November 11, 2026
A pricing page looks like a weak page in most reports. Of the visits to the MachineTranslation.com pricing page between 5 August and 3 November, 62% clicked nothing at all, and the median visit lasted 24 seconds.
I'm the CMO at Tomedes, the translation company that develops MachineTranslation.com. The paywall opens most of our sales. The pricing page does a different job, and the reports we had barely credited it. Thirteen weeks of its data show what that job is, and why I think the usual analytics undercount the page.
The page closes sales that start somewhere else, it closes larger ones than the paywall does, and a landing-page report cannot see either.
Most buyers reach the MachineTranslation.com pricing page already on their way to paying. 72% of the first-time buyers who bought there, a small group over the 13 weeks, had been on the sign-up or log-in pages earlier in the same visit, and only 7% started their visit on the pricing page itself.
How a visitor arrives predicts what they do. Visitors from their account page click a paid plan 45% of the time, often through the Change Plan button, and visitors from the sign-up or log-in pages click 25% of the time. Visitors who land from outside the site click 3.9% of the time.
Here is how I read the spread. Visitors from the account page and the sign-up pages have already decided to pay, and they are looking for the plan. Visitors from outside the site are still deciding whether the product deserves a look. A pricing page can close the first group, and it has little to offer the second.
Device matters too. On desktop the plan cards sit on the first screen, and nearly all of the first-time buyers who bought on the page were on desktop.
A landing-page report credits the pricing page with almost nothing. Our own traffic report showed no first-time buyers who landed on it, because only 7% of the page's first-time buyers started their visit there. The sale opens on the paywall, the account page, or the sign-up pages, and the pricing page records the close.
That makes the page look weaker than it is. By the usual measures it reads as a dead end, since most visits are a quick look. The visits that click a plan tell a different story. They stay a median 3.4 minutes against 19 seconds for visits with no plan click, and 31% of them pay within the hour.
The page also works across visits. In the 7 days before paying, a further group of first-time buyers visited the pricing page and then completed a checkout on a paywall or through the account page. Counting them, about one in five first-time buyers either bought on the page or visited it within a week before buying.
A report that looks only at landing pages, or only at the session where the payment happened, will miss most of this. Counting visits in the week before buying adds about a quarter more buyers to the page's name, and the standard views cannot show it.
The paywall opens 74% of new payments, and the pricing page opens at least 13%. The page's payments are larger, though. Its median payment is about 2.7 times the paywall's because most of them are Pro, and it carries at least 19% of revenue.
The paywall is the screen that appears when a translation needs a paid plan. It catches people at the moment they need more, and it sells what that moment needs, which is a pass or a single unlock. The day pass is the clearest case, because 81% of day-pass payments come through the paywall. The pricing page sells the plan decision.
| Route | All new payments | Pro payments | Day-pass payments |
|---|---|---|---|
| Pricing page | 13% (at least) | 29% | 11% |
| Paywall | 74% | 41% | 81% |
| Other routes | 5% | 16% | 0% |
| Not traceable | 8% | 14% | 8% |
A common path runs from a paywall to sign-up, then to the account page, then to the Change Plan button, and finally to the pricing page. In that path the paywall opens the sale and the pricing page closes it. 95% of pricing page checkouts come from signed-in accounts, against 32% for the paywall.
Each surface does a different job. Marketers who compare them on payment count alone will rank the paywall first and stop there. They will miss that the two are linked.
The visitors who stay are checking limits and trust. Visits that opened the FAQ clicked a paid plan 21% of the time, against 11% for the rest, and the most opened questions are about download limits, free word limits, how the service compares with free tools, and which plan fits.
Those questions are the objections buyers raise before paying. The next most opened cover which AI models each plan includes, how human verification works, and cancellation. The page earns its keep by answering them where the buyer is already looking.
Signed-in visitors click a plan 35% of the time against 7% for guests. Buyers also move quickly once they decide. The first plan click comes a median 19 seconds after the page opens, and buyers finish paying a median 1.5 minutes after the checkout opens. I read that as a decision made before the page loaded, with the page there to remove the last doubt.
Measure intent by arrival path, and count the pricing page's help across visits, including the days before the buying session. Trace each payment to the checkout that opened it, split signed-in visits from guest visits, and report a floor when some payments cannot be traced.
We used the analytics we already had and our payment records, and the rest was method.
Two launches change the page, and we are watching both. The day picker went live on 9 October. Before it, every day-pass checkout opened on the pricing page was the 1-day pass, and after it two thirds were 3-day or 5-day passes. The numbers are small, so we read that as a direction.
The Team card went live on 28 October and is too new to read. The week it launched also brought more direct visits and fewer visitors from the account and sign-up pages, the groups that click most, so the conversion numbers for that week say little.
You can see the page at the MachineTranslation.com pricing page. The plans are explained in the post on pay once or go unlimited pricing, and the reasoning behind the price points is in AI translation pricing strategy.
It depends on how many translations you need and for how long. For a handful of translations in a short window, a single Unlock at $1.50 is cheaper, and above about four translations the $6 one-day pass wins. Monthly Unlimited at $19 pays off only for recurring volume across months.
A professional linguist reviews the AI output on top of the consensus translation. Every paid plan, meaning Unlimited, a day pass, or an Unlock, includes free human verification for up to 2,000 words a month, and beyond that allowance verification is available per word.
Yes, there are two limits to know. Files can be up to 70MB, and layout downloads carry a fair-use limit of 100,000 words a day. EPUB and Kindle formats are not supported, so books need to be converted to DOCX, PDF, or TXT first.
The free plan draws its consensus pick from standard, lighter model variants, and paid plans draw from the newest, most capable variants, with both shown side by side the same way. Paying customers get more advanced models because internal testing showed quality gaps widen on complex or idiomatic content.

By William Mamane
Connect on LinkedInWilliam leads marketing and growth at Tomedes, where he turns SEO experiments and real campaign data into the strategy behind MachineTranslation.com's reach. He writes about how AI products actually get found and trusted plus the tactics that work, the ones that don't, and what search is rewarding right now.