Post-mortem
Why it died
Not one cause. Six, arriving close enough together that the model had no time to adapt to any of them. Each section below is linked from the relevant entries in the graveyard.
1. The link revenue collapsed
Of the three revenue streams that funded post2host, the one that paid best was selling links from an aged, active forum. When search engines moved decisively against paid links, that income did not shrink. It ended.
This is the least discussed cause and arguably the most important, because it hit the funding side rather than the demand side. A host could absorb fewer signups. It could not absorb losing a third to a half of its revenue while its server bill stayed exactly the same.
2. Paid hosting fell below the price of a coffee
The post2host bargain was simple: your time instead of your money. That trade only makes sense while hosting is expensive enough to be worth an hour of your month. Once shared hosting from the large budget hosts dropped to a couple of pounds a month, the calculation flipped for almost everyone. Ten posts a month is not obviously cheaper than two pounds.
Worse, the free hosts that asked for nothing at all, 000webhost and ByetHost among them, were competing for the same users while demanding no participation whatsoever. Post2host was squeezed from both sides at once.
3. The forum stopped being where people were
The model did not just use a forum, it was a forum. Everything depended on people wanting to be in that room. Facebook, Reddit, Twitter and later Discord absorbed exactly the casual discussion traffic that quotas were designed to generate, and they did it without requiring anybody to earn anything.
You can watch this happen in the archive. x10Hosting's community survived, but its boards were merged and hidden as usage fell, and it settled into being a support venue rather than a destination. Frihost's admins and moderators departed one by one after 2012. The forums did not usually die dramatically. They thinned.
4. The licensing shock of 2019
On 27 June 2019 cPanel, by then owned by Oakley Capital and later folded into WebPros alongside Plesk and WHMCS, replaced its flat server licence with account-based pricing. A licence that had cost roughly forty-five dollars a month for unlimited accounts could cost several hundred for a server carrying around a thousand.
For an ordinary commercial host this was a painful cost increase to pass on to customers. For a free host it was categorical. The entire proposition of free hosting is a very large number of accounts each generating no revenue, which is precisely the shape the new pricing punished hardest. Dormant accounts were billable too.
By 2019 the post2host scene was already well into decline, so this was not the thing that killed it. It is the thing that made rebuilding it impossible, which matters more. It closed the door behind the era.
The most instructive response in this whole archive belongs to iFastNet, which runs ByetHost and InfinityFree: when the panel became the cost, they shipped their own panel. VistaPanel is cPanel-shaped and costs them nothing per account. That single decision is why their free hosting outlived nearly everything here.
5. The abuse burden never stopped growing
Free hosting is, from an attacker's point of view, free infrastructure. There is no card to trace, nothing to lose when an account is banned, and no cost to creating another. Security researchers consistently find that a large share of phishing sites sit on free hosting and free tools, because when a domain is blocklisted the attacker simply starts again at no cost.
For a small host run in someone's evenings, this is relentless: phishing kits, spam scripts, miners, mass mailers, and a mailbox full of abuse reports. It also poisons the well. A free host that sends spam ends up on blocklists, at which point the legitimate users cannot send a password reset email, and the service quietly becomes worthless.
The related lesson from the wider industry is Freenom, which gave away domains and by 2021 accounted for five of the ten most abused top-level domains anywhere. Years of abuse reports did not stop it. A trademark lawsuit did.
6. Trust, spent all at once
In October 2015 it emerged that 000webhost had lost around 13.5 million user records, with the passwords stored in plain text, in a breach that had happened roughly five months before anyone was told. It was attributed to an exploit in an outdated PHP version.
The damage was not confined to one company. It became, and remains, the thing people reach for to explain why free hosting cannot be trusted with anything that matters. Every host in this archive paid part of that bill.
And the quieter one: acquisition
The precursors mostly did not fail. They were bought, and then switched off when they stopped fitting a portfolio. Yahoo closed GeoCities in 2009. Vistaprint closed Webs in 2023. Lycos, which had owned both Tripod and Angelfire since 1998, closed them on the same day in April 2026. No market forces required any of this. Somebody simply decided the line item was not worth keeping, which is its own kind of lesson about who you let host your history.
The one that never appears in the write-ups
Most post2host hosts were one person. One person paying the server bill, reading the abuse reports, moderating the quota-fillers, answering support at midnight, and doing all of it around a job. The archives record hosts that "declined". A fair number of them simply ran out of the evenings that were keeping them alive.
What this adds up to
Read together, the six causes split cleanly into two groups. Three of them are gone: link revenue has been priced out permanently, the paid market has finished falling, and the audience has finished migrating.
The other three are now tractable in ways they were not. Panel licensing has an answer, and iFastNet demonstrated it. Moderation at scale has an answer that did not exist in 2012. So does abuse detection: reading every uploaded file to spot a phishing kit was impossible for a one-person host and is now merely a matter of wiring it up. The obituary is accurate about 2013. It is not obviously accurate about 2026.