We didn't promote the AI translation track. People found it anyway.

A few weeks ago I wrote about the decision behind Tomedes' AI translation pricing: a flat $0.01 per word, no human proofreading in the default workflow, launched deliberately without knowing what buyers would actually do with it. I said we were watching the engagement data with genuine curiosity, and that the data would shape real decisions rather than our assumptions shaping it for us.

We have not run a single campaign for the AI translation track. No paid ads pointing to it, no email push, no banner announcing it. It sits on the quote page we rewrote to be honest about what each track actually is, labeled clearly next to Human Translation, and that is the only place anyone has seen it.

Volume on that track has grown every week since launch, without a dollar of paid promotion behind it.

Where is this traffic actually coming from?

Every order on the AI translation track so far has come through the quote page itself, not from a campaign directing someone there. That distinction matters more than the volume number would on its own. A promoted feature getting traffic tells you the promotion worked. An unpromoted feature getting traffic tells you the demand was already there, waiting for a clear, honestly labeled option to buy.

That is the more useful signal. It means the earlier decision, naming the track "AI Translation" instead of burying it inside a vague "fast delivery" option, was not just a transparency exercise. It removed friction between people who wanted this specific product and the moment they could actually purchase it.

Are buyers treating this as a one-time test, or coming back?

They are coming back. People who order AI translation once are returning to order it again, not treating it as a one-time experiment with something new. That is a different signal than curiosity. Curiosity gets you a single order. Repeat orders get you a buyer who found the product did what they needed the first time and came back for more of it.

Is the open question from last time getting answered?

Slowly, yes. In the pricing post, I said we launched without a default human proofreading layer specifically to see what buyers would ask for before we built that assumption into the product. Some buyers are now requesting MTPE, human post-editing, as an add-on. Not a large share yet, and that is not surprising, because we have not made that option visible or easy to find. It exists, but we have not told anyone it exists in any deliberate way.

That is the next honest thing to say publicly: the low add-on rate right now reflects that we haven't asked the question yet, in a place where buyers would see it, rather than reflecting a lack of interest. The next step is straightforward: surface the MTPE option clearly at the point of purchase, the same way we surfaced the AI Translation option itself, and see what changes.

What does this actually change, and what's still unresolved?

This is not a finished story. Early demand and repeat orders are encouraging, but we still don't know what the MTPE add-on rate looks like once it's actually visible to buyers instead of buried. We still don't know whether the buyers ordering AI translation today are the same buyers who would have ordered Human Translation before we split the tracks apart, or whether this is genuinely new demand that a vague, undifferentiated quote page was leaving on the table.

What I can say honestly is this: removing ambiguity from the quote page did not just make our pricing page more honest. It appears to have made a real, previously invisible demand visible, on both sides of the line we drew. That was the bet in the first post in this series. So far, the data is backing it.

I'll keep reporting back on what we learn, including the parts that don't go the way we expect.

Get a quote and see the AI Translation option live: https://www.tomedes.com/quote