A 30-day content test works when you change one thing per week and judge every post against your own median. Run four tests in order: the first line, the format, the posting hour and the call to action. Do not waste a week on hashtag counts.
TLDR: Spend days 1 to 3 setting a baseline, then run one test a week. Compare each post to your own median, keep what wins in at least 3 of 5 posts, and test again. Skip variables that barely move results.
Days 1 to 3: set the baseline
Open your last 20 posts and write down reach and engagement rate for each. Find the median, the middle value, not the average. One viral post wrecks an average.
Then note the format and the posting hour of each. You are looking for your normal, because every test later is "better or worse than normal".
For context, Mydrop's data put the median Instagram engagement rate at 3.1% and the median TikTok rate at 3.0%. Your own median matters more than either number.
The four tests
| Week | Variable | Keep the same | Posts | Win |
|---|---|---|---|---|
| 1 | First line of the caption (the hook) | Format, time, topic | 5 | 3 of 5 above your median |
| 2 | Format (carousel vs single image vs video) | Hook style, time | 5 | 3 of 5 above your median |
| 3 | Posting hour | Format, hook style | 5 | 3 of 5 above your median |
| 4 | Call to action (save, comment, send, click) | Format, hook, time | 5 | 3 of 5 above your median |
The 3-of-5 rule is a rule of thumb, not a statistical test. With five posts, results are directional. That is fine: you are looking for big differences, and you will confirm a winner by repeating it.
Week 1: test the hook
The first line decides whether anyone expands the caption or stops scrolling. Write three styles and rotate them: a number ("3 mistakes that..."), a direct claim, and a question to the reader's situation. Keep the topic and format identical so the hook is the only change.
Week 2: test the format
If you post mostly photos, try carousels. In Mydrop's data, carousels ran 10% above each account's typical post, while photos and videos sat at the median. That is a clue about where to start, not a guarantee for your audience.
Week 3: test the posting hour
Move the same type of post to a different hour in your audience's time. The one clear result in Mydrop's data was the bad slot: posts published between midnight and 6 am local time ran 17% below the median across platforms. Try morning, lunch and evening, and avoid the night.
Week 4: test the call to action
Change only the ask at the end: "save this", "send this to a friend", "tell me below", or a link. In the same dataset, a link in the caption cost 7% on average, so check whether a "send this" or "save this" ask beats it for you.
How to read the results
After each week, put the five posts next to your baseline median. Ignore the single best and single worst post. Look at the middle three.
If three of five beat your median, adopt the change for the next week and keep testing the next variable. If the results are mixed, keep the old way. A test that shows nothing is still a result: it tells you not to spend effort on that variable.
What not to test
Some variables are not worth a week. Hashtag count moved median engagement by 5% at most in either direction in our data, and the weekday made almost no difference, with every day within 2% of the median. Spend your tests on the hook, format, hour and ask.
Run it without a spreadsheet
Plan the 20 posts in Mydrop's calendar with the variable noted in each post, then read results per post in Performances, which compares each post to that account's own median. The method is in our 30-day Instagram reach recovery plan.
Next step: do days 1 to 3 today. Write your median, pick the week 1 hook styles and schedule five posts. To keep every test in one calendar, start free with Mydrop.
Google review
Trustpilot review
Product Hunt comment
