Can a business reduce form spam without putting customers off?
Measure how many genuine enquiries come through alongside how much spam your protection blocks.
Read the key findingsTL;DR
Some CAPTCHA challenges created accessibility barriers. In a 2022 survey task, participants with visual impairments had more failures with an interactive checkbox CAPTCHA than participants without visual impairment.
See what the accessibility study measuredChoose a check that asks as little as possible of customers. Cloudflare describes managed, non-interactive and invisible modes for Turnstile. Try the mode you choose on your real form and record who can complete it.
Review the documented widget modesMeasure both spam and real enquiries. Track unwanted submissions, false positives and completed legitimate forms before and after changing protection.
See what a fair comparison includes
- Participants reporting visual impairment0.31
- Participants without visual impairment0.12
Ombretta Gaggi studied reCAPTCHA in a survey with 479 participants. The comparison is specific to the checkbox challenge and the groups in that survey.
See how it was measuredProtect the form and keep the next step usable
A form that receives spam wastes staff time. A challenge that real customers cannot complete can block quote requests and appointment enquiries. The available evidence supports checking both sides of that trade-off rather than assuming any one tool is friction-free.
A study of Google reCAPTCHA shows how an interactive check can affect different users. Cloudflare documents other ways to run checks. Use a test on your own form to choose a setup.
Accessibility survey, 479 participants
The checkbox caused more failures for participants with visual impairments
In this study flow, the author reported greater failure burden for visually impaired participants on the checkbox challenge: an average of 0.31 failures per user compared with 0.12 for participants without visual impairment on the reported page.
- Participants reporting visual impairment0.31
- Participants without visual impairment0.12
The survey recruited through disability associations and social media. It included 194 respondents reporting visual impairments. In the same study flow, 99.42% passed the reCAPTCHA v3 check, which ran without a task for participants to complete.
If your form uses a challenge, try it with a keyboard and a screen reader as well as on a phone. Keep a way to reach your business visible when verification fails.
Ombretta Gaggi, “A Study on Accessibility of Google reCAPTCHA Systems,” OASIS 2022, 28 June 2022. Read Gaggi’s accessibility study
Product documentation, accessed 21 November 2026
Choose a mode, then measure its effects on the real form
Cloudflare documents managed, non-interactive and invisible modes. In managed mode the service decides whether interaction is needed; if a prompt is required, the documented interaction is a checkbox rather than image or text recognition.
These product descriptions do not show how many spam submissions a particular business will stop, how many genuine visitors it might reject, or whether form completion will improve. Invisible mode also does not mean no data processing or no privacy obligations; Cloudflare’s documentation calls for the Turnstile Privacy Addendum to be referenced in the privacy policy when using invisible mode.
Before and after a change, record spam submissions, legitimate enquiries received, form starts and successful submissions, and any reports of verification failure. Review false positives and completion by device and assistive-technology testing. Keep the form’s labels, error messages and a human contact route usable if verification cannot be completed.
Cloudflare, Turnstile widget concepts and modes (accessed 21 November 2026). Review Cloudflare’s Turnstile mode guidance
Use the lightest protection that works for your form.
Record how the current form performs, then compare genuine submissions and spam after a change.
- Record a baseline for spam, legitimate form submissions and completed enquiries before changing verification.
- Test the chosen configuration on mobile, by keyboard, and with a screen reader. Make the error state and alternative contact route clear.
- Review false positives and successful submissions after launch; adjust or remove the challenge if it blocks the people you want to hear from.
Keep tracking whether customers can complete the enquiry you want them to send.
Explore website sales optimisation