Thumbnail-as-Product Test
If most people only see the package, the package IS the message.
Asked about his Boomers documentary, Nicholas describes rejecting an obvious clickbait thumbnail — an old hand giving the middle finger. It would have performed well as a click-through. He killed it because most people who see a thumbnail never watch the thing — so the thumbnail itself functions as a standalone product. Putting an aggressive image in front of millions who won't watch the film does brand damage that outlives any CTR gain.
The framework: treat the thumbnail (or poster, hero image, social card, headline) as a product in its own right that ships to the entire audience, not just the click-through subset. Optimise it on two axes simultaneously — does it earn clicks from the target viewer, AND does it leave a fair impression on the larger group who only ever see the package?
This reframes thumbnail decisions from 'maximise CTR' to 'maximise CTR subject to brand integrity at the impression level'. It's a direct counter to the bait-and-switch incentives baked into algorithm-driven distribution.
- Most viewers of a thumbnail never watch the content — the thumbnail IS what they consume.
- A thumbnail that wins on CTR but alienates non-clickers is a brand-cost trade you may not see.
- The package ships to a much larger audience than the content does — design accordingly.
- Clickbait that contradicts the content damages trust on the next thumbnail you ship.
- The right test is two-axis: does it earn the click AND does it leave a fair impression on the rest?
- Mock up the most aggressive optionMake the version that most directly bait-pulls clicks — the middle finger, the all-caps shock claim, the stark binary. You need it as the boundary case to compare against.
- Imagine the impression-only viewerPicture the people who will see the thumbnail in their feed and never click. For each major audience segment (target viewer, adjacent viewer, the subject of the content), ask what impression the thumbnail leaves.Pro tipIf your content is about a specific group, imagine a member of that group seeing it without context. Their reaction is the brand-cost signal.
- Score each option on the two-axis testFor each candidate, score (1) likelihood of click from target viewer, (2) impression quality across the full feed audience. The aggressive option usually wins axis 1; the right answer is the option that's competitive on axis 1 and strong on axis 2.WarningDon't average the two axes into a single number — the trade-off should remain visible so you make it consciously.
- Test the headline / title under the same lensThe same logic applies to titles, social-post first lines, and email subject lines. Apply the test to all package surfaces, not just the visual.Pro tipIf you can't write a title that's honest and earns the click, the issue is usually the content positioning, not the headline.
- Track downstream effects, not just CTRAfter publishing, look at sentiment in comments, email replies, and unsubscribe rates — not just click-through. A high-CTR / high-unsubscribe combination is a clear signal the package is over-baiting.
The Boomers film had a mocked-up poster of an old hand giving the middle finger. It would have been an excellent thumbnail in pure CTR terms. Tom killed it after picturing a 70-year-old seeing it in their feed and refusing to watch the film he'd spent months making for them.
Many DOAC-adjacent thumbnails use shock-claim text ('THIS DESTROYS YOUR HEART') that wins clicks. The brand-cost shows up over time as a trust deficit when the content doesn't deliver on the threat-level promise.
Nicholas explains the moment in production where the team had a strong middle-finger poster mocked up. He sat with it, then asked: 'if you're 60, 70, 80 and you see that poster, are you going to watch the film?' Answer: no. He then generalised: most viewers of any thumbnail never watch the content, so the thumbnail is the product they actually consume. The framework is that realisation made into a repeatable test.