Most survey problems are not about technology. They happen before anyone opens a link: unclear objectives, poorly worded questions, or no plan for what to do with the data. This guide covers the full process, from defining what you actually want to learn, through question design, distribution, and analysis using AhaSlides.
Why run surveys online at all?
Before getting into the how, it's worth being clear about the trade-offs.
Online surveys eliminate printing and distribution costs and let you reach a distributed workforce instantly. Results arrive in a dashboard rather than a spreadsheet you have to build by hand. For most HR and L&D use cases: training evaluations, pulse checks, post-event feedback, that operational simplicity is reason enough to go digital.
The response rate trade-off is real though. Research shows online surveys average around 33–44% completion, compared to 56–75% for paper-based equivalents [1]. That gap matters most for population-level research. For internal surveys where you control the context (end of a training session, company all-hands, team meeting), response rates tend to be much higher because participants are already in the room and engaged.
The global online survey software market was valued at around $3.6 billion in 2023, and more than 78% of enterprises now use these platforms for employee feedback and market research [2]. The tools have caught up with the workflow.
Step 1: Define your objective first
The most common survey mistake is starting with questions instead of a goal. Before opening any tool, answer three things:
A training evaluation survey where the goal is "understand whether participants can apply the content on the job" produces very different questions than one aimed at "check whether people enjoyed the session." Both are legitimate, but conflating them produces a survey that answers neither cleanly.
Step 2: Set up your AhaSlides account
Go to ahaslides.com and create a free account. From the dashboard, go to 'Survey'.
Step 3: Design your questions
AhaSlides Survey supports several question types suited to different data needs:
Multiple choice works for close-ended questions with a defined answer set: department, role, attendance at a previous session.
Rating scales are the right format for attitude and satisfaction data. You set the number of points and the labels. Use a 5-point scale for most professional contexts; move to 7-point when you need to detect finer differences between groups or track change over time.
Open-ended questions let respondents write free-text answers. Useful for follow-up questions ("What would you change about this training?") where you want qualitative input rather than numbers.
Matrix questions group similar statements into a single table so you don't need to clone the questions.
Step 4: Distribute the survey
AhaSlides generates a shareable link and a join code. For live sessions, go to your presentation and add a Survey slide. Link your Survey here. When you present on AhaSlides and move to this slide, the audience will be able to see the questions all in one page.
For async distribution, paste the link into Slack, email, or your LMS.
Step 6: Analyze and act on results
AhaSlides shows results in real time as a visual dashboard. For live sessions, you can share the screen and walk through results with participants while context is fresh.
For deeper analysis, export responses to Excel. From there you can cross-tabulate by department or role, calculate average scores across cohorts, or run the data through an AI tool for trend identification across large open-text responses.
Build the analysis step into your workflow before you run the survey, not after. If there is no plan for what happens when results arrive, the data sits in a dashboard that nobody revisits.
One practical approach: before distributing the survey, write down the two or three questions you want the data to answer. Then, once results are in, map your findings directly to those questions. This keeps analysis focused and makes it easier to communicate results to stakeholders in plain language rather than sending around a raw export that most people will not open.
Practical tips for getting better data
A well-designed survey can still produce thin or unreliable results if the surrounding context is not right. These tips apply regardless of which platform you use.
Time the survey deliberately. Ask for feedback when the experience is still fresh. A training evaluation sent three days after a session competes with everything else in a participant's inbox and memory. Running it in the final 10 minutes of the session, or immediately after, produces more accurate recall and consistently higher completion rates.
Write a clear introduction. Before the first question, tell participants how long the survey takes, whether responses are anonymous, and what the data will be used for. One or two sentences is enough. Participants who understand the purpose are more likely to give thoughtful answers rather than rushing through to finish.
Match the question type to the data you need. Rating scales are good for comparing across groups or tracking change over time. Open-ended questions surface context and nuance. Multiple choice gives you clean counts. A common mistake is using open-ended questions for everything because they feel thorough, when a rating scale would actually produce more comparable, usable data across a cohort.
Keep the language simple. Write questions at the reading level of the widest range of respondents, not the most senior. Avoid jargon, acronyms, and double negatives. "Did the session not fail to meet your expectations?" is a real type of question that turns up in corporate surveys and produces data nobody can interpret. Read each question aloud before publishing it. If you pause or stumble, rewrite it.
Send a reminder for async surveys. Response rates for async surveys typically improve by 10–20 percentage points with a single follow-up reminder sent 48 hours before the deadline. Keep the reminder short and restate the deadline clearly. Do not send more than one reminder — it creates resentment that carries over to future surveys.
Common mistakes to fix before you launch
Skipping a pilot. Send the survey to one or two colleagues first. They will catch ambiguous wording, broken links, and questions that produce unusable data — things you will not notice yourself after staring at the draft for an hour.
Too many open-ended questions. One or two is useful. Six is a burden. Most respondents will abandon a survey that reads like a written exam.
No follow-through communication. If people complete a survey and never hear what happened with the results, response rates on the next survey drop. Close the loop with a short summary: "Here's what we heard, here's what we're changing."
Frequently asked questions
How many questions should an online survey have?
For live sessions where participants are answering in real time, aim for 5 to 8 questions. That range typically takes 3 to 5 minutes and keeps completion near 100% when people are already engaged in the context. For async surveys sent by link, 10 to 12 questions is a reasonable ceiling. Research on survey fatigue consistently shows that completion rates drop after about 10 minutes of response time, so the practical limit is less about question count and more about how long it actually takes to answer. Time your own survey before distributing it.
Should surveys be anonymous?
It depends on the topic. For questions about management effectiveness, workplace culture, compensation, or anything where respondents might worry about being identified, anonymous collection produces more honest answers. For training evaluations where you need to tie results to individual completion records or follow up with specific people, named responses are appropriate. The key is being transparent: if you tell participants a survey is anonymous, it needs to actually be anonymous. Trust, once broken on that point, is difficult to rebuild and will affect every survey you run afterward.
What should I do with open-text responses?
Start by reading a sample of 15 to 20 responses to get a feel for the range of answers. Then group responses into themes manually or use a spreadsheet or AI tool to tag recurring ideas. For small surveys (under 50 responses), manual coding is faster and more accurate than any automated approach. For larger datasets, tools like ChatGPT or Claude can summarize themes from a pasted block of text in a few seconds. Either way, open-text data is most useful when you can anchor it to something: "40% of open responses mentioned pacing as an issue" is more actionable than a wall of quotes.
Running surveys with AhaSlides
AhaSlides works for both live facilitated surveys and async ones sent in advance or after an event. Rating scales, multiple choice, matrix, and open-ended questions are all available.
For L&D teams running regular training evaluations, the combination of fast setup, live results, and Excel export covers most of what you need without requiring a dedicated survey platform.
Sources
[1] Nulty, D.D. (2008). "The adequacy of response rates to online and paper surveys." Assessment & Evaluation in Higher Education. University of Alaska Fairbanks PDF — meta-analysis showing online surveys average 33% vs 56% for paper-based equivalents.
[2] Grand View Research. Online Survey Software Market Size & Share Report, 2030. https://www.grandviewresearch.com/industry-analysis/online-survey-software-market-report — market sizing and enterprise adoption data.
