Rank viral social posts and send top picks overnight
A post in your niche pulled 2,158 likes and 5,834 saves, and you never saw it. Task: A post in your niche pulled 2,158 likes and 5,834 saves, and you never saw it. Native analytics only cover your own posts. Trending pages show you what has already peaked. By the time a format lands there, the people who caught it early have published their version. Here’s how to track viral posts across X, TikTok, and Instagram with a Grok Bot that works while you sleep. It runs three Apify Actors through the Apify plugin: Tweet Scraper, TikTok Scraper, and Instagram Scraper. Then it puts all three platforms’ engagement numbers on one scale, ranks the posts by save rate, and sends the top three to Telegram before you wake up. The full build is on video if you'd rather watch it: Why rank by saves, not likes Likes tell you a post was enjoyed. Saves tell you someone intends to come back and do the thing. If you're deciding what to make next, the second signal is worth more than the first. The numbers from the run behind this article make the point. The top post had 2,158 likes and 5,834 saves, a save rate of 2.7 (saves divided by likes). A product launch post in the same batch had far more likes and a save rate under 1. Ranked by likes, the launch wins. Ranked by save rate, the post people are actually planning to copy wins. What you'll need A Grok Bot account. Grok Bot is xAI's always-on agent app, and it keeps running after you close your laptop, which is the whole point here. An Apify account. Actors used here charge per result. The free plan is enough to test this. A Telegram account if you want the morning digest on your phone. Step 1: Install the Apify plugin in Grok Bot Open the marketplace in Grok Bot, install the Apify plugin, and sign in with your Apify account. You do this once. The plugin brings Apify MCP Server and a set of scraping skills with it. That's what lets the bot run any Actor in Apify Store by asking rather than by writing code, and it's why the rest of this tutorial is just prompts. Installation commands for other agents, including Claude Code, Cursor, and Codex, are at apify.com/plugin. Step 2: Test the scrapers and check the field names Before you build anything, give the bot one instruction: That last sentence decides whether the bot is worth anything. You’re checking what each platform calls the number you care about. Same human action, three different names, and one platform missing the best signal entirely. If the field names are wrong, the board ranks the wrong numbers. Instagram has no public save count, so this build substitutes comments divided by likes and labels that column separately, so nobody reads an Instagram row as a save. If shares matter more to you than comments, Instagram Reel Scraper can return a shares count for Reels. Step 3: Build the ranking in one prompt With the fields confirmed, let the bot build the board: Two filters in that prompt do most of the work. The 24-hour filter exists because tag pages come back newest first. Some of what you scrape is minutes old with nothing on it yet. In this run, the bot dropped four posts for being under a day old. Without the filter, brand-new posts with no engagement end up in the ranking. The second filter drops posts with no usable save rate. A post with no save count can't be ranked by save rate at all, and pretending otherwise puts it at the top or the bottom for no reason. This run dropped 18 more posts that way. Of 150 posts scraped, 128 made it onto the board. Step 4: Read the board Here's the top of the board from the run behind this article: All three top rows are X posts. That isn't the ranking favoring X: X returned the highest volume of posts in this run, so it had the most candidates near the top. TikTok and Instagram posts sit further down the same board, and a different set of search terms can change the order. Every row links back to the original post, so you can open it, see the format, and decide whether it's worth your version. Step 5: Make it run every morning A page you have to remember to open is still work. Set a weekday digest in Grok Bot, and the bot re-scrapes and re-ranks every morning on its own. Nothing else changes: same Actors, same prompt, same board, fresh rows. Step 6: Send the top three to Telegram Create a Telegram bot with BotFather and copy the token. In Grok Bot, install the Composio plugin and connect Telegram with that token.; tools: Grok Bot. Human role: The original post does not say how a person stayed involved.; limitations: Requires Grok Bot, Apify, and Telegram accounts. Apify Actors charge per result, though a free plan is available for testing. Instagram lacks public save counts, necessitating a proxy metric (comments/likes). The workflow depends on accurate field mapping; incorrect names lead to wrong rankings. Proposed advice: Always validate scraper output fields before automating the ranking logic..
Tools
Published on Agent Work Atlas:
Overview
A post in your niche pulled 2,158 likes and 5,834 saves, and you never saw it.
