41. A pre-ticked consent box
Marketing wants the "Send me offers" checkbox ticked by default at sign-up for users in the EU. What is the right response?
- A.Agree; pre-ticked boxes are standard practice
- B.Explain that under GDPR consent must be an active, informed choice, so the box must start unticked; offer to improve the value proposition copy instead
- C.Hide the checkbox so nobody sees it
- D.Tick it by default but make the text very small
Show answer
Answer: B. Explain that under GDPR consent must be an active, informed choice, so the box must start unticked; offer to improve the value proposition copy instead
EU law and court rulings say pre-ticked boxes do not count as valid consent. Beyond legal risk, tricking people into lists raises unsubscribes and complaints. Clear, honest benefits are the ethical way to increase opt-in.
42. Limits of heuristic evaluation
With no time for user testing, a lead suggests one designer does a heuristic evaluation instead. What is the accurate view?
- A.Useful and cheap, but one evaluator finds only a fraction of issues; use 3 to 5 evaluators and treat results as expert judgement that still needs validation with real users
- B.It finds every problem real users would find
- C.It is useless and should never be done
- D.It is the same thing as a usability test
Show answer
Answer: A. Useful and cheap, but one evaluator finds only a fraction of issues; use 3 to 5 evaluators and treat results as expert judgement that still needs validation with real users
Different experts spot different problems, so combining several evaluators improves coverage. Experts also miss domain-specific issues and can flag things users do not mind. It is a complement, not a replacement, for testing.
43. Swipe-only actions
In a mobile mail app, delete and archive are available only by swiping a message left or right. Testing shows many users never find them. What is the best change?
- A.Add an onboarding tutorial and keep swipe as the only way
- B.Increase the swipe distance needed
- C.Remove the actions completely
- D.Keep swipe as a shortcut but also offer visible ways, such as buttons on the open message or a long-press menu, and make sure screen reader users can reach the actions
Show answer
Answer: D. Keep swipe as a shortcut but also offer visible ways, such as buttons on the open message or a long-press menu, and make sure screen reader users can reach the actions
Hidden gestures lack signifiers and are not discoverable or accessible on their own. Tutorials are often skipped and forgotten. Gestures work best as accelerators on top of visible controls.
44. One category fails in a tree test
In a tree test, 80% of users succeed on most tasks, but only 25% find "Download Form 16", most going to "Documents" instead of "Payroll". What should you do?
- A.Move it or cross-link it under Documents, then re-test
- B.Keep it under Payroll since that is correct
- C.Remove Form 16 from the app
- D.Rename Payroll to Documents
Show answer
Answer: A. Move it or cross-link it under Documents, then re-test
When most users look in a different place, their mental model is the guide. Place the item where users expect, or list it in both places, and re-test to confirm success improves.
45. Browser back inside a modal
In a web app, a user opens a large modal with a three-step form, then presses the browser back button. The whole page navigates away and the form is lost. What is a better approach?
- A.Disable the browser back button with script
- B.Make back close the step or modal, keeping data
- C.Warn users that back is not supported here
- D.Reload the whole page when back is pressed
Show answer
Answer: B. Make back close the step or modal, keeping data
Users treat big overlays like pages and expect back to undo the last step. Pushing history entries for the modal and its steps, and keeping entered data, matches that model. Disabling back breaks a core browser control.
46. Designing all upload states
You are designing a document upload for KYC. Which set of states is most complete?
- A.Idle, uploading and success, with no other states
- B.Idle, uploading, success, failed with retry, cancelled
- C.Uploading, success and archived, with no idle state
- D.Idle, success, failed and archived, without progress
Show answer
Answer: B. Idle, uploading, success, failed with retry, cancelled
Mapping upload as a state machine makes sure each state has clear feedback and actions: progress and cancel while uploading, a retry and reason when it fails, and confirmation when done. Missing states lead to dead ends.
47. Calculating lostness
Lostness L = √((N/S − 1)² + (R/N − 1)²), where S = total pages visited, N = distinct pages and R = minimum pages needed. A user visits 10 pages (8 distinct) on a task needing 4. What is L, and what does it suggest?
