The email that asks a client how the project went almost never works. It arrives from the person who ran the project, lands in a thread that person will read, and asks for criticism of work that person did. The reply comes back as "really happy with it, thanks again", which is pleasant and tells nobody anything. Six months later the account does not renew and the reason turns out to have been the same thing three times over.
A form fixes part of that, not because forms are magic, but because they change who the answer feels addressed to. A structured form reads as a process rather than a personal question, and a client who would never write "the weekly calls were a waste of time" in a reply will happily tick 2 out of 5 on a communication scale. This guide covers the question set that gets usable answers, the timing that decides whether the answers arrive in time to matter, and the two things agencies actually want out of this: renewals and a quote worth publishing.
What the form has to overcome
Client feedback has three specific distortions that a general satisfaction survey does not.
The first is the relationship. The client likes the account manager and does not want to cause trouble for them. Anything that reads as a personal question gets a polite answer. This is why the sender and the recipient of the form matter as much as the questions, and why a form that says the responses go to someone other than the project lead gets more candid answers.
The second is the buyer being different from the user. The person who signed the contract often did not do the work of being a client. Feedback from the marketing director and feedback from the coordinator who chased assets every week are both worth having and will disagree, so send to both and keep the answers separate rather than asking one person to represent the company.
The third is timing against memory. Clients remember the last two weeks of a project and the worst single moment. A form sent at the end of a six month engagement measures the handover, not the engagement. That is a reason to ask during the work as well as after it.
None of these can be fixed with better question wording. They are fixed by deciding who sends, who receives, who is asked and when.
The question set
Six rated questions, three open, two permission questions. That is the whole form, and it takes a client four minutes.
Did the work achieve what it was meant to? This is the only question the client can answer that nobody inside the agency can answer for them. Ask it first, and name the objective from the brief rather than asking in the abstract.
Was it delivered when it was promised? Separate from quality, and the item most likely to be rated lower than the agency expects, because internal deadline changes that felt reasonable at the time read as slippage from outside.
How was communication? Frequency, clarity and responsiveness are three things and can be three questions for a long engagement. For a single project, one question plus the open box is enough.
How easy was the agency to work with? Effort, in other words. A client who got an excellent result after chasing for it every week rates quality highly and effort poorly, and that gap is where accounts are lost.
Was the work worth what was paid? Ask it plainly. The answer read against the actual fee is the single most useful input into how the next proposal is priced.
Would the work be given to the same team again? A yes or no with an open box. More honest than any scale, because it forces a decision.
Then the open questions. What worked, what should be done differently, and what almost stopped the project. The third one is the unusual one and it is the most productive, because it surfaces the near misses that never became complaints.
The two permission questions
May this response be quoted, and if so under what name? Offer four options: full name and company, first name and role, company only, or not at all. Asking at the moment the client is feeling positive is the only reliable way to collect testimonials, and specifying the attribution level up front removes the second round of emails.
Would you be willing to refer this work to someone? A yes, a maybe and a no. The yeses are a list worth working through and most agencies never build it, because the question is never asked in writing.
Questions to leave out
Skip anything the client cannot judge and anything that invites a compliment. "How would you rate the creative team?" gets a kind answer, because the client is rating people rather than describing an experience. "Were you satisfied overall?" adds a figure that then competes with the individual items and disagrees with them.
Also skip the question about what other services the client might buy. It reads as a sales question inside a feedback form, and it lowers the honesty of everything above it. Keep the commercial conversation in a separate conversation, where it belongs and where it will do better.
When to send it
| Point | Purpose | Length |
|---|---|---|
| Two or three weeks in | Catch a process mismatch while it can be changed | Two questions |
| Midpoint of a long project | Course correct before the result is fixed | Four questions |
| Two weeks after delivery | The full picture, with results starting to show | Full form |
| Retainer, every quarter | Renewal risk, not project quality | Full form |
| After a loss or non renewal | Why, while the client will still say | Three questions, mostly open |
Two of these are usually skipped and both are the valuable ones.
The check two or three weeks into a new engagement catches the mismatches that otherwise run for the whole project: too many meetings, the wrong person on calls, reports in a format nobody reads. These are trivially fixable in week three and permanent by week twelve.
The one sent after a non renewal is uncomfortable and it is the only feedback that comes from someone with no reason to be polite. Keep it to three questions, send it from someone senior rather than the account lead, and accept a low response rate. The few answers that come back are worth more than a year of end of project forms.
