Customer Feedback Survey Template (with AI Follow-Up Questions)

The Bottom Line

A good customer feedback survey template has three parts: a satisfaction metric you can track over time, a few diagnostic questions that explain the score, and at least one open-ended question that surfaces what you did not think to ask. Below is a copy-ready template you can use today. But the single biggest upgrade is not the wording — it is the medium: run the template as an AI-moderated interview so every "it was fine" gets a follow-up "what would have made it great?" With a platform like Koji, the same template auto-charts the metrics and auto-themes the open answers, so you skip the spreadsheet entirely.

The Template (Copy-Ready)

1. Overall satisfaction (scale, 1–5)

"How satisfied are you with [product] overall?" — 1 = Very dissatisfied, 5 = Very satisfied

2. Likelihood to recommend / NPS (scale, 0–10)

"How likely are you to recommend [product] to a friend or colleague?"

3. The diagnostic open-ended (open_ended)

"What is the main reason for your score?"

4. Value perception (scale, 1–5)

"How well does [product] meet your needs?"

5. Feature/area satisfaction (single_choice or multiple_choice)

"Which part of [product] do you rely on most?" — list your core areas

6. The improvement question (open_ended)

"If you could change one thing about [product], what would it be?"

7. Effort (scale, 1–7)

"How easy was it to get what you needed today?" — 1 = Very difficult, 7 = Very easy

8. The open mic (open_ended)

"Is there anything else you want us to know?"

That is eight questions: three quantitative metrics you can trend, two diagnostics, and three open-ended prompts. Keep it to 5–8 questions — past that, completion rates fall off a cliff.

Why Question Type Matters

In Koji, the question type decides how the AI asks it and how the report visualizes it. The template above maps cleanly onto Koji's 6 structured question types:

• scale → satisfaction, NPS, value, and effort (distribution charts; NPS gets the standard 0–10 range)
• single_choice / multiple_choice → "which area do you rely on" (frequency bar chart)
• open_ended → the reason, the one change, and the open mic (themed analysis with quotes)

Using the right type means your dashboard builds itself. A scale question becomes a distribution chart; a choice question becomes a frequency chart; an open-ended question becomes a ranked list of themes. See the structured questions guide for the full set.

The Upgrade: Add AI Follow-Up to Every Answer

A static template captures one answer per question. The questions that matter — "what is the main reason for your score?" and "what one thing would you change?" — are exactly the ones a flat survey under-serves. People type "support" and move on. You needed to know which support interaction, when, and why it stuck with them.

This is where converting the template into an AI interview changes the economics. For each open-ended question, set the AI's probing depth (maxFollowUps from 0 to 3). For your two highest-value questions, set 2–3 follow-ups so the AI naturally asks "Can you give me an example?" or "What would have made that a 10?" On the satisfaction and NPS scales, enable an anchor follow-up so a 6/10 automatically triggers "What would move that closer to a 9?" — turning a number into a prioritized fix list. Platforms like Koji do this automatically; with a static form you would have to email every detractor by hand.

Voice or Text, Multilingual by Default

Koji runs the template as voice or text, participant's choice. Voice answers to "what would you change?" tend to be richer because talking is lower-effort than typing. If you sell internationally, the AI conducts and analyzes the interview in the participant's language and reports back in yours — no separate translation pass.

Reading the Results

As responses arrive, Koji aggregates in real time. Your satisfaction, NPS, value, and effort scales render as distribution charts you can trend release over release. The "which area do you rely on" choice question becomes a frequency chart. And the three open-ended questions are coded into themes — each with the count of how many people raised it and the verbatim quotes that prove it. Instead of reading 200 spreadsheet rows, you open a report that already says "32% mentioned slow load times; here are their exact words."

How Often to Send It

• Post-interaction (transactional): fire the effort + satisfaction questions right after a support ticket or onboarding milestone.
• Quarterly (relationship): send the full 8-question template to your active base every quarter to trend NPS and satisfaction.
• Always-on: embed the link in-product so feedback trickles in continuously rather than in a once-a-quarter spike.

Tailoring the Template by Goal

The eight-question template is a strong default, but trim it to the decision you are trying to make:

• Measuring loyalty? Lead with NPS (scale, 0–10) and the segment-aware reason follow-up. Drop the effort question. This is your quarterly relationship pulse.
• Diagnosing a specific feature? Replace the generic "which area do you rely on" with a single_choice scoped to that feature's sub-parts, and make "what would you change?" the deep-probed question (maxFollowUps 3).
• Reducing churn? Add a forward-looking open_ended: "What would make you keep using [product] a year from now?" Probe it hard — the answer is your retention roadmap.
• Post-support follow-up? Keep only the effort scale and one open-ended ("what would have made that easier?"), triggered immediately after ticket resolution.

The rule of thumb: every question must map to a decision. If you cannot name the action a question's answer would drive, cut it.

Common Mistakes That Sink Feedback Surveys

• Double-barreled questions. "How satisfied are you with speed and support?" forces one answer to two things. Split them.
• Leading wording. "How great was your experience?" biases the score upward. Stay neutral: "How was your experience?"
• Too many open-ended questions. Three is the ceiling for a static form. With AI follow-up you need fewer open questions, not more, because each one goes deeper.
• Asking and never closing the loop. Customers who give feedback and see nothing change stop responding. Report back what you fixed.
• Scale inconsistency. Mixing 1–5 and 1–10 across questions confuses respondents and muddies trending. Pick a convention (1–5 for satisfaction, 0–10 for NPS, 1–7 for effort) and keep it stable.

The template above plus AI follow-up handles all five by design: neutral wording, a capped open-question count, consistent scale ranges, and themed reporting that makes closing the loop trivial.

Boosting Response Rates

The template is only as good as the responses it collects. Three levers move the needle most: timing (send transactional questions within minutes of the interaction, while it is fresh), brevity (a visible "3 questions, ~2 minutes" promise lifts starts), and medium (an AI interview that lets people talk instead of type feels lighter and tends to complete better than a long static form). Combine all three and a feedback program that used to scrape single-digit response rates can comfortably reach the double digits — with far richer answers per response.

Related Resources

• Structured Questions Guide — the 6 question types behind this template
• Customer Satisfaction Survey Questions — a deeper question bank
• CSAT Survey Guide — designing and benchmarking CSAT
• NPS Survey Guide — running and scoring NPS
• Customer Feedback Questions — more question examples by goal
• From Survey to Conversation — why an AI interview beats a static form