- A.About 0.54; the user was likely lost
- B.About 0.20; navigation was efficient
- C.About 0.80; the task failed
- D.Exactly 0; the user took the best path
Show answer
Answer: A. About 0.54; the user was likely lost
N/S = 0.8, so (−0.2)² = 0.04. R/N = 0.5, so (−0.5)² = 0.25. L = √0.29 ≈ 0.54. Studies suggest values above about 0.5 indicate users who look lost, while values below about 0.4 look comfortable.
48. Comparing benchmarks across releases
Last year's benchmark showed 65% success on "change mobile number". This year it is 82%, but tasks were reworded and participants came from a different panel. What is the problem?
- A.An 82% score is always a real improvement over 65%
- B.Benchmarks should never be repeated a second time
- C.Changed tasks and samples make the scores not comparable
- D.The old score must have been wrong
Show answer
Answer: C. Changed tasks and samples make the scores not comparable
Benchmarks only compare well when tasks, success criteria, recruiting profile and protocol stay the same. Otherwise the change may come from the method, not the design. Keep a fixed benchmark protocol.
49. Interviewing users who missed loan payments
A lending app wants to interview customers who recently missed EMI payments. Which plan shows the most care?
- A.Have the collections team phone them to schedule the research sessions
- B.Offer to waive late fees if they take part
- C.Record the sessions and share them with the credit risk team
- D.Make it opt-in, let them skip questions, and keep it apart from collections
Show answer
Answer: D. Make it opt-in, let them skip questions, and keep it apart from collections
Participants in financial stress are vulnerable. Separating research from collections avoids pressure, opt-in and skip rights protect them, and a fair fixed incentive avoids coercion. Linking participation to fees or sharing with risk teams would make consent not truly free.
50. Recruiting from app store reviewers
To recruit quickly, a team invites only people who left app store reviews in the past month. What bias does this sample most likely carry?
- A.It over-represents first-time users who never opened the app
- B.It over-represents people with strong, often negative, feelings
- C.It under-represents people with strong opinions
- D.It has no bias because reviewers are real users
Show answer
Answer: B. It over-represents people with strong, often negative, feelings
People who write reviews are self-selected and usually feel strongly, often after a bad experience. Their views matter, but they do not represent the quiet majority. Mixing recruitment sources gives a more balanced view.
51. Testing an AI assistant before building it
You want to learn how users would talk to a tax-filing assistant before investing in a model. Which approach fits best?
- A.A survey asking users whether they like chatbots
- B.Wizard of Oz: a hidden person writes the assistant's replies
- C.Launching the real model to all users at once
- D.A card sort of possible tax topics
Show answer
Answer: B. Wizard of Oz: a hidden person writes the assistant's replies
In a Wizard of Oz study a person simulates the system behind the scenes. It reveals what users ask, how they phrase things and what answers they trust, before costly building. Participants should be debriefed afterwards about the simulation.
52. Quote marks that dent the left edge
A large pull quote starts with a curly quote mark, and the text edge looks indented on the first line. What is the typographic fix?
- A.Hang the punctuation outside the text edge
- B.Remove all quote marks from the pull quote
- C.Indent every line to match the first
- D.Centre the whole pull quote
Show answer
Answer: A. Hang the punctuation outside the text edge
Hanging punctuation places the opening quote mark in the margin so the letters form a clean left edge. It is optical alignment: the letters line up even though the bounding boxes do not.
53. Stripes in a smooth gradient
A large dark blue background gradient shows visible stripes (banding) on some screens. What is a common fix?
- A.Use fewer colour stops at sharper angles
- B.Save the background as a low-quality JPG
- C.Add subtle noise or dithering to the gradient
- D.Increase the contrast between the stops
Show answer
Answer: C. Add subtle noise or dithering to the gradient
Banding happens when a gradient spans a small colour range over many pixels, so 8-bit colour steps become visible. A little noise breaks up the step edges. Wider colour ranges and avoiding heavy compression also help.
54. Cramming everything above the fold
A stakeholder asks for all eight product benefits above the fold because "nobody scrolls". What does research suggest, and what is the better layout?
