Ask any director considering digital assessment for their biggest worry, and the answer is rarely about features. It is: "Will the results still mean something?"
It is the right question. A digital test that students can game is worse than paper, because it produces confident-looking numbers that are wrong. But the response to that fear is often a pile of controls chosen for how reassuring they sound rather than what they do. Let us separate the two.
Measures that actually work
A time limit calibrated to the task. The single most effective integrity measure, and the simplest. If your test is designed so a prepared student needs most of the allotted time, there is little room left for searching answers elsewhere. Generous time limits, not screen locks, are what make cheating practical.
Question and option shuffling. When every student sees questions in a different order, with answer options shuffled too, shoulder-surfing and message-group coordination lose most of their value. This costs students nothing and is invisible when done right.
Question pools. Writing three variants of each question and serving a random one per student means two students sitting side by side take genuinely different tests. This is more preparation work, but it compounds: the pool you build this year serves every future session.
One submission, visible audit trail. Recording when the test started, when it was submitted, and (where relevant) answer timing gives you evidence when something looks wrong, instead of suspicion.
Proportionate stakes. The strongest measure is not technical. Frequent low-stakes assessment reduces the payoff of cheating on any single test. A student facing a weekly 10-minute quiz has little reason to orchestrate fraud; a student facing one exam worth 100 percent of the grade has every reason.
Measures that are mostly theater
Blocking copy-paste and right-click. Defeated by a phone sitting next to the keyboard. Annoys legitimate students who want to re-read their own answer.
Full browser lockdown for low-stakes tests. Lockdown browsers have a place in formal certification. For a weekly quiz, they add installation friction and IT tickets while the actual risk they address is minimal.
Webcam proctoring everywhere. Beyond the privacy and consent questions, always-on proctoring signals distrust to every student to deter a small minority. Reserve it for the rare assessments where stakes genuinely justify it.
The pattern: controls that change the test itself (timing, shuffling, pools) work quietly and scale. Controls that try to control the student's environment are expensive, invasive, and easiest to defeat.
The mindset shift
Paper exams were never as secure as memory suggests. Answers were whispered, notes were smuggled, and graders were inconsistent in ways nobody measured. Digital assessment does not need to achieve perfect integrity. It needs to be at least as trustworthy as what it replaces, while being faster, fairer, and visible.
Design the test so cheating is hard and low-value. Keep evidence so rare cases can be handled with facts. And spend the energy you save on what actually improves results: better questions.