Migrating off Nolt.io, Frill.co or Upvoty

Migrating off Nolt.io, Frill.co or Upvoty is the same job whichever one you are leaving: get your posts, votes and comments out, and land them somewhere else without losing the thread. None of the three publish confirmed detail on their own export format, so check each vendor's file before you commit to a date. That is the step people skip.

You have already decided to go, so this page will not talk you out of it. For the Canny.io version, the one vendor whose CSV-versus-API split we could confirm, see migrating off Canny.io. This page is one part of a larger guide: for the whole subject, see what feedback tools actually cost.

What do Nolt.io, Frill.co and Upvoty actually cost?

Cost is usually why someone starts looking for the door, so start there. All three run on a trial rather than a genuine free tier: Nolt.io gives ten days, Frill.co and Upvoty both give fourteen. None of the entry prices below are directly comparable, since each vendor prices by a different unit.

Say the billing basis every time you repeat one of these numbers. Nolt.io and Frill.co each publish a lower rate for paying yearly and a higher one for monthly, the pattern most of this category uses; over three years, on the figures above, Nolt.io totals $1,044 and Frill.co totals $792. Upvoty publishes only a monthly figure with no confirmed annual rate, so we will not invent one to set against those two totals.

Upvoty is also priced differently in kind: by project, with unlimited tracked users at every tier, the opposite of the tracked-user model most of the category runs on.

What can you actually expect to export?

Honestly, we do not know. We have only confirmed the shape of this problem for one vendor: Canny.io's CSV export carries post titles, statuses and vote counts, but not comments, which sit behind a separate JSON API instead.

For Nolt.io, Frill.co and Upvoty, no equivalent detail is confirmed. We have not verified the file format, whether comments and status history travel with it, or whether vote counts keep the names attached. Assuming any of the three behaves like Canny.io, or assuming it doesn't, would both be a guess dressed up as a fact.

Test it yourself before you set a date. This is the step almost everyone skips: read "export your data any time" on a pricing page, take it as settled, and only open the file the week you actually leave, the worst week to find a gap. Open your export now, while the account is live, and read what an export file actually contains for the columns worth checking.

What never comes across cleanly, whatever the vendor?

Four things are worth expecting trouble with, regardless of which of the three you are leaving.

• Vote attribution to real people. A vote count survives almost any export. Knowing which named customer cast each vote, so you can follow up or weight a request by who asked for it, is a separate, harder claim, and the one most likely to arrive as a number with the names stripped off.
• Comment threading. Even where comments are included, whether replies keep their parent and reply structure or land as one flat list is a different question from whether they made it into the file at all.
• The original timestamps. A request that looks brand new on your new board when it was actually filed two years ago misleads the customer who filed it and quietly rewrites your own sense of what is genuinely urgent.
• Public URLs already linked from elsewhere. Every request referenced in a changelog entry, a support reply or your own help pages has a URL doing real work. When the board moves, that link breaks unless something takes its place.

None of this means the move isn't worth making. It means treating "we support export" as a starting question, not a finished answer.

What order should you actually do this in?

The mechanics matter more than the vendor.

1. Export first, and read the file before you announce anything. Check what is actually in the columns against what the admin panel shows you. An announcement you have to walk back costs more trust than a quiet delay. 2. Decide what a vote is worth once it is detached from the voter. If the file gives you a count but not a name, decide now whether you carry that count across as-is, discard it, or manually reconcile the ones that matter, rather than deciding under pressure later. 3. Keep the old board readable while the new one fills. Customers keep old links in old emails and bookmarks longer than you expect. A board that still loads, even read-only, avoids a dead link turning up in someone's inbox six months on. 4. Set up redirects so the links already pointing at your board survive. Every request URL shared publicly should point somewhere real after the move, not a 404 on a domain you have already cancelled. 5. Only then, cancel. There is no upside to closing the account early, and every gap above is cheaper to catch before you do than after.

Does the EU Data Act give you any leverage here?

Some, but not the leverage it is usually described as giving. The EU Data Act has applied since 12 September 2025, and it requires a switching contract to set out three periods running one after another: a notice period of up to two months, then a transitional period of at most 30 calendar days for the port itself, then a retrieval window of at least 30 calendar days. The short clock caps the switch, not the wait, so a compliant vendor can spend two months before it starts.

We are not lawyers and this is not legal advice. The obligation binds providers of data processing services, a defined term in the Act, and whether a feature request board falls inside it is a judgement we are not qualified to make. Either way, no vendor in this category, Nolt.io, Frill.co or Upvoty included, mentions the Act anywhere in its own materials. If support is slow, or the file you get back looks thinner than the admin panel, asking which notice period and which transitional period the contract sets out is a concrete line for the email.

Frequently asked questions