- A.Nobody scrolls, so shrink everything to fit
- B.People scroll if the top promises more below
- C.Hide all benefits in a hamburger menu
- D.Split the page into eight separate tabs
Show answer
Answer: B. People scroll if the top promises more below
Studies show people scroll when the top of the page promises valuable content, though attention is highest near the top. A strong opening with one clear message and visual cues to continue beats cramming everything into the first screen.
55. Remote calls or field visits
You are researching how small kirana shops reorder stock. A colleague suggests video calls only, to save time. What is the strongest reason to do some in-person visits?
- A.Video calls are not considered a valid research method when studying small retail shops in India
- B.Shop owners nearly always give false answers on a video call
- C.In-person visits are usually cheaper and faster than video calls
- D.You see the real setting: interruptions, shelves, paper notes and the phones they use
Show answer
Answer: D. You see the real setting: interruptions, shelves, paper notes and the phones they use
For a physical, interrupted task like reordering stock, context matters: the notebook under the counter, the distributor who calls, the shared phone. Remote calls are efficient for many topics, but they miss the environment, which is often where the key insight sits.
56. A redesign already in flight
A big app redesign is half built with no user research. Launch is in eight weeks. What is the best way to reduce risk now?
- A.Stop the project at once and restart it after six months of fresh user research
- B.Launch on time to everyone and fix issues after complaints
- C.Run quick usability tests on the riskiest flows and stage the rollout
- D.Ask the design team to review their own work again
Show answer
Answer: C. Run quick usability tests on the riskiest flows and stage the rollout
You cannot redo discovery, but you can still test. Fast usability tests on the most used or riskiest flows catch major problems, and a staged rollout (for example, 5% of users first) limits damage while you measure real behaviour.
57. Only unhappy users replied
A satisfaction survey for a new HR portal got a 9% response rate. Most replies are complaints. What is the risk in using these results to set requirements?
- A.Non-response bias: happy users may not reply
- B.The sample is too large to be useful
- C.The survey must have had leading questions
- D.Complaints are never useful for requirements
Show answer
Answer: A. Non-response bias: happy users may not reply
When few people respond, those with strong views, often negative, are over-represented. The complaints may be real, but their scale is unclear. Comparing with usage data, a short targeted follow-up or interviews with a random sample gives a fairer picture.
58. A free trial that is hard to cancel
A streaming client wants a 'free 7-day trial' that needs a card, auto-renews at AED 59 a month, and can be cancelled only by phone during office hours. What should you recommend?
- A.Show the renewal terms clearly and allow easy online cancellation
- B.Keep it; customers agreed to the terms by entering their card
- C.Keep phone-only cancellation but add a chatbot that delays users
- D.Hide the renewal price until after the trial has ended
Show answer
Answer: A. Show the renewal terms clearly and allow easy online cancellation
Making cancellation much harder than sign-up is a 'subscription trap' dark pattern, restricted in India's dark pattern guidelines, UK consumer law and many US state auto-renewal laws. Clear terms, renewal reminders and easy cancellation build trust and reduce chargebacks and complaints.
59. "No thanks, I hate saving money"
A discount pop-up's decline button reads "No thanks, I prefer paying full price". What is wrong with it?
- A.It shames users into agreeing, a dark pattern that erodes trust
- B.It is too short to be understood
- C.It should be in all capital letters
- D.Nothing; it is a proven, ethical way to raise conversions
Show answer
Answer: A. It shames users into agreeing, a dark pattern that erodes trust
Confirmshaming guilt-trips people for declining. It may lift short-term clicks but annoys users and harms trust. Regulators also target dark patterns; for example, India's 2023 guidelines on dark patterns list confirm shaming. Use a neutral "No thanks".
60. Copy for an AI-generated summary
Your app adds AI-written summaries of long contracts. What should the interface copy make clear?
- A.That the summary is legally binding, complete and final
- B.Nothing at all, so that users trust the new feature more
- C.That it's AI-generated, may have errors, and how to check the source
- D.That a human lawyer is writing each summary in real time
Show answer
Answer: C. That it's AI-generated, may have errors, and how to check the source
Users should know when content is machine-generated and how much to rely on it. A short label plus an easy way to see the original text supports trust and safety. Overclaiming accuracy can cause real harm, especially in legal and financial tasks.
