Feedback tool pricing in 2026: what six vendors actually meter

July 28, 2026

Most comparisons of feedback tools line up the monthly prices and declare a winner. That is the least useful thing you can know about them.

The monthly price is what you pay in month one. What decides your bill in month eighteen is the meter — the thing the vendor counts. And in this category the meter is almost always some measure of how much feedback you are collecting. Which means the bill goes up precisely when the programme is working.

Here is what each of six tools counts, with the numbers read off their own pricing pages.

What each tool counts

Those are each vendor's cheapest annual rate — the number most favourable to them.

The three shapes this takes

1. Paying per person who participates

Canny's meter is "tracked users": people who post, vote or comment. The $79 plan includes 100 of them.

Think about what that means for a public roadmap. The entire goal is to get lots of customers voting on what you build next. A hundred voters is not an ambitious number — it is one decent week after you put the board in your product's navigation. Success on the metric you care about is the same event as crossing a billing threshold.

There is a defensible logic to it. More participants genuinely is more value delivered, and it scales the price with the size of the customer. But it does put you in the position of quietly hoping fewer people vote, which is a strange incentive to build into the tool you bought to get more votes.

2. Paying per person who was merely present

Beamer's meter is monthly active users, defined on their pricing page as unique users who access any page where the script is embedded.

Read that carefully, because the distinction matters: it is not people who opened your changelog. It is people who loaded a page that contains it. Put the widget in your app's header — which is the recommended installation, and the only one that gets a changelog read — and your meter is now your product's MAU.

The $49 plan covers 1,000. Their ceiling is $249/mo at 50,000, and above the cap it is $50 per extra 5,000 users.

So a changelog on a product with 60,000 monthly users costs more than $249/mo, whether or not a single person clicks it. You are not paying for changelog engagement. You are paying for your own traffic.

3. Paying per unit of the thing you collect

Survicate meters responses: $114/mo for 500 a month, rising to $569, with overage fees when you go past your allowance.

This is the most direct version. A survey that nobody answers is cheap. A survey that works is expensive. If you run NPS across a customer base of any size, response volume is not a dial you control precisely, and a good quarter is an unplanned invoice.

Marker.io adds a variation worth noticing: seats and projects, but also a 5,000 page-view cap on the entry plan. For a bug-reporting tool that lives on a staging site, that is usually fine. For one installed on production, 5,000 page views is a small site.

The stack problem underneath all of this

The other thing the table hides is that these tools each cover a slice. If you want the full set of feedback surfaces — a feedback widget, surveys, visual annotation for bug reports, a changelog, a public voting board, and testimonials — no single vendor above sells you all of it.

The cheapest honest combination, using each vendor's best annual rate:

Beamer is deliberately left out rather than double-counted, since Canny already includes a changelog.

Four vendors. Four dashboards, four logins, four invoices, four support queues — and three of the four bill you more as more people respond. Feedback from the same customer arrives in four different places with no way to connect it.

What we do, and where the argument runs out

We built Feedback Magic with a flat price: $29/mo for five widgets on one project, $79/mo for twenty-five widgets across five projects. Responses, voters, page views and reviewers are unlimited and never metered. The same $79 that covers a quiet month covers the month your NPS survey gets ten thousand answers.

We think that is a more durable argument than being cheap, because it stays true even if we raise prices. But three honest caveats, because a comparison that wins every row persuades nobody:

• Metered pricing is not a scam. It ties cost to value and it lets a vendor offer a genuinely free entry point. Four of the six tools above have a free tier. We have none — a 7-day trial and then a plan. If free matters more to you than flat, several of these beat us outright.
• Several of them do things we do not. Hotjar and Userback record sessions; we only capture what someone chose to tell us. Senja collects video testimonials; we collect text. Survicate has native iOS and Android SDKs; we are web-only. Canny's integration ecosystem is far deeper than our seven adapters.
• They have a track record and we do not. Years of customers and public reviews are a real thing to weigh, and no amount of writing changes it.

If you want the full comparison — including the rows where they win — it is on our comparison page, with every price dated and linked to the vendor's own page.

How to check this for yourself

The most useful ten minutes you can spend before buying any of these: open the vendor's pricing page and find the sentence that defines their unit. Not the price — the unit. "Tracked user", "MAU", "response", "seat", "page view". Then work out what that number will be for you in a year if the tool does exactly what you bought it to do.

That is the number that determines what you pay. Everything else is the first month.

One warning from doing this ourselves: when we built our comparison data, we checked six vendor prices against the "best alternatives" listicles that rank for these searches. Three of the six were materially wrong — one by more than $300 a month at the ceiling. Go to the vendor's own page. Including when you are checking ours.

On the numbers: every competitor price here was read off that vendor's own pricing page on July 26, 2026, not from a comparison listicle — three of the six we first checked were materially wrong. Pricing changes; if you are reading this much later, check the vendor's page and tell us if we are stale.