. Not Shown	560	0.83×	0.63–1.11×	0.352
Multiple Cars vs. One Car	222	0.66×	0.47–0.94×	0.114
Large Faces within Stories	165	0.79×	0.43–1.44×	0.663

No adjusted comparison met the 5% threshold after correction. The broader screen of 92 contrasts across 79 unique partitions (4,999 draws) also produced no discoveries, including under the more conservative Benjamini–Yekutieli correction.

Sensitivity checks

Formats: Technical Breakdowns versus Formula 1 News fell from 2.09× to 1.36× after removing videos longer than 30 minutes. Stories versus News fell from 1.40× to 1.09×. Only five channels supplied both technical and news formats, and only two had at least five videos on each side.
Drivers and cars: Verstappen titles versus other person-naming titles remained around 1.77–1.82× in selected checks. Multiple cars versus one car moved from 0.66× to 0.76× within the 3–30 minute subset.
Title promises: The raw success-versus-problem ratio was 1.44×; the adjusted estimate was 1.01× with a wide 0.62–1.66× interval. This does not establish equal performance.

These estimates are exploratory and did not pass the multiple-testing correction. The large-face comparison was also inconclusive. Showing Verstappen within Verstappen-title videos, depicting a named person and technical-component focus within technical videos lacked enough eligible channel coverage.

Limits and intended use

The results describe associations in this collection. They do not show that changing a title, adding a driver or removing a car causes more clicks. The data does not include impressions, click-through rates, retention, traffic sources or title and thumbnail histories. With only 14 channels and uneven category coverage, statistical power and cluster-based uncertainty remain limited.

The findings are best used to prioritise research: separate subject choice from image choice, distinguish explanations from predictions, and compare formats within channels and similar age and duration ranges. A packaging intervention still needs a controlled test.

Sources and reproducibility

The downloadable aggregate CSV contains the chart values, not the individual observations required to reproduce the adjusted models.

Source file fingerprints

f1_thumnnail_prod_analysis.csv0aca34337bb17649cb674e6f7bc3914bbd06559b69eca7ebd29821a470be9f11
f1_thumnnail_prod_classifications.json6a45c78bc696011b1bd3f3c076a5d627b6d1e5988f53c6072d96bce3cd7d606a

I analysed 560 Formula 1 YouTube videos. Here’s what performs best.

1. The Formula 1 video formats that perform best

2. The Formula 1 drivers that attract the most clicks

3. The title promise that performs best