Workflow
A post in your niche pulled 2,158 likes and 5,834 saves, and you never saw it. Native analytics only cover your own posts. Trending pages show you what has already peaked. By the time a format lands there, the people who caught it early have published their version. Here’s how to track viral posts across X, TikTok, and Instagram with a Grok Bot that works while you sleep. It runs three Apify Actors through the Apify plugin: Tweet Scraper, TikTok Scraper, and Instagram Scraper. Then it puts all three platforms’ engagement numbers on one scale, ranks the posts by save rate, and sends the top three to Telegram before you wake up. The full build is on video if you'd rather watch it: Why rank by saves, not likes Likes tell you a post was enjoyed. Saves tell you someone intends to come back and do the thing. If you're deciding what to make next, the second signal is worth more than the first. The numbers from the run behind this article make the point. The top post had 2,158 likes and 5,834 saves, a save rate of 2.7 (saves divided by likes). A product launch post in the same batch had far more likes and a save rate under 1. Ranked by likes, the launch wins. Ranked by save rate, the post people are actually planning to copy wins. What you'll need A Grok Bot account. Grok Bot is xAI's always-on agent app, and it keeps running after you close your laptop, which is the whole point here. An Apify account. Actors used here charge per result. The free plan is enough to test this. A Telegram account if you want the morning digest on your phone. Step 1: Install the Apify plugin in Grok Bot Open the marketplace in Grok Bot, install the Apify plugin, and sign in with your Apify account. You do this once. The plugin brings Apify MCP Server and a set of scraping skills with it. That's what lets the bot run any Actor in Apify Store by asking rather than by writing code, and it's why the rest of this tutorial is just prompts. Installation commands for other agents, including Claude Code, Cursor, and Codex, are at apify.com/plugin. Step 2: Test the scrapers and check the field names Before you build anything, give the bot one instruction: That last sentence decides whether the bot is worth anything. You’re checking what each platform calls the number you care about. Same human action, three different names, and one platform missing the best signal entirely. If the field names are wrong, the board ranks the wrong numbers. Instagram has no public save count, so this build substitutes comments divided by likes and labels that column separately, so nobody reads an Instagram row as a save. If shares matter more to you than comments, Instagram Reel Scraper can return a shares count for Reels. Step 3: Build the ranking in one prompt With the fields confirmed, let the bot build the board: Two filters in that prompt do most of the work. The 24-hour filter exists because tag pages come back newest first. Some of what you scrape is minutes old with nothing on it yet. In this run, the bot dropped four posts for being under a day old. Without the filter, brand-new posts with no engagement end up in the ranking. The second filter drops posts with no usable save rate. A post with no save count can't be ranked by save rate at all, and pretending otherwise puts it at the top or the bottom for no reason. This run dropped 18 more posts that way. Of 150 posts scraped, 128 made it onto the board. Step 4: Read the board Here's the top of the board from the run behind this article: All three top rows are X posts. That isn't the ranking favoring X: X returned the highest volume of posts in this run, so it had the most candidates near the top. TikTok and Instagram posts sit further down the same board, and a different set of search terms can change the order. Every row links back to the original post, so you can open it, see the format, and decide whether it's worth your version. Step 5: Make it run every morning A page you have to remember to open is still work. Set a weekday digest in Grok Bot, and the bot re-scrapes and re-ranks every morning on its own. Nothing else changes: same Actors, same prompt, same board, fresh rows. Step 6: Send the top three to Telegram Create a Telegram bot with BotFather and copy the token. In Grok Bot, install the Composio plugin and connect Telegram with that token.
Reported result
The source reports a specific run where 150 posts were scraped and 128 were ranked. The top post had a save rate of 2.7 (5,834 saves, 2,158 likes). No broader business impact or conversion metrics are provided. Acceptance check: Verify that the bot correctly identifies the save rate calculation and filters out posts with zero engagement or age under 24 hours.
Human role
The original post does not say how a person stayed involved.
Limitations
Requires Grok Bot, Apify, and Telegram accounts. Apify Actors charge per result, though a free plan is available for testing. Instagram lacks public save counts, necessitating a proxy metric (comments/likes). The workflow depends on accurate field mapping; incorrect names lead to wrong rankings. Proposed advice: Always validate scraper output fields before automating the ranking logic.
Try it
Act as a social media analyst. I need to track viral content in [replaceable niche keywords] across X, TikTok, and Instagram. Use Grok Bot with the Apify plugin. First, run Tweet Scraper, TikTok Scraper, and Instagram Scraper on [replaceable search terms/URLs]. Output a table showing the raw engagement fields returned by each Actor to verify field names (e.g., bookmarkCount for X, collectCount for TikTok). Do not proceed until I confirm the fields. Once confirmed, normalize the data into a single table with columns: Platform, Likes, Saves, Save Rate. For Instagram, calculate Save Rate as Comments/Likes and label it 'Proxy Save Rate'. Filter out any posts created less than 24 hours ago and any posts with zero likes. Rank the remaining posts by Save Rate descending. Select the top 3 posts. Format the output as a Telegram message containing the rank, platform, save rate, and direct link to the post. Send this message to [replaceable Telegram Chat ID] only after I approve the draft. If any scraper fails or returns empty data, report the error and stop. Do not invent data.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.