A founder posted a four-minute video on X in late October 2026 with a big promise. He runs thousands of social media accounts, and he will show you how to rebuild his system in 30 days. He says he put out more than 5,200 posts on TikTok and Instagram in the last 30 days, and that companies using his method pay 22 cents per thousand views. In the trade, systems like this are called TikTok account farms, even when real people run every account. I read his playbook the way I read a lab protocol, line by line, with a PhD in machine learning and neuroscience and years of format testing for clients behind me. Half of it is a real method, and the other half, the planted questions and the unlabelled accounts, works as a disguise. If you are a founder tempted by a thousand accounts, the question is which half you are buying.
TL;DR
- The account-farm playbook has a real core: copy formats that already work, test them at volume, and separate the person who creates from the people who publish.
- Its test mixes two rules that can disagree, outliers and consistency, and two new accounts measure the accounts as much as the format.
- The cheapest views come from one vendor case study, and the best-documented app growth of the past two years ran on paid creators posting from their own accounts.
- TikTok’s own rules target networks of similar accounts that push people to a business, and the FTC expects paid endorsements to be disclosed.
- A thousand rented accounts find what the platform already wants. A founder gives it something new to want, and that part compounds.
What the 1,000-account playbook says
Miles Feldstein co-founded OthersideAI (BusinessWire), the company behind HyperWrite (Marketing AI Institute), and co-founded HumanPost. His video lays out four moves. He says it is not sponsored, that the tips work without his service, and that he is biased because he sells part of it.
First, copy. Search TikTok for photo slideshows in your niche, keep the ones above 25,000 views from the last three months, and make 40 posts in that format, on at least two accounts, 20 each, before you call the test. When posts consistently break 2,000 views, you have a winner. Second, produce at scale: reaction clips with the hook written on top and one of five product demos, or slideshows from a small tool you can build with Claude or Codex, reusing the same images with new text. Third, convert harder: put the App Store screenshot inside the slideshow, have other people comment “what app is this?” under your best videos, and answer with a photo of the product. Fourth, scale with people: one real person per account, with a profile picture, bio and username he says should be optimized to your brand, accounts warmed up, never several phones on the same household Wi-Fi.
HumanPost sells that fourth step. According to its website, people across the US create TikTok and Instagram accounts dedicated to one brand, warm them up on niche keywords and post the brand’s approved content by hand from their own phones (HumanPost). The sample accounts on the site use first-name handles and personal-style bios. The same day as the video, the company published an official MCP integration on GitHub that lets AI agents order that work (HumanPost MCP). Ok, I know, it sounds like science fiction: your AI agent now manages a crowd of humans with phones.
Can one person run a thousand accounts?
The best line of the video is the oldest idea in media: never have the same person create the content and post it. A newspaper has a newsroom and delivery trucks. A founder who writes, films, edits, posts and replies runs both, and the trucks always eat the newsroom.
So yes, one person can run the brain of a thousand accounts: the formats, the slides, the tests, the reading of the numbers. The arms, a thousand phones and the people to hold them, are exactly what the market now rents. The skill that stays with you is knowing which formats deserve the arms.
Is 40 posts on two accounts a real test?
The instinct is right. A single video tells you almost nothing, because a video is a throw and a format is the die. Volume is how you read the die. I built my whole method on that sentence (board game theory and TikTok).
The protocol has two problems, and both are classic statistics.
The first is the winning rule. “Watch for outliers” and the bar of posts that consistently break 2,000 views are two rules that can disagree. An outlier is one throw far from the others. Consistency is the median and the spread. Views on short video are heavy-tailed: in one large study of YouTube (2007 data), the top 10% of videos drew nearly 80% of all views (Cha et al.). If you crown formats on outliers, you crown luck. Judge each format on its median and its floor once the data is mature, and treat the outlier as the hypothesis for your next variant.
The second is the baseline. 2,000 views means nothing on its own. On a fresh account it can be a great result, on an established one a failure. When you test on two new accounts, the accounts themselves differ: one gets pushed, the other stalls, for reasons that have nothing to do with your format. Read the format inside each account first, against that account’s own baseline, and only then compare accounts. Otherwise your test measures the accounts.
Voilà. With these two fixes, the 40-post test becomes what it should be: a cheap experiment that tells you which dice to keep.
Do TikTok account farms work, and for whom?
The method has self-reported wins for some apps, and the best-documented story comes from a different model: paid creators. Cal AI, the calorie scanner built by teenagers, paid about 250 influencers on monthly retainers, according to its founder (Latka). MyFitnessPal bought the app in a deal that closed in December 2025, citing more than 15 million downloads and over $30 million in annual revenue (TechCrunch). Those were real creators on their own accounts.
Slideshow farms have their own success stories, mostly self-reported. Glam Up, an AI makeup app, tested before-and-after slideshows on six TikTok accounts its founders ran by hand, and Yuma Ueno reports $150,000 of monthly recurring revenue for it, without a source (Yuma Ueno). Other self-reported cases show the gap between views and money. Larry, an AI agent that makes TikTok slideshows, drove 500,000 views and $714 of monthly recurring revenue in five days for its creator’s app, by his own account (Stack Junkie). Roy Lee, Cluely’s co-founder and CEO, reported that one of its UGC videos passed 49 million views and brought fewer than 100 downloads (Cluely).
Look at the products that win, such as a calorie scanner or an AI makeup app. Their value shows on screen in three seconds: you see the photo, you see the result, you get it. A faceless slideshow account can sell that, because the product is the proof.
Now take a B2B AI tool sold through demos at a few hundred dollars a seat. The buyer does not install it from a slideshow. They ask who built it, they read the founder, they check whether the company will exist next year. For that buyer, a hundred anonymous accounts add reach and subtract trust. The method works where the product carries the trust. Where the founder carries it, a network can only amplify a voice that already exists.
What does a view really cost?
The 22 cents per thousand views comes from a case study on HumanPost’s website: a travel app, 19.2 million views from 827 posts on 22 accounts, at $5 a post. HumanPost says its Scaling customers averaged 99 cents over the last 90 days (HumanPost). Both are vendor figures. For comparison, paid TikTok in-feed ads cost roughly $3 to $15 per thousand impressions (Business Model Analyst), and I use $27 per thousand as the paid benchmark for a US AI and SaaS audience.
Now apply the video’s own bar. HumanPost’s Scaling plan costs $1,125 a month, billed quarterly, for 250 posts, which works out to $4.50 a post. If a winning post does 2,000 views, a thousand views cost $2.25, ten times the headline. The headline needs posts that average about 20,000 views, and to be fair to the vendor, the travel-app case study averaged about 23,200.
Gaps like that do not last. Either the platform closes them, or everyone piles into the same formats until they stop working. In a report published in June 2026, Kapwing set up a brand-new TikTok account and classified 294 of the first 500 videos in its For You feed, 59%, as AI slop (Kapwing). In my reading, the same thing happens to every cheap format: it gets copied until it is noise.
So yes, the cost per view can be tiny, and we don’t care, because the number that pays the bills is the cost per customer who stays. The video has a conversion step, and its only business number is the cost per thousand views. Installs, signups and revenue are the missing link of the playbook, because between cheap views and a company that grows sits trust.
Are TikTok account farms allowed?
Three parts of the playbook deserve a careful read before you copy them.
The planted comments. Having other people ask “what app is this?” under your videos stages a customer question. Viewers believe it comes from a stranger. Under the FTC’s Endorsement Guides, a message people read as an independent opinion is an endorsement, and a paid or other material connection has to be disclosed clearly (FTC). TikTok also bans the trade of services that artificially increase engagement (TikTok). I wrote about some of the largest influence operations ever exposed, and most of them never reached a real community (the comment section essay). You can get that question for free. Show the result, keep the tool’s name off the slides, and give real viewers a reason to ask it.
The accounts kept apart. One person per account, warmed up, never two phones on the same Wi-Fi. He gives account health as the reason, and the effect is that nothing technical ties the accounts together. TikTok allows several accounts for “authentic creative expression” and forbids using them for deceptive purposes, and its spam rules list networks of accounts that post similar content to lead others to other accounts, websites or businesses (TikTok). Saying on every account who it works for answers the disclosure rules. The spam rules on networks still apply to a labelled network, so read them with a lawyer before you scale.
