1. Spotify's "Because you liked…" labeling serves what strategic function in AI personalization?
✓ Correct — Correct. Spotify's explanation interface solves the transparency condition: by making the recommendation's reasoning legible, it shifts the perceived relationship from surveillance to helpful curation — the same data, experienced differently.
The "Because you liked…" label addresses the transparency condition of the trust architecture. By explaining its reasoning, Spotