What should a client satisfaction survey for consulting services ask?

A client satisfaction survey for consulting services should ask four scored questions and one open one. Score the consultant's expertise, their communication, how they worked with the client's people, and whether the delivery met expectations. Then ask one open question about what to do differently. Five questions. Two minutes. The same wording every single time.

That last part is the one people skip, and it is the one that decides whether the exercise was worth doing. A survey you rewrite for each client produces a number that is true on its own and comparable to nothing. A survey you never touch produces a trend, and a trend is the thing you can act on and the thing a buyer will believe.

This guide covers which areas to score, how to word each question, what scale to use, what to leave out, and what a finished survey looks like. If you are still deciding whether to measure at all, start with how to measure customer satisfaction for consulting services. For who to send it to and when, read the best practices for collecting feedback from business clients.

What should the survey ask, in short?

Five questions, in this order. The four scored ones come first because they are easy. The open one comes last because it takes thought, and thought is what people run out of.

Four scored areas is not an arbitrary number. It is small enough to answer in two minutes and wide enough that a low score points somewhere. One overall rating tells you the client is unhappy. Four tell you which part to fix.

Why should the questions be identical every time?

Because a score only means something next to another score.An 8.2 on its own is a number. An 8.2 against last year's 7.6, on the same question, asked the same way, is evidence that something you did worked.

Rewording breaks that comparison more thoroughly than most people expect. The Pew Research Center's guidance on writing survey questions (opens in a new tab)puts it plainly: even small wording differences can substantially affect the answers people give. Change “satisfied with the quality” to “happy with the result” between two surveys and you have measured two different things with the same label.

There is a second reason, and it is less comfortable. When you write the questions fresh for each client, you write them to your strengths. Nobody does this deliberately. Everybody does it. A fixed set removes the temptation before it arrives, which is also the answer you give a buyer who asks whether you chose the questions to flatter yourself.

The trade is real, and worth naming. Fixed questions cost you the ability to ask about the one thing unique to this engagement. Put that in the open question, or ask it in a conversation. Keep it out of the scored set.

How many questions should you ask?

Four to six, and the ceiling is set by your responder's patience rather than your curiosity. SurveyMonkey analysed 100,000 surveys run on its platform (opens in a new tab) — two thousand at each length from one to fifty questions — and found drop-off climbing with every question added, most steeply across the first fifteen.

Abandonment is not even the worst outcome. The worse one is that people finish anyway, badly. Survey methodologists call this satisficing: rather than doing the work of recalling and judging, a tired responder picks the first answer that looks acceptable and moves on. Jon Krosnick and Stanley Presser describe the mechanism in their chapter on question and questionnaire design (opens in a new tab), and note that responders often satisfy whatever motivated them early on, then grow fatigued as the questionnaire continues.

For a consulting client that ceiling arrives fast. They are not a research panel. They are a busy professional doing you a favour, usually between two meetings. A survey that takes two minutes gets answered honestly. A survey that takes ten gets a row of sevens.

Which four areas should a consulting survey score?

Score only what the client can see.That single rule eliminates most bad consulting survey questions, because it rules out anything about your process, your effort or your internal quality — none of which the client observed.

Expertise

Ask how the client rates your professional skill and knowledge. This is the one area every engagement shares, whatever its shape. It is also the score most likely to be high, which makes movement in it worth watching: a slip here usually means you were sold into work slightly above your level.

Communication

Ask about availability, clarity and kept promises. In professional services this is the score that separates a renewed contract from a polite ending. Clients rarely end an engagement because the work was wrong. They end it because they stopped being able to tell whether it was right.

Collaboration

Ask how you worked with the client's own people, inside and across teams. This is the area a project manager cannot rate for you, which is exactly why it belongs in a survey sent to more than one person. It is also the area where a consultant's self-assessment is least reliable.

Delivery against expectations

Ask whether the outcome matched what was agreed. Phrase it around expectations rather than around quality, because “quality” invites the client to rate an abstraction while “did this meet what you expected?” invites them to rate an event. A low score here is usually a scoping failure, and scoping failures are fixable before they happen.

Should the questions change with the type of engagement?

The question set should change, but only between a small number of fixed sets — never per client. A workshop attendee cannot rate your collaboration; they watched you for three hours. A client who never sits beside you cannot rate your effect on their working environment.

Three engagement shapes cover almost all knowledge work, and each needs its own fixed four. These are the client satisfaction surveys for consulting work in Feedback Journey, and they are a reasonable template even if you build your own.

Notice what stays constant across all three: expertise is scored every time. That is deliberate. It gives you one line you can follow across every kind of work you do, even when the other three questions differ.

How should each question be worded?

There is a published checklist for this, and it is short. Krosnick and Presser summarise the accumulated conventional wisdom of survey methodology in eight rules. Applied to a consulting survey they read like this.

1. Use simple, familiar words. No methodology jargon, no internal vocabulary. If your client would not use the word in a meeting, do not put it in a question.
2. Use simple syntax. One clause where possible. A question a responder has to re-read is a question they will guess at.
3. Avoid ambiguous words.“Regularly”, “proactive” and “value” mean different things to different people, so the answers will not be comparable even between two people in the same room.
4. Be specific and concrete.“How satisfied are you with the communication and availability throughout?” beats “How was the engagement?”
5. Make response options exhaustive. Every responder needs somewhere honest to land, including the one who cannot judge.
6. Avoid leading or loaded questions.“How happy were you with our excellent delivery?” is not a question. It is a request.
7. Ask about one thing at a time.Double-barrelled questions are, in Pew's words, difficult for respondents to answer and often lead to responses that are difficult to interpret. “Communication and availability” survives because they are one behaviour; “quality and speed” does not, because they trade off.
8. Avoid negations.“Was anything not delivered as expected?” makes the responder do algebra before answering.

Add one more rule that is specific to consulting: name the person. Write “How would you rate the professional skills of Sara?” rather than “of the consultant”. Naming makes the question concrete, and it stops a responder from quietly rating the supplier when you asked about the individual.

What rating scale should you use?

Use 0 to 10, and label what the numbers mean. The labelling is the part that gets dropped, and it is the part that does the work.

The research is not unambiguously on the side of a ten-point scale. Krosnick and Presser's review concludes that seven-point scales are probably optimal in many instances, and that reliability is higher when every point is labelled with words rather than only the ends. An unlabelled 0 to 10 leaves each responder to invent their own meaning for a 7, and they will not all invent the same one.

A 0 to 10 scale still wins in business feedback for two practical reasons. Business responders already know it, and a wider scale leaves room for a trend to be visible year over year. The way to recover the lost precision is to label the ranges at the moment someone picks a score, so the scale is defined when it is used rather than in a preamble nobody read.

Label them roughly like this, and keep the labels identical across every survey you send:

That top label is the anti-inflation mechanism, and it matters more than the scale length. Told nothing, a polite client gives everyone a 9 or 10 and your data flatlines. Told that a 10 is reserved for world class, the same client gives an 8 and adds a sentence about what would make it a 9. The second answer is the useful one.

Should you offer a “don't know” option?

Yes for consulting surveys, even though general survey research is sceptical of them. The disagreement is worth understanding, because it shows you when to use one.

Krosnick and Presser argue that offering “don't know” on attitude questions does not consistently improve data quality. Their reasoning: most people who choose it are not missing information at all. They are ambivalent, or confused by the question, or taking the low-effort exit. The option becomes a cue to satisfice, and real opinions go unrecorded.

A consulting survey is a different case. You are not asking one person for an attitude they definitely hold. You are asking several people about specific behaviour, and some of them genuinely did not see it. A junior developer who worked beside you daily can rate your collaboration precisely, and may have no basis at all to rate how you handled the steering group.

Without an opt-out, that person does not skip the question. They split the difference and enter a 7, and your average now contains a number that describes nothing. One invented score in a survey of five responders moves the category average by a fifth. Give people the exit, and make it dignified rather than apologetic: “I don't know”, not “not applicable”.

Two conditions keep the option honest. Score it as absent rather than as a zero, so an opt-out never drags an average down. And watch the rate: if a category collects a lot of them, you have either asked the wrong people or asked about something invisible.

What open-ended question should you ask?

Ask one, put it last, and make it optional.One question, phrased forwards. “What is the one thing I could have done better?” works better than “Do you have any other comments?”, because a blank invitation gets a blank answer.

Keep it at the end for a mechanical reason. SurveyMonkey's analysis found completion rates fall when a survey opens with an open-ended question, and fall again when it carries more than a handful of them. Scored questions are cheap to answer; typing is expensive. Collect the cheap answers first, so a responder who quits has still given you something.

Read the text before the scores. A category average tells you where to look, and the sentence next to it usually tells you why. This is also the material worth keeping verbatim in your records — documenting client feedback for service improvement covers what a record needs to contain.

Should you borrow NPS, CSAT or CES questions?

Borrow the wording discipline, not the whole instrument. All three are single-question metrics designed for high-volume consumer relationships, and a consultancy has neither the volume nor the anonymity that makes them work.

The recommendation question, popularised by Fred Reichheld in “The One Number You Need to Grow” (opens in a new tab)in 2003, is still worth asking — just not as your only question. Its real value in consulting is not the score. It is that a client willing to answer it positively is a client willing to be quoted, which is a different and more useful asset. Whether one number is enough as a metric is argued out in the guide to measuring customer satisfaction.

The Customer Effort Score, introduced by Matthew Dixon, Karen Freeman and Nick Toman in “Stop Trying to Delight Your Customers” (opens in a new tab)in 2010, is the worst fit of the three. It measures how hard a customer had to work to get a problem solved. Hiring a consultant is meant to involve the client's people; low effort would mean you worked alone.

Which questions should you leave out?

Cut anything the client did not observe, anything with two answers, and anything that sells. Most bloated consulting surveys are bloated with these.

• Price and value-for-money. The person who answers your survey usually did not sign the contract. Asking them to rate your rate produces an opinion about budgets you cannot act on.
• Anything about your internal process. The client did not see your estimation, your tooling or your effort. They saw the output.
• “How did we do?”Too broad to score and too vague to compare. It reliably returns “great, thanks”.
• Double-barrelled questions.“Was the work delivered on time and to budget?” has two answers and one box.
• Demographics and firmographics. You already know who the client is. Every field you ask for is a field between them and finishing.
• Next-sale questions.“Would you be interested in further services?” turns a feedback request into a pitch, and responders adjust their scores once they notice.
• A comment box under every question. One at the end, not five throughout. Five boxes read as homework.

Should you ask about you, or about your firm?

Ask about the named individual, because in professional services the individual is the product. A client does not experience your consultancy. They experience the person in the room.

Firm-level questions also destroy the granularity that makes feedback actionable. “How satisfied are you with Acme Consulting?” averages a strong architect with a weak project manager into a number that describes neither and improves nothing. Per-person, per-engagement scores tell you which person, which client and which category needs attention.

This is also what makes the results portable. A firm-level satisfaction score belongs to the firm. A score attached to your name and your engagements travels with you into the next proposal, and client feedback for consultancy leaders aggregates the same per-person records upwards when the firm needs a view.

Should you ask for a testimonial in the same survey?

Yes, but as a separate, optional, clearly consented question after the scoring. The moment right after someone has rated you well is the best moment you will ever get to ask for a quote, and almost nobody asks.

Three things make the request work rather than backfire:

1. Put it last. After the scores are locked in, so the request cannot influence them.
2. Ask for name and title with the words. An unattributed quote is worth very little to the next buyer.
3. Say where you will use it,and get explicit agreement to that use. “May I use this in proposals and on my website?” is a complete consent question.

Keep the testimonial and the scores separate in your own head, too. A quote is marketing; a score is evidence. The scores are what you show when someone asks whether the quotes were cherry-picked.

What does a complete consulting client satisfaction survey look like?

Here is the whole thing, in order. Six items. Two minutes. Copy it and change nothing but the name.