The missing label. TikTok requires all commercial content to be disclosed, promotion of your own brand included, and applies the disclosure setting itself when it finds undisclosed marketing (TikTok). In a separate investigation, Indicator found mass-produced “organic” campaigns with more than 2 billion views and no paid label, and sent more than 60 accounts to TikTok and the FTC (Indicator). In Entrepreneur’s write-up of a Wall Street Journal report, advertising attorney Rob Freund says enforcing ad disclosure requirements on social media “doesn’t seem” to be a top priority right now (Entrepreneur). Weak enforcement today is a bet on tomorrow, and the stake is your brand.
The video also promises accounts that will not be shadow banned. No vendor can promise distribution. I build my work on a different promise: I guarantee the system and the volume of work, and the data judges the views.
What is left after six months?
Here is the test I apply to every growth tactic. Stop paying, wait six months, and look at what you own.
With a rented network, you own the formats you learned and a spreadsheet of results. Unless your contract says otherwise, the accounts, their followers and their history live with the people who run them, or with nobody once they get restricted. Each account is a mimetic die: it throws what the platform already distributes. Only the founder can produce an idea the market has yet to see, and the founder is the one person missing from the network.
I ran the opposite experiment with a 57-year-old pizza maker who believed hashtags were the secret. One face, more than 400 videos edited from January 2024 to April 2026. He went from 2,000 to 1 million followers in two years, 1.6 million today, and he now has a two-hour show on national TV and franchise requests from five countries. Volume found his formats, and his face made them his.
The objection I hear most
“My competitors already run fifty accounts, I cannot wait for my face to compound.” D’accord. Then run the network the way a scientist runs a lab, in the open. Test formats at volume with your own images. Put your brand in the name and the bio of every account, or pay creators who post from their own accounts with the platform’s label. Cut every planted comment. And put the founder in at least one of the formats, so the winners carry a name the market can remember.
How to use the account-farm method without renting your reputation
- Check the fit. A network sells products whose value shows on screen in three seconds, with a short path to install. If you sell through demos, start with the founder.
- Scout formats like a scientist. In TikTok’s Photo results, keep posts from the last 90 days far above your niche’s usual post, five to ten references per format, and add one original format built on you.
- Produce 40 posts per format with your own images, one variable at a time, the hook first.
- Test on two accounts, 20 posts each. Read on day 7 and judge each format’s median against each account’s baseline.
- Convert with real proof: the App Store slide, the demo trick, a pinned comment from the account, the product in the bio.
- Scale with people whose accounts say who they work for, and judge the network every month on the cost per customer who stays.
So here is my question for you: if you had a thousand accounts tomorrow, what would they say that only you could say?
A thousand accounts can tell you which dice the platform already loves, and only a founder can carve a die nobody else can throw.
FAQ
What is a TikTok account farm?
A network of TikTok accounts run by one company to publish its content at volume, often photo slideshows or short clips about an app. Some are run by hired people on their own phones, others by software running hundreds of phones, the phone farms. TikTok’s rules prohibit networks of accounts that post similar content to lead others to other accounts, websites or businesses.
Do TikTok account farms actually work?
They buy exposure for consumer apps whose value shows on screen in seconds. The revenue effect is mostly unproven: the public cases report views or self-reported revenue, and the best-documented app growth of the period ran on paid creators who post from their own accounts.
Are TikTok account farms legal, or is it astroturfing?
TikTok allows several accounts for “authentic creative expression” and forbids them for deceptive purposes. Fake personas and undisclosed paid posts break its rules, and as I read them, planted comments fall under its ban on artificial engagement. Labels answer TikTok’s disclosure rule and the FTC, and TikTok’s spam rules on networks of similar accounts and high-volume commercial content apply to a labelled network too. In the US, undisclosed paid endorsements fall under the FTC’s rules. Check with a lawyer before you pay people to post.
Should a startup use UGC or founder-led content?
Run the six-month test: stop paying and look at what you own. Rented accounts leave formats and a spreadsheet. A founder’s face leaves an audience that knows a name, and that part compounds.
Dr. Louis Korczowski is a short-form content strategist for SaaS and AI founders, with a PhD in machine learning and neuroscience. The full TikTok slideshow app marketing playbook, with the platform rules and open-source AI skills, is on STATUR.