41. Nobody clicks the offer banner
Click maps show almost no clicks on a large, colourful offer banner at the top of a page. Users later say they "never saw any offer". What is the most likely cause?
- A.The banner is too small to be seen
- B.Banner blindness: users skip ad-like areas
- C.Users do not like offers in general
- D.Click tracking is always inaccurate
Show answer
Answer: B. Banner blindness: users skip ad-like areas
People learn to ignore elements that look like ads, especially large banners in typical ad positions. Placing key information in content-like formats near the task usually works better than louder banners.
42. A result that flips by segment
Design B has a higher overall conversion than A, yet A wins on both mobile and desktop separately. How is this possible?
- A.The data must contain a calculation or tracking error
- B.Mobile users always convert better than desktop users do
- C.B got a larger share of desktop traffic, which converts better
- D.This cannot happen with real data
Show answer
Answer: C. B got a larger share of desktop traffic, which converts better
This is Simpson's paradox: when groups with different base rates are mixed in different proportions, the combined result can reverse the within-group result. Check that traffic mix is balanced and analyse by key segments.
43. Reading cohort retention
The January cohort had 1,000 sign-ups and 300 still active in week 8. The March cohort had 2,000 sign-ups and 500 active in week 8. What does this show?
- A.March retention (25%) is worse than January (30%)
- B.March retention is better because 500 is more than 300
- C.Both cohorts retained exactly the same share
- D.Retention cannot be compared across cohorts
Show answer
Answer: A. March retention (25%) is worse than January (30%)
Retention is a rate: 300 ÷ 1,000 = 30% and 500 ÷ 2,000 = 25%. Larger absolute numbers can hide a worse rate. Cohort analysis compares groups at the same age to see whether changes help or hurt.
44. Tracking that breaks on resize
A heading uses letter-spacing: 2px. When the heading grows from 24px to 64px on desktop, the tracking looks far too tight. What is the better approach?
- A.Specify letter-spacing in vw so it scales with width
- B.Specify letter-spacing in em so it scales with size
- C.Remove letter-spacing and use word-spacing
- D.Use a pixel value large enough for both sizes
Show answer
Answer: B. Specify letter-spacing in em so it scales with size
A pixel value stays fixed while the font grows, so its relative effect shrinks. em values are proportional to the current font size, so tracking stays visually consistent across breakpoints. The same applies to tracking in design tools, which is relative to size.
45. Measurement with privacy signal loss
After stricter consent banners and iOS tracking limits, your ad platforms report 35% fewer conversions, but CRM sales are flat. What is the best long-term response?
- A.Turn off the consent banner so tracking returns to normal
- B.Use browser fingerprinting to identify users without consent
- C.Strengthen first-party data: implement consent mode, server-side tagging and conversions APIs with hashed consented data, and validate with CRM data and lift tests
- D.Cut ad budgets by 35% because performance has fallen
Show answer
Answer: C. Strengthen first-party data: implement consent mode, server-side tagging and conversions APIs with hashed consented data, and validate with CRM data and lift tests
Flat CRM sales show the drop is lost measurement, not lost performance. The durable fix is privacy-safe first-party measurement: consent mode, server-side tagging, offline or API conversion uploads and modelled conversions, cross-checked with CRM and lift tests. Removing consent or fingerprinting breaks privacy laws and platform policies.
46. Defend the brand budget
The CFO wants to cut a Rs 20 lakh brand video budget to zero because it shows almost no direct conversions in the dashboard. What is the best response?
- A.Propose a measured test, such as a geo holdout or brand lift study, and track branded search volume, direct traffic and blended CAC to show its real effect
- B.Agree immediately, since only direct conversions matter
- C.Argue that brand marketing cannot be measured, so it should not be judged
- D.Move the money to retargeting, which shows the highest ROAS
Show answer
Answer: A. Propose a measured test, such as a geo holdout or brand lift study, and track branded search volume, direct traffic and blended CAC to show its real effect
Brand spend mainly creates future demand that shows up in branded search, direct visits and better conversion rates on other channels. A geo holdout or lift study turns the debate into evidence. Saying it cannot be measured loses trust, and retargeting only harvests demand that brand work created.
47. Measuring YouTube's real impact
You want to know how many extra sales a YouTube campaign really causes. Which approach gives the most reliable answer?
