Skip to content

Hitfactor Labs

The YouTube space runs on guru advice, myths, and anecdotes. Most “data-backed” advice is poorly designed and quietly misleading, which costs you time and budget. Worse still: the gurus telling you, “trust me, bro, I have a billion views.”

Hitfactor Labs exists to test conventional YouTube wisdom and broscience against actual data and report what survives. We believe YouTube is the single greatest marketing and attention asset at your disposal and you can’t waste that on unsubstantiated advice. Better data leads to better decisions.

Study 006 Does the 'I Asked AI to Make Me Money' Outlier Format Actually Work? A viral post says steal this format and you'll go viral. We found every version we could — 97 videos. The typical one got about 1.4x its channel's usual views, 41% did worse than the channel's own average, and on channels with no audience it got 1–87 views. A better slot machine, not a reliable formula. 97 videos · 85 channels Read the study Study 005 Does a More Colorful or Higher-Contrast Thumbnail Get More Views? We tested "make the thumbnail pop" on 1.5M videos. Within a channel, more color does nothing for views, and more contrast helps only trivially. Neither is big enough to act on. 1.53M videos · 16,681 channels Read the study Study 003 Does Posting More Get You More Views? We tested advice you hear constantly: just post more. Within a channel, posting more goes with proportionally more total reach (roughly double the uploads, double the total), but through volume, not a per-video lift: each video earns about the same. Correlation, direction unknown. 1.24M videos · 16,282 channels Read the study Study 004 Does Where the Face Sits on the Thumbnail Get More Views? We tested "put the face on the left, not center" on the videos that have one face. Within a channel, left vs right moves views by no more than a few percent, and our best estimate is none. 17K videos · 381 channels Read the study Study 002 Do YouTube Thumbnails with Larger Faces Get More Views? We tested "fill the frame with a face" on the videos that already have one. Within a channel, a bigger face changes views by essentially nothing: confidently near zero. 68K videos · 2,192 channels Read the study Study 001 Do YouTube Thumbnails with Faces Get More Views? We analyzed 1.25M videos to test "put a face on it." Within a channel, a face on the thumbnail is associated with about 6% fewer views, the opposite of the advice. 1.25M videos · 16,737 channels Read the study

Plan first, data second. We commit the analysis and tag it publicly before we query a single outcome. The timestamp is the audit trail.

A holdout we never touch. A fixed slice of channels is set aside before exploration. The headline number runs on it once, same code, after the plan is locked.

We resample channels, not videos. Videos from one channel aren’t independent, so every confidence interval treats the channel as the unit.

Magnitudes, not significance. Past a million videos, anything non-zero clears the bar. So we report effect sizes, intervals, and a practical threshold set in advance.

The data ships with the study. Binned tables behind every chart, plus a de-identified row dataset. Replicate it yourself.


Looking for product documentation instead? See the Docs.