Knowledge · HRD & Training

Satisfaction Survey Design · Reaction Evaluation

Training Satisfaction Survey Design

📂HRD & Training⏱2-min read📊Beginner🔗10 related

What is satisfaction survey design?

In one line
A survey that measures learners' reactions right after training. Depending on how its questions are designed, it can produce information you can act on or leave you with a score that is just a formality.
Bring this topic in-house📚 See related course💬 Ask about training
⚡ Summary in 3 lines
  • Maps to Kirkpatrick Level 1 (Reaction)
  • Ask about usefulness and intent to apply, not just satisfaction
  • Comparisons and changes tell you more than absolute scores

Survey questions are almost the only record a company keeps once a training session ends. The material pulled out at the budget review three months later, and the material the next cohort’s designer opens, is this one survey form. That is why changing the questions raises the quality of the meeting discussion. An average score of 4.4 ends with “It went fine, so we keep going as is,” but 40 responses naming “one thing to apply next week” become follow-up coaching topics on the spot. The starting point is to treat the survey not as score collection but as input for the next design.

Questions should be separated by type. Satisfaction (was it good?), usefulness (is it relevant to my work?), intent to apply (will I use it?), and likelihood to recommend (would I suggest it to a colleague?) each measure something different. Of these, usefulness and intent to apply are the ones that connect, to a degree, with later on-the-job behavior, while satisfaction scores, which swing with fun and atmosphere, connect only weakly. A 5-point scale is fine, but if responses cluster heavily at the neutral midpoint, use a 6-point scale with no middle value. Realistic upper limits are 8 to 12 multiple-choice items and under 3 minutes to complete. Response rates drop sharply when surveys go out by email, so collecting responses on site right after the session works better.

You also need to draw boundaries with the other levels. A satisfaction survey cannot measure whether learning took place. People who give 5 points to 'I understood' and people who can actually explain the procedure rarely overlap. That check belongs to pre- and post-assessments or grading of practice deliverables, while reaction evaluation is closer to explaining why the results came out as they did. Open-ended comments have the same problem: if you leave them open as 'areas for improvement,' half the answers are 'none.' You only get real sentences when you narrow the scope to the one thing you would change in the next session.

The most common misstep is using this score to manage instructors. Once it becomes known that a score below 4.5 puts a contract renewal under review, instructors drop the hard exercises, lift the mood with anecdotes, and finish ten minutes early. Scores rise while learning falls, a backfire effect. Looking only at the average is also risky. Two courses at the same 4.2 are in very different shape if one has a standard deviation of 0.3 and the other 1.2, and the latter is often a sign that participants were selected wrongly, not that the teaching was poor.

Training that covers this term · HR & Leadership
Recommended courses · Recommended instructors · Quick quote
📚AI Training Pre-Assessment and Customized Training DesignOnline assessment + 2 hours · In-person corporate training · John KimView course →📚In-House Instructor Development Program12 hours · In-person corporate training · John KimView course →
Instructor Jihye ChoiJihye ChoiExpert in design thinking, new-hire onboarding, and train-the-trainer programsInstructor John KimJohn KimAuthor of B2B Marketing Design and CEO of Nanum Consulting
🧮 Mini quote · Based on instructor fees
About KRW 1.2 million (≈ US$900)2-hour talk · up to 20 people · Korean
Get a quote for these requirements →
💡 In plain English
It is the difference between asking only whether the meal was tasty and asking whether they would come again and what you should change.
Why it matters in practice
A survey where every course scores 4.5 tells you nothing. What you ask determines what you can improve.
Origin and sources

It originates from Level 1, Reaction evaluation, of the four-level evaluation model Donald Kirkpatrick published in 1959. For response scales, the Likert scale proposed by Rensis Likert in 1932 is widely used.

Sources · Donald Kirkpatrick's four-level evaluation model (1959), Rensis Likert's scale (1932)
An example

Company A is an automotive parts manufacturer with 620 employees (a hypothetical example). It runs 32 courses a year, yet its average satisfaction score had been stuck between 4.4 and 4.7 for two years. Unable to show which courses to cut at budget review, it had copied the previous year's plan every year.