- A.Count conversions reported in YouTube's view-through column
- B.Compare sales this month with the same month last year
- C.Ask customers in a survey if they saw the ad
- D.A geo-holdout or conversion lift test that compares exposed and unexposed regions or groups
Show answer
Answer: D. A geo-holdout or conversion lift test that compares exposed and unexposed regions or groups
Incrementality tests create a fair control group, so the difference in sales is caused by the ads. Platform-reported view-throughs include people who would have bought anyway. Year-on-year comparisons mix in seasonality, pricing and other changes.
48. How much of cart-flow revenue is incremental?
You hold out 10% of cart abandoners from the cart flow. Holdout shoppers convert at 8%, flow recipients at 11%, with the same order value. What share of the flow's attributed revenue is incremental?
- A.About 27%
- B.About 73%
- C.About 38%
- D.100%
Show answer
Answer: A. About 27%
Of the 11% who bought after getting the flow, 8 points would have bought anyway, so only 3 points are caused by it: 3 ÷ 11 ≈ 27%. The flow is still valuable, but attributed revenue overstates its effect almost fourfold.
49. Proving the nurture programme works
Leadership asks whether the B2B nurture programme creates pipeline or just emails people who would buy anyway. What is the strongest evidence?
- A.The open rate of the nurture emails over six months
- B.Total revenue from leads who ever received a nurture email
- C.Sales team feedback collected after each quarter
- D.A random holdout of new leads, comparing opportunity rates later
Show answer
Answer: D. A random holdout of new leads, comparing opportunity rates later
Comparing randomised nurtured and non-nurtured leads on opportunities and revenue after a set period isolates the programme's effect. Totals from nurtured leads include buyers who would have bought anyway, and opens are unreliable.
50. Saturation curves in MMM
An MMM fits a saturation (diminishing returns) curve to Meta spend. The current spend is on the flat part of the curve. What does that imply?
- A.Meta has the highest average ROI of all channels
- B.Meta spend should rise until the curve turns steep
- C.Extra Meta spend will add little; marginal ROI is low
- D.Meta results have not been measured correctly
Show answer
Answer: C. Extra Meta spend will add little; marginal ROI is low
On the flat part of a saturation curve, each extra rupee adds very little response, so marginal ROI is low even if average ROI looks good. Budget decisions should be based on marginal returns, not averages.
51. A holdout that was not held out
In an email holdout test, the control group was excluded from emails but still saw the same offer through retargeting ads and SMS. What does this do to the measured email lift?
- A.Overstates it, as control saw fewer messages overall
- B.Understates it, as control was nudged by the same offer
- C.Has no effect, as only email is being measured
- D.Doubles it, as two channels are now counted
Show answer
Answer: B. Understates it, as control was nudged by the same offer
If the control group gets the same offer through other channels, its conversion rate rises and the gap to the test group shrinks. The test then underestimates email's effect alone. Keep other channels consistent across groups or measure the combined programme.
52. iOS app install measurement
For iOS users who decline App Tracking Transparency, how do ad networks receive install attribution?
- A.Through the IDFA, which is still shared at install time
- B.Through GA4 user-level data shared with the ad network
- C.They get nothing, since iOS installs cannot be measured
- D.Through Apple's delayed, aggregated and private postbacks
Show answer
Answer: D. Through Apple's delayed, aggregated and private postbacks
Apple's SKAdNetwork, and its successor AdAttributionKit, sends postbacks that are delayed, coarse and limited in detail, with no user-level ID. Teams combine them with modelled data and incrementality tests to judge iOS campaigns.
53. Profit-aware conversion values
Your store sells phones (5% margin) and phone cases (60% margin). You bid on revenue ROAS. What is the risk and the fix?
- A.Bidding favours cases; send fewer conversions
- B.Bidding ignores both; switch to manual CPC only
- C.Bidding favours low-margin phones; send margin-based values
- D.Bidding is fine; raise the ROAS target instead
Show answer
Answer: C. Bidding favours low-margin phones; send margin-based values
A ₹30,000 phone looks far more valuable than a ₹1,000 case, even though the profit is ₹1,500 vs ₹600 and the case may be much cheaper to win. Sending profit or margin-adjusted values lets value-based bidding optimise for what the business keeps.
54. Modelled conversions after consent
After advanced Consent Mode goes live for EEA traffic, Google Ads shows conversions with a note that some are modelled. What does that mean?
