What I found from 765 posts
I pointed AI at six creators in my niche and had it pull every post from the last 90 days. 765 posts total. Then it downloaded every video, turned the audio into text, and ran the numbers on all of them. 106 breakout posts. 99 outliers. 522 average posts. Here is what separates the ones that blow up from the ones that die.
Finding 1: Negative hooks get double the views
Posts with negative framing averaged 34,335 views. Positive framing averaged 16,846. That's a 2x difference just from how you open.
The words that showed up in breakout hooks were killed, dead, scam, replacing, stop, ruining, cooked, and warning. The words that showed up in average posts were powerful, best, easy, amazing, incredible, and game changer.
Lead with the pain, the cost, or something being destroyed. Every time you're tempted to say something is "amazing" or "powerful," flip it. Say what it replaced. Say what it killed. Say what it cost you before.
Finding 2: Breakouts are a third shorter than average posts
Average posts run about 90 seconds. Breakout posts hit a median of 54 seconds. That's 33% shorter.
The word count tells the same story. Breakout median is 128 words. Average posts run much longer. Everyone is trying to pack more information into their videos but the data says cut a third of it and you land in breakout territory.
Finding 3: Posting 3 times a day is killing your reach
This is the one that surprised me most. Every guru says post more. The data says the opposite.
Spacing your posts 12 to 48 hours apart massively outperforms posting multiple times per day. One creator in the dataset saw 42 times more views just by spacing posts out. The sweet spot is once per day, ideally once every 24 to 36 hours.
8 more patterns from the outlier data
Layering two hook types doubles your odds
Posts that combine two or more hook types averaged 49,099 views. Posts with a single hook type averaged 25,839. That's a 90% lift. The best combos were controversy plus a named enemy, how-to plus a specific stat, and bold claim plus a comparison.
Urgency language nearly triples breakout rates
Posts with urgency phrases hit a 35% breakout rate versus 13% without. Average views jumped from 19,970 to 58,613. Phrases like "nobody is talking about this," "right now," "most people don't know," and "if you're not doing this" were the most common in outliers.
93% of breakouts use zero emojis
Average posts use emojis at three times the rate of breakouts (23% versus 7%). Emojis signal average content to the algorithm and the audience. Drop them entirely.
Sentence variety is mandatory
Back to back sentences with the same structure appear in only 6% of sentence pairs across all breakouts. If two sentences in a row feel like bullet points from a slide deck, the algorithm treats them like one. Vary every sentence.
Listicle format outperforms unstructured walkthroughs
13% of breakouts use a listicle structure like "3 systems" or "5 steps" versus only 3% of average posts. Listicle breakouts averaged 115,028 views compared to 65,548 for non-listicle. When the content supports it, number your points.
Exactly 5 hashtags is the sweet spot
Posts with exactly 5 hashtags averaged 185,420 views. Posts with 3 to 4 averaged 149,000. Posts with 1 to 2 averaged 90,000. More is not better and fewer leaves reach on the table.
Tagging other creators hurts your reach
Posts without @mentions had a 15% breakout rate and averaged 23,050 views. Posts with tags dropped to 11% breakout rate and 15,252 average views. Stop tagging people thinking it gets you exposure.
Post between 5pm and 9pm Mountain time
Posts in this window averaged 255,260 views. That's 2.2 times the next best window. Most creators post in the morning but the data shows evening dramatically outperforms. The best individual hours were 5pm Mountain (389,000 avg) and 6pm Mountain (278,000 avg).
How to run this yourself with Apify
Apify is a tool that pulls public data from social platforms. You give it creator profile URLs and it grabs every post with all the metrics. The total cost for my run across 765 posts was $1.87.
Step 1: Get your Apify token
Create a free account at apify.com. Go to your settings and copy your API token.
Step 2: Pull the data
Open Claude Code and paste this prompt. Replace the example URLs with creators in your niche.
I want to analyze the short form content strategy of these creators. Pull every post from the last 90 days for each one using the Apify Instagram Scraper. Here are the profiles:
1. [creator profile URL]
2. [creator profile URL]
3. [creator profile URL]
4. [creator profile URL]
5. [creator profile URL]
6. [creator profile URL]
My Apify API token is: [paste your token here]
For each post, I need: post URL, post date, caption text, view count, like count, comment count, and whether it's a video or image. Save everything to a JSON file called creator-raw-data.json.Step 3: Turn videos into text and run the numbers
Once the data is pulled, paste this into Claude Code.
Download every video from creator-raw-data.json. Turn each video into text. Then analyze the full dataset and tell me:
1. What hook patterns get the most views? Compare negative vs positive framing with exact multipliers.
2. What words appear in high performing hooks that don't appear in average ones?
3. What's the average duration and word count of top posts vs average ones?
4. Does posting frequency affect reach? Compare posts spaced 12-48 hours apart vs posts within 12 hours of each other.
5. What caption patterns correlate with higher views? Length, emoji usage, hashtag count, tagging.
6. What posting times and days perform best?
Give me specific numbers for every finding. Save the report to creator-analysis-report.md.Claude Code handles the downloads, transcription, and analysis on its own. The whole thing costs a couple dollars and runs without you touching it.
What this costs
Apify charges based on usage. My run across 765 posts from six creators cost $1.87. Most runs for 3 to 6 creators land between one and five dollars.
