The 30-Minute Monthly Content Audit

Checking your analytics every morning is the least informative way to read them. A day holds one or two posts, a single post sits inside a wide band of normal variance, and the story you tell yourself about why it did well is almost always invented after the fact.

A month is the smallest unit that tells the truth. Here is the direct answer if you want to know how to analyze your own TikTok analytics without drowning in them: once a month, put every post in one sheet, label each one by format, and run four comparisons. This month’s median against last month’s. Format against format. Your top quarter against your bottom quarter, on the opening seconds only. What you planned against what you actually shipped. Then leave with exactly three decisions: one format to keep, one to cut, and one thing to test. The whole pass takes about thirty minutes.

Why per-video post-mortems mislead

The instinct after a flop is to open that one video and find the reason. You will always find one. The hook was slow, the lighting was flat, the caption was long. The trouble is that the video that did three times better last week had the same slow hook and the same lighting, and you did not look at it, because it did not hurt.

One video has dozens of variables and no control group. Any explanation you pick is a guess, and guesses made in a bad mood tend to change the wrong thing. This is the same trap covered in finding the repeatable part of a video that blew up: a single result, good or bad, cannot tell you which of its parts mattered.

A group can. If you posted five tutorials and five opinion videos this month, the gap between those two groups survives the noise that ruins any single comparison. That is why the audit is organized around formats rather than videos. Per-format reviews average out the week you posted, the sound you used and the one lucky share. Per-video reviews are mostly made of those things.

Five minutes: build the sheet

Open a spreadsheet and give every post from the last calendar month one row. Leave out anything published in the final seven days, because those videos are still being tested and their numbers are not finished.

The columns that matter:

• Platform. TikTok and Instagram numbers are never compared raw. They sit in the same sheet so you can see the month, not so you can rank one against the other.
• Format. Your own plain label for the shape of the video: tutorial, list, story, opinion to camera, day in the life, transformation, reaction. Keep the vocabulary to six or seven words and reuse it every month, or the comparison breaks.
• Hook shape. One or two words: question, bold claim, visual reveal, disqualifier.
• Length in seconds.
• Views, then average watch time and, where the platform shows it, the share of viewers who watched to the end.
• Shares, saves and new follows from that post.

Now add the three columns that do the real work, because raw counts measure distribution rather than the video. Divide shares, saves and follows by views and multiply by 1,000. Every comparison below uses those ratios, and the reason is laid out in the metrics worth ignoring: a number without a denominator describes your reach, not your content.

Read it like a stranger would

Before you compare anything, hide the caption column and sort by saves per 1,000 views.

This step sounds cosmetic and is the most important one in the audit. You remember which video took four hours to edit and which one you filmed in the car. You remember the one you were proud of. A stranger reading your sheet remembers none of that, and a stranger is exactly who your videos are shown to. Reading the rows as labels and ratios, stripped of the memory of making them, is the difference between an audit and a mood.

Comparison one: this month’s median against last month’s

Use the median, never the average. One video that ran forty times your usual number will drag an average upward and hide a month where most posts slipped. The median is the video in the middle, and it does not care about outliers in either direction.

Compare this month’s median views, median watch-through and median saves per 1,000 views with last month’s. Then count your hit rate: how many of this month’s posts beat last month’s median. If you posted sixteen times and nine beat the old median, your baseline moved up. If four did, it moved down, even if one of those four went viral.

That single number is more honest than your best video. Viral is a multiple of your own median rather than a round figure, which is the whole argument in how many views counts as viral, and a month with one breakout and a falling hit rate is a month where the underlying work got worse.

Comparison two: format against format

Group the rows by your format label and take the median of each ratio per group. You need at least three posts in a format to say anything about it. With fewer, write “not enough” in the cell and move on rather than drawing a conclusion from two data points.

What you are looking for is a format that is quietly outperforming its share of your effort, and one that is quietly eating it. It is common to find that the format you post most often is not your best, simply because it is the easiest to make.

