A hashtag experiment should help answer a small question: does a particular set of relevant tags appear useful for this content and audience? It should not pretend to reverse-engineer TikTok from two posts.

The practical obstacle is that videos differ. A better hook, a topical event, or a more useful demonstration can affect performance far more than a tag change. Publishing different videos with different hashtags is not a clean randomized test, even when a spreadsheet labels the groups A and B.

This worksheet offers a modest alternative: matched content groups, fixed observation windows, and a decision rule written before the results arrive. All numerical examples are hypothetical.

Start with a question narrow enough to test

Avoid “Which hashtags make videos go viral?” It is too broad and implies a guarantee no tag list can provide.

Try this instead:

“For short beginner sewing demonstrations, does adding a specific problem tag produce more relevant discovery than using only broad sewing tags?”

Define the intended viewer. A beginner searching for help threading a machine has a different need from an experienced pattern designer. A tag is useful when it describes the actual content and audience, not merely because it has a large historical view count.

Build two relevant tag sets

Create a baseline set and one challenger set. Change as little as practical.

Set Illustrative tag roles What changes
A Sewing topic, beginner community, tutorial intent Baseline
B Sewing topic, beginner community, specific problem One intent tag

Use real tags only after checking current search results inside the platform. A tag that sounds appropriate may have shifted into unrelated content. Check recent posts, language, and audience context rather than relying on its name alone.

There is no universal magic number of tags. Keep the set concise enough that every tag has a clear reason to be present. Do not add irrelevant trending tags to manufacture a test condition.

Plan matched content rather than duplicate uploads

Prepare several genuinely different videos with similar difficulty, length, format, and audience intent. Pair comparable topics, such as two beginner machine-maintenance tips.

Rotate the tag sets across your normal publishing schedule so one condition is not always assigned to your strongest slot. Avoid giving all the best ideas to the challenger because you hope it will win.

Do not repeatedly upload identical videos just to change tags. Duplicate content can introduce audience fatigue and other platform effects, and it still does not create a clean test.

Record what changed beyond hashtags. If a video suddenly uses a new presenter or stronger opening, that result deserves a note rather than being attributed entirely to tags.

Copy this observation template

Create one row per post using these fields: