# Transcription

By Munawar Khel / Independent Practice / Freelance UI/UX Designer / 2016–19 / 5 min read

Published at https://munawarkhel.com/work/independent-practice/transcription/

Delivered a complete responsive marketing and service website for a multi-service language business as a solo freelance engagement

- Company: Independent Practice
- My role: Freelance UI/UX Designer
- Timeline: January 2016 – December 2019
- Platform: Responsive marketing and service website
- Category: Language Services / Transcription and Translation / Marketing Website
- Status: New design, delivered
- Tools: Sketch
- Team: Solo, working directly with the client

## Context

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 problem

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.

A third issue is trust in an era of automation. Buyers increasingly wonder whether they are paying a human premium for machine output, and the site had to address that head-on rather than ignore it.

## What I did

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.

Accuracy, being unprovable in the abstract, is made **procedural**: a four-step process, transcription, review, proofreading, and quality check, presented as a numbered sequence. Rather than claiming accuracy, the page shows the machinery that produces it, which is a far more credible argument.

Audience breadth is handled by a **"Great For" grid** naming each customer type explicitly, business, research, call centres, medical, translation, education, so visitors self-identify in seconds without the page needing a variant per segment. The five service lines sit in a horizontal card carousel with one card highlighted, so breadth is visible without the page fragmenting.

The automation question is met directly with a "machines versus humans" comparison, turning a potential objection into a demonstration of why the human service is worth paying for.

## Outcomes

- Delivered a complete responsive marketing and service website for a multi-service language business as a solo freelance engagement
- Converted the category's two blocking unknowns, price and quality, into self-service tools: a per-audio-minute cost calculator and inspectable sample transcripts
- Reframed accuracy as a visible four-step production process rather than an unverifiable percentage claim
- Enabled six distinct customer segments to self-identify on a single maintainable page rather than through per-segment variants
- Addressed the automation objection directly with a machines-versus-humans comparison

## Decisions

### Replace the quote request with a self-service cost calculator

What: Gave the visitor a calculator that generates their own per-audio-minute price, rather than routing every pricing question through an enquiry form.

Why: A flat rate is dishonest because price genuinely depends on length, turnaround, and verbatim style, and a quote form asks the visitor to give up contact details before they know whether they can afford the service, which loses the cautious majority. A calculator answers the blocking question immediately and privately, and a visitor who has seen a number they accept is far closer to enquiring.

### Prove accuracy with process rather than claims

What: Presented a four-step production chain, transcription, review, proofreading, quality check, as a numbered sequence, positioned as the accuracy guarantee.

Why: A percentage is unverifiable and every competitor quotes a similar one, so it carries no information. Showing the steps that produce accuracy is concrete, differentiates on operations rather than assertion, and implicitly justifies a price above automated alternatives.

### Let the audience self-identify instead of building per-segment pages

What: Named the customer types explicitly in a single "Great For" grid, business, research, call centres, medical, translation, education, rather than creating separate landing pages or a generic page.

Why: Five service lines multiplied by six audiences is unmaintainable as separate pages, and a neutral page leaves every visitor unsure the service is for them. Naming the segments lets each visitor recognise themselves in seconds while the page stays single and maintainable, which matters for a solo-delivered site with no ongoing content team.

## What I learned

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.

Naming your audiences out loud is underrated. It felt almost too plain to simply list the customer types, but a page cannot make six different segments each feel addressed through tone alone. Explicit naming did in one grid what an entire restructure into separate pages would have done, at a fraction of the maintenance cost.

Tags: Art Direction, Responsive, Marketing Website, Landing Page, Conversion Design