Surveys are a great way to hear what your customers think — but they often get bad press due to poor implementation and design. Done well, they're one of the most powerful tools in your analytics stack. Here's how to get started.
At their core, surveys do something that no amount of behavioural analytics can: they tell you the why behind the what. Web analytics can tell you that 70% of users abandon your checkout on step three. A well-designed survey can tell you they left because they couldn't find a delivery option that suited them. That's the gap surveys close.
The most common mistakes
Surveys fail almost always for the same handful of reasons: they're too long, they ask leading questions, they fire at the wrong moment (a pop-up the instant someone lands on a page), or the results are collected but never acted upon. None of these are hard to fix — they just require a bit of upfront discipline.
✓ Do
- Keep it short — 3 to 5 questions maximum for on-site surveys
- Trigger on exit intent or after a key action (purchase, abandonment)
- Use a mix of rated and open-text questions
- Pilot the survey internally before going live
- Close the loop — act on what you find
✗ Don't
- Fire the survey on page load — it's intrusive and misleading
- Ask leading questions ("How great was your experience?")
- Use double-barrelled questions ("Was delivery fast and accurate?")
- Collect months of data before analysing
- Ignore open-text responses — they contain your best insight
Free tools vs. proper implementation
Free survey tools — Google Forms, SurveyMonkey's free tier — are fine for a quick temperature check. But they come with real limitations: no targeting by user segment, no integration with your analytics data, no ability to suppress the survey for users who've already responded, and very limited analysis beyond basic charts.
The moment you want to do something more sophisticated — say, show a specific survey only to users who've visited three or more times, or cross-reference responses with transaction data — you need a proper platform and a bit of custom implementation. This is where most organisations reach a ceiling with free tools, and where a structured Voice of Customer programme starts to earn its keep.
Surveys should be part of a wider customer feedback mechanism — not a standalone exercise. On their own, they tell you what people say. Combined with behavioural data, they tell you what people do and why.
The key integration pointWhat to measure
Start simple. For most organisations, three questions cover 80% of what you need to know at the start of a programme:
1. Task completion: "Were you able to complete what you came to do today?" (Yes / Partially / No) — this is your headline satisfaction proxy.
2. Ease: "How easy was it to do what you needed?" (rated 1–5) — this correlates with effort and identifies friction.
3. Open comment: "Is there anything we could do to improve your experience?" — this is where the gold is.
Once you have a baseline, layer in an NPS question and begin to track trends over time. After six months of consistent data collection, you'll have something genuinely useful to act on.