Guides
How to fix translation agency intake before new accounts disengage
Improving translation client onboarding: four measurable symptoms, what each one says is missing, and a diagnostic table for the first ninety days.
What to take away
- Diagnose before redesigning. Onboarding failures show up as four measurable symptoms, and each points at a different missing step.
- Count the reviewer's first-round changes on projects one, two, and three. If the count is not falling, terminology is not being captured.
- Count queries answered after the deadline. That number is a client-behavior problem you can fix with process, not persuasion.
- Count deliveries that needed a format fix. That is an intake failure, always.
- Fix one symptom per account per month. Changing everything at once tells you nothing about what worked.
The four symptoms
Symptom 1: reviewer changes are not declining
Record how many changes the client's in-country reviewer makes on each of the first three projects, and sort them into corrections and preferences.
If corrections dominate and stay flat, the linguist brief is thin. If preferences dominate and stay flat, decisions are being made repeatedly and never written down. The second is more common and more fixable: every preference change should end up in the glossary or the style note the same day, with the reviewer's reasoning attached.
Symptom 2: queries are answered late or not at all
Count queries sent, answers received, and days to answer.
Late answers are usually a format problem rather than an attitude problem. Questions arriving one at a time, phrased as open questions, with no proposed reading, are genuinely hard to answer. Batch them, propose a reading for each, and state what you will do by default if no answer arrives. Most accounts improve immediately.
Symptom 3: deliveries need a format fix
Count how often the client had to convert, reformat, or rebuild what you sent.
Every instance traces back to a missing intake answer: which format, what the file does next, who opens it. Fixing it after delivery is expensive and visible. Fixing it during intake costs one question.
Symptom 4: the same question arrives from a different person
Count how many people on the client side have asked you the same thing.
That is a sign that nothing you sent has been circulated internally. Give the client one short account document they can forward: what was agreed, who does what, and the review windows. It is one page and it saves you from being the client's internal memory.
A diagnostic table
| Symptom | Most likely missing step | The fix |
|---|---|---|
| Reviewer changes stay flat | Preferences never captured | Log every change into the glossary with its reason |
| Queries answered late | Queries not batched or not answerable | Batch, propose readings, set a default |
| Format fixes after delivery | File specification never gathered | Add it to intake and confirm before the first batch |
| Repeated questions from new people | No forwardable account summary | One page the client can circulate |
Measure the first ninety days, then stop
Onboarding metrics are for onboarding. After the account is stable, tracking reviewer change counts monthly turns into a ritual that nobody reads.
Run the four counts across the first three projects, act on the worst one, and then keep only whichever one showed a real problem for that client.
Two things that look like onboarding problems and are not
A client whose source text is poor. That is a content problem, and no amount of onboarding process fixes it. Say so early, in writing, with examples, and offer source editing as a paid step.
A reviewer who disagrees with the brief. Sometimes the reviewer is right and the brief was wrong. Sometimes the reviewer was never shown the brief. Find out which before changing your process, because those need opposite responses.
An account that also needs interpreting. Spoken work is scheduled by appointment and written work is scheduled by volume, and the occupational description of interpreting and translation treats them as separate professions with separate practice. An onboarding process built for document flow will not hold an interpreting request; give it its own intake.
Handling material that has to be handled carefully
New accounts often arrive with confidential or regulated content. Onboarding is when you say how it is stored, who can reach it, and what happens at the end. Write that down once and check it against what your subcontractors have signed.
What a specific client's sector requires is a question for their own compliance function. What applies to you is a question for a qualified adviser.
Where the translated output will be seen by the public, the client may carry accessibility obligations that change the deliverable format. The Department of Justice small business guide to the ADA is a starting point for the general topic in a United States context and is not a ruling on any particular organization. Ask during intake rather than after delivery.
If you ask new clients for feedback that becomes public, apply one process to everyone, disclose material connections, and leave critical feedback visible; the FTC guidance on soliciting and paying for online reviews covers the practices to avoid.
The full intake and delivery model these symptoms are measured against is in the onboarding and delivery guide. When reviewer changes point to a linguist selection problem, not a briefing problem, see how the bench is built and trained.
Unbilled fixing shows up in the numbers in the pricing and profit guide. An account that onboards cleanly is the strongest asset behind the marketing and growth guide.
Common questions
Which symptom should I fix first?
The one with the highest count, unless format fixes are happening at all, in which case fix that first because it is the cheapest and the most visible to the client.
How many projects before I judge an account?
Three. One tells you nothing, two shows a direction, three shows whether the direction holds.
Should the client see these measurements?
Some of them, framed usefully. "Your reviewer's changes dropped by half once we captured the terminology" is a good conversation. A dashboard of internal counts is not.
What if onboarding goes well and the account still stalls?
Then the problem is upstream: the content program itself may have changed, or the person who bought may have moved. Ask directly rather than improving a process that is already working.