  1. It started when Kim, a manager on the training team, spread two years of surveys, 38 in all, across a single Excel sheet and said in a team meeting, "With these scores, we can't decide what to cut and what to grow."
  2. From the existing 12 questions, we dropped 5 general-impression items such as “The lecture was useful,” reorganized the rest into 8 questions across four areas (instructor, content, operations, environment), and added one open-ended question asking for one thing to apply to next week’s work.
  3. We changed the 5-point scale to 6 points to remove the midpoint and collected responses on site by QR code 10 minutes before the end. Before starting, we announced up front that “names are not recorded and results are not used for instructor contract renewal decisions.”
  4. In the three pilot courses, scores diverged for the first time. Content scored 4.9 and instructor 4.7, but operations scored only 3.6. Eleven open-ended comments clustered around two complaints: "30 minutes of practice is too short" and "I'd like the materials the day before."
  5. In the next session, with practice extended to 90 minutes and materials distributed two days ahead, the operations score rose from 3.6 to 4.4. The completion rate for the application questions went from 62% to 88%, and in the follow-up a month later, 41% of respondents said they had actually applied what they learned.
Related terms
Kirkpatrick Four-Level EvaluationA classic training evaluation model that assesses training effectiveness in four levels: reaction, learning, behavior, and results.Learning TransferApplying what was learned in training to actual work and sustaining it. The hurdle training must clear to become real results.Formative and Summative EvaluationFormative evaluation happens “during” learning to improve it; summative evaluation judges results at the “end.” Their purposes differ.Phillips ROI ModelAn evaluation model that adds a fifth level, “Return on Investment,” to Kirkpatrick’s four levels to measure the financial value training creates.Training Vendor Selection · Training Provider EvaluationThe process of comparing external providers and instructors against set criteria, then recording results afterward to inform rehiring decisions.Training Result Report · Annual Training Performance ReportA report summarizing a year of training spend and hours, delivery results, and changes on the job, submitted as the basis for deciding next year’s training budget and direction.Basic Statistics · Mean, Median, Standard DeviationBasic tools that summarize a data set in a few numbers: the mean and median show where the center is, and the standard deviation shows how widely values spread around it.CSAT · CES Customer Satisfaction MetricsCustomer survey metrics: CSAT (Customer Satisfaction Score) asks how satisfied customers were with a specific experience, and CES (Customer Effort Score) asks how much effort it took to get something done.AI Training Application Rate · Measuring AI Training EffectivenessThe share of people who received AI training and are actually using what they learned at work after a set period. It measures whether behavior changed after training, not satisfaction or completion rates.Internal Training Promotion · Training AnnouncementsDesigning announcement copy, distribution channels, timing, and manager recommendations so employees sign up for in-house courses on their own. Here the training team plays the role of a marketer.
Training linked to this term
📚AI Training Pre-Assessment and Customized Training DesignNanum Consulting corporate training · On-site/OnlineView course →
Need in-house training on this topic?
Get a curriculum and quote tailored to your needs right away.
Ask About Training
📚 View related course🗂 All programs📄 Free resources🔎 Glossary home
Last updated 13 Sep 2026 · Reviewed by John Kim

FAQ

No. It has long been pointed out that reaction evaluation does not predict learning or on-the-job application. Enjoyment and usefulness are different things, so look at intent to apply and post-training behavior checks as well.
There is no single right answer. But to compare courses within an organization, it is important to use one consistent scale. Choose an odd or even number of points depending on whether you want to allow a neutral response, and keep it once decided.
The most effective approach is to set aside time for responses on site, right after the session ends. Fewer questions, a stated completion time, and sharing in the next course how the results were used all raise participation.
Because of leniency bias. Look at rankings across courses and changes over time rather than absolute scores, and draw real insight from open-ended questions that ask what to improve.