Note what is missing from the table. Feedback at the moment of delivery is too early. The client has not used the work yet, so the answer measures the handover meeting.
Who reads it, and who must not
The form has to go somewhere other than the inbox of the person being rated. This is the single design decision that determines whether the answers are honest.
For a solo freelancer that is not possible, and the honest workaround is to say in the request that the answers will be used to change how the next project runs, and to ask the uncomfortable questions in a way that gives permission to criticise. "What almost stopped the project" is easier to answer than "what did the freelancer do wrong".
For an agency with more than two people, route client feedback to someone who was not on the delivery team, then have them share it. That routing has to be built into the form rather than being a habit, because habits fail in the week somebody is on holiday.
The mechanics are the same ones that make any response queue work. Each response needs an owner, so a low score is a specific person's job. It needs a status, so the difference between read and acted on is visible. And the reply has to go out from the same place the response is read, so what was promised to the client stays with the record rather than living in one person's sent folder. That is the practical gap a form tool with response management fills, and it is worth checking what the tools after the form actually do rather than comparing question types, since every form builder can render a five point scale.
Keep one history per client, not one file per project
Agencies lose the value of client feedback by filing it per project. The third project for the same client produces a form that sits in a folder next to two others nobody opens, and the pattern that was visible across all three is never seen.
Responses keyed on the email address should join up into one contact, so the client who rated communication 2 out of 5 on the last two projects arrives with that already visible. The same applies at company level for accounts with several contacts.
Alongside the client's answers, record what the team did about them. Whether the complaint was raised internally, what changed, and whether the next project scored differently on the same item are three facts the client never sees, and they are the ones that turn feedback into a process change. Keeping them as fields on the response rather than as notes in a document means they come out as columns in an export and can be counted at the end of the year. The way feedback is received and worked through has more to do with whether anything changes than the question set does.
What to do with a bad answer
Decide the threshold in advance and write it down, because judgement made in the moment always finds a reason to wait.
Below the threshold, three things happen in order. Someone senior who was not on the project reads it the same day. They contact the client to listen rather than to explain, and the distinction matters, since a client who wanted to be heard and got a defence of the decisions stops answering forms. Then one concrete change is named, with a date, and sent back to the client in writing.
That last step is the one that gets dropped and it is the one that keeps the account. Clients rarely leave because something went wrong. They leave because they said something went wrong and nothing visible happened.
Resist the urge to argue with a rating. A client who rates the timeline poorly on a project that was delayed by their own asset approvals is still describing their experience accurately. The internal record can note the cause. The reply should not.
What to change first
Add the two permission questions to whatever is being sent now, because a testimonial and a referral list cost nothing to ask for and are never collected by accident. Then set up a check two or three weeks into the next new engagement, which is the cheapest feedback an agency can buy. When the responses start arriving faster than one person can track, see what one list from response to reply looks like in Halict.
Q1. What should a client feedback form ask?
Six rated questions covering outcome, timeliness, communication, ease of working together, value for the fee, and whether the same team would be used again. Then three open questions on what worked, what should change, and what almost stopped the project. Finish with permission to quote the response and willingness to refer.
Q2. When is the best time to send a client feedback form?
About two weeks after delivery, once the client has used the work but before the detail fades. For engagements longer than a couple of months, add a short check two or three weeks in, which is the only point at which a process mismatch can still be corrected cheaply. Retainers are better served by a quarterly cycle than by project triggers.
Q3. Should client feedback be anonymous?
Not for a small client base, where the answers are identifiable anyway and anonymity only removes the ability to act. Send it to a named person, say who will read it, and route it to someone who was not on the delivery team. Naming the recipient as someone other than the project lead does more for candour than anonymity does.
Q4. How do you get a testimonial from a feedback form?
Ask inside the form, at the moment the client has just written something positive, and offer a choice of attribution: full name and company, first name and role, company only, or no quote. Asking later as a separate request is a second favour and gets a much lower yield. Store the permission level with the response so nothing is published beyond what was agreed.
Q5. What response rate should a client feedback form get?
Higher than a consumer survey, because the relationship is real, and it depends almost entirely on who sends it and whether a previous round visibly changed anything. Track your own rate over time rather than a benchmark. A falling rate on an account is itself a warning sign, often the earliest one available.