- A.Google copies conversions from other advertisers in the same sector
- B.Google counts only users who accepted and adds nothing
- C.Google moves conversions from GA4 into the Ads account
- D.Google estimates conversions from declined users using cookieless pings
Show answer
Answer: D. Google estimates conversions from declined users using cookieless pings
When users decline, tags send cookieless pings. Google uses patterns from consented users to estimate conversions for unconsented ones, and the estimate appears in reports and feeds bidding. It is an estimate, so validate trends against backend sales.
55. Three methods, three answers
Platform data says Meta drove 900 sales, your MMM says 500, and a lift test says 450 in the tested period. How should a senior marketer use these?
- A.Treat the test as causal truth and calibrate the others to it
- B.Pick the highest number to protect the budget
- C.Average the three numbers and report 617
- D.Ignore all three and rely only on last-click GA4
Show answer
Answer: A. Treat the test as causal truth and calibrate the others to it
Each method has a role: lift tests measure causation for a period, MMM covers all channels over time, and platform data is fast for daily optimisation. A common practice is to compute a ratio such as 450 ÷ 900 = 0.5 and apply it to platform numbers until the next test.
56. Last week looks terrible
A B2B campaign's cost per conversion looks twice as high for the last 7 days as for earlier weeks, yet nothing changed. Leads usually convert 10–20 days after the first click. What is happening?
- A.The platform has started charging more per click
- B.The audience became saturated in exactly one week
- C.Conversions older than 7 days were deleted
- D.Conversion lag: recent clicks have not had time to convert yet
Show answer
Answer: D. Conversion lag: recent clicks have not had time to convert yet
Platforms credit conversions to the date of the click, so recent days fill in later as delayed conversions arrive. Judging fresh data too early leads to pausing good campaigns. Compare periods only after the usual lag has passed.
57. Measuring click-to-WhatsApp ads
A Dubai car rental firm runs Meta click-to-WhatsApp ads. Ads Manager shows many "messaging conversations started", but the owner cannot tell which ads bring bookings. What fixes this?
- A.Count every conversation started as one booking made
- B.Switch to the Awareness objective to get more chats
- C.Tag chat leads and bookings and send them back as events
- D.Turn off WhatsApp and send all traffic to the website
Show answer
Answer: C. Tag chat leads and bookings and send them back as events
Conversations started only shows that a chat opened. Tagging chats in the WhatsApp Business platform or CRM, and sending lead or purchase events back (for example through the Conversions API), links ads to real outcomes. Meta can then optimise for chats that turn into bookings.
58. One conversion set for many accounts
A UAE retail group runs separate Google Ads accounts for its Dubai, Abu Dhabi and Riyadh stores, all selling on one website. Each account has its own purchase tag, and counts overlap. What set-up is cleaner?
- A.A separate website for each Google Ads account
- B.Cross-account conversion tracking from the manager account
- C.Copying the same tag into each account and adding them up
- D.Turning off conversion tracking in two accounts
Show answer
Answer: B. Cross-account conversion tracking from the manager account
With cross-account conversion tracking, conversion actions are created in the manager account and shared with client accounts, so one tag and one definition serve them all. Each conversion is credited to the account whose ad drove it. It avoids duplicate tags and mismatched definitions.
59. UTM tags on internal links
The site's internal banners link to other pages with ?utm_source=homepage parameters. What problems does this cause?
- A.None, because Google ignores all parameters
- B.It improves rankings by adding keywords to the URL
- C.It creates duplicate URL versions that split signals and crawling, and it overwrites the visitor's real traffic source in analytics
- D.It only affects paid search reports
Show answer
Answer: C. It creates duplicate URL versions that split signals and crawling, and it overwrites the visitor's real traffic source in analytics
Each parameter version is a separate URL that Google may crawl and must canonicalise. In GA4, a new UTM starts a new attributed source, so organic visits look like 'homepage' traffic. Use clean internal links and track internal promotions with events instead.
60. Content and last-click attribution
Your blog shows almost no conversions in a last-click report, and leadership wants to cut it. Many buyers read posts weeks before converting through branded search. What is the best argument?
- A.Last-click gives all credit to the final touch, so look at assisted conversions, data-driven attribution and paths that include the blog before deciding
- B.Blogs are for brand only and should never be measured
- C.Switch the report to first-click so the blog gets all the credit
- D.Remove branded search from the reports
Show answer
Answer: A. Last-click gives all credit to the final touch, so look at assisted conversions, data-driven attribution and paths that include the blog before deciding
Top-of-funnel content usually influences buyers early, and last-click attribution ignores that. Comparing conversion paths and data-driven or position-based models shows a fairer picture. Swapping to first-click just creates the opposite bias.