1. Expertise.“How would you rate the professional competence and expertise of [name]?” Scored 0–10, with 0 as “I don't know”.
2. Communication.“How satisfied are you with [name]'s communication and availability throughout?” Scored 0–10.
3. Collaboration.“How would you rate [name]'s collaboration with you and your team?” Scored 0–10.
4. Expectations.“How well did [name] understand and meet your needs and expectations?” Scored 0–10.
5. One open question.“What is the one thing [name] could have done better?” Optional, free text.
6. The recommendation, optional.“Would you be willing to recommend [name]? If so, in your own words — and may I use it in proposals and on my website?” Collect name and title with it.

Send it to everyone who worked with you, not only the person who signed the contract. The signatory rates the outcome; the team rates the working. Sampling only the signatory hides exactly the information you are trying to surface, and the guide to collecting feedback from business clients covers who to invite and how many.

Before you send it, check

• Every question asks about one thing.
• Every question is about something the responder actually saw.
• The scale meaning is visible at the moment of scoring, not buried in a preamble.
• There is one open question, and it is at the end.
• The wording is identical to your last survey. Word for word.
• It takes under two minutes. Time yourself.

What do the Feedback Journey question sets ask?

Feedback Journey ships the fixed question sets so you do not have to write or defend them.Three engagement types, four scored questions each, on a 0–10 scale where 0 means “I don't know”. The questions do not change, which is the point.

Four design decisions follow directly from what this guide argues:

1. The questions are not configurable.That is a deliberate trade of flexibility for comparability — results line up across clients, colleagues and years without anybody having to normalise anything.
2. Scoring guidance appears as the responder picks. The 10 is described as world class and rare, so scores land where they mean something instead of clustering at the top.
3. Zero means “I don't know”, so someone who never saw a category can say so instead of inventing a middle score.
4. Free text and an optional recommendation sit after the scores,with the responder's name, title and consent attached to anything you may reuse in marketing.

What comes after the survey is where the questions earn their keep. A weak category becomes a documented improvement plan tracked through later surveys, and a range of completed surveys becomes a verifiable proof of client satisfaction that a buyer can check with a QR code. Neither is possible if the questions moved between surveys.

If your firm is certified, the same fixed sets do double duty as clause 9.1.2 evidence — documenting customer satisfaction for an ISO 9001 audit covers what an auditor looks for.

Frequently asked questions

Sources

• Jon A. Krosnick and Stanley Presser, “Question and Questionnaire Design” (opens in a new tab), in Handbook of Survey Research(2nd edition), Wright and Marsden (eds.), Elsevier, 2009 — the eight conventional-wisdom wording rules, satisficing and respondent fatigue, evidence that seven-point scales are probably optimal in many instances, higher reliability when all scale points are verbally labelled, and the case against routinely offering “don't know” on attitude questions.
• Pew Research Center, “Writing Survey Questions” (opens in a new tab)— small wording differences substantially affecting answers, double-barrelled questions being difficult to answer and to interpret, question-order contrast and assimilation effects, and how open versus closed formats change the distribution of responses.
• SurveyMonkey, “Does adding one more question impact survey completion rate?” (opens in a new tab) — an analysis of 100,000 surveys run between January 2009 and September 2010, two thousand at each length from one to fifty questions, finding drop-off rising with each added question and most steeply over the first fifteen.
• Fred Reichheld, “The One Number You Need to Grow” (opens in a new tab), Harvard Business Review, December 2003 — the origin of the Net Promoter recommendation question.
• Matthew Dixon, Karen Freeman and Nick Toman, “Stop Trying to Delight Your Customers” (opens in a new tab), Harvard Business Review, July–August 2010 — the introduction of the Customer Effort Score and its original effort question.
• ISO 10004:2018, Quality management — Customer satisfaction — Guidelines for monitoring and measuring (opens in a new tab) — guidance on identifying what to measure and designing the method that collects it.

Start asking

The question sets are already written. Send your first survey in a few minutes — free, no credit card required.

Start free — no credit cardSee the question sets