Read length alongside format, not separately. A 45-second tutorial and a 12-second reaction should not be held to the same watch-through, and your ideal length is a format question rather than a platform rule. Where a format underperforms, check whether the videos in it ran long for their shape before you blame the shape itself.

Comparison three: top quarter against bottom quarter, on the opening

Sort by watch-through and take the top four rows and the bottom four. Now ignore everything except the first three seconds of each.

Pull the retention curves and look for where the two groups separate. If the bottom quarter lost most of its viewers before second three and the top quarter did not, your problem is the opening, and the hook shape column will usually show you which move is failing. If both groups held about the same at second three and split apart later, the hook was fine and the middle is where the work is.

This matters because the fix for each is different, and creators routinely rewrite hooks that were working. Total views will not tell you which case you are in, as covered in how to tell if your hook worked. The retention curve will.

Comparison four: what you planned against what you shipped

Open last month’s plan next to the sheet. Count how many posts were the format you intended, and how many were substitutions made on the day because the planned video was too hard to film.

Drift is information. If the planned carousel or the long tutorial kept getting replaced by a quick opinion video, one of two things is true: the planned format is too expensive for your real schedule, or the substitute is what you actually enjoy making. Either answer should change next month’s plan. A calendar built around pillars and formats rather than specific captions survives this kind of pressure far better, which is the case made in the content calendar that survives real life.

Check cadence here too. If you posted less than you planned and your hit rate rose, that is worth knowing before you force the volume back up. How often to post is a function of hit rate, not a quota.

The output: one keep, one cut, one test

An audit that ends in a list of twelve observations changes nothing. End it with three lines.

Here is what that looks like on a month of sixteen posts. Illustrative figures.

• Keep: tutorials. Five posts, median 14 saves per 1,000 views against a sheet median of 6, and four of the five beat last month’s median views.
• Cut: trend audio remakes. Three posts, the lowest watch-through in the sheet, and none of them brought a single follow above the sheet median.
• Test: the opinion videos held viewers well past second three but lost them around second twenty. Next month, three opinion videos with a re-hook placed just before that point, and nothing else changed.

Notice what the test does not do. It does not change the format, the hook and the length at once. If three things change and the numbers move, you learn nothing about which one did it.

Four mistakes that ruin the audit

• Auditing posts that are still being tested. Anything from the last week is unfinished. Leave it for next month.
• Comparing TikTok and Instagram rows directly. Run the comparisons within each platform, then look at both summaries side by side.
• Letting the best video define the month. It goes in the sheet like every other row. Hit rate and medians decide the verdict.
• Changing the format labels every month. If “tutorial” becomes “how-to” next month, you have lost the ability to compare. Pick the vocabulary once and keep it.

Where Blossom fits

We build Blossom, so treat this section as interested and the behavior as checkable. The slow part of the audit is the labeling: deciding what format a video is, what its opening move was, and where viewers left, consistently, across sixteen posts and two platforms.

Paste a video into Blossom and it comes back with a named format, a hook scored 1 to 10 with the reasoning behind the score, and a timeline marking the moments attention builds and drops. Because the format names come from one shared vocabulary, the grouping in comparison two stays consistent from month to month without you policing your own labels. The library then answers the question your own sheet cannot: whether a format that worked for you is also holding up across other analyzed videos in your category, or whether your five tutorials were a small sample having a good month.

None of that replaces your native analytics, and no tool can tell you a video will go viral. The numbers live in your dashboard. What changes is how quickly they become three decisions.

The half-hour version

List last month’s posts. Label each one by format and hook shape. Divide saves, shares and follows by views. Compare medians and hit rate against last month, compare formats against each other, compare the opening seconds of your best and worst quarters, and hold the result against your plan. Write one keep, one cut and one test, and change nothing else.

Do it on the same day every month and the sheet becomes the most useful document you own. Ready to skip the labeling? Start a Blossom trial and paste in last month’s posts, or read the FAQ first.