Transcription

A transcription and translation service selling across five related offers: transcription, translation, writing, localisation, and subtitling. Its customers are unusually varied, spanning business, research, call centres, medical, legal-adjacent work, and education, and they arrive with very different volumes, accuracy requirements, and budgets.

Language services are sold on two things that are hard to show: accuracy and price. Accuracy is invisible until you receive the work, and price depends on audio length, turnaround, and verbatim style, so a single headline figure is either misleading or meaningless.

The category's central marketing problem is that **buyers cannot verify quality before purchase and cannot easily work out what they will pay**. Both unknowns stall the decision. A visitor who cannot estimate cost will not enquire, and a visitor who cannot assess accuracy will default to whichever provider looks cheapest, which is a race the service does not want to run.

There is a second structural problem. Five service lines and at least six distinct customer types produce a combinatorial mess: writing a page for each audience is unmanageable, and writing one generic page speaks to nobody. The visitor has to find themselves quickly.

The site turns both unknowns into interactive answers rather than claims.

For price, it provides a **transcription cost calculator**, letting a visitor generate their own per-audio-minute figure instead of requesting a quote and waiting. For quality, it offers **sample transcripts**, so the visitor can inspect real output and learn the vocabulary of the category, full verbatim, clean verbatim, timestamps, at the same time. Together these convert the two blocking unknowns into self-service tools, which suits a buyer who wants to evaluate privately before speaking to anyone.

Turning an unknown into a tool beats explaining it. Both blockers on this page, what will it cost and will it be any good, could have been addressed with reassuring copy, and copy would have changed almost nothing. A calculator and a sample viewer let the visitor answer their own question privately, and self-generated answers are trusted in a way that vendor claims are not. I now look for the blocking unknown on any service page and ask whether it can be made interactive.

Showing the process is a stronger proof than stating the outcome. Claiming high accuracy is free, so it reads as noise. Naming four production steps costs the business something real and is checkable, which makes it credible. Where a quality claim cannot be verified, describing the machinery behind it is usually the more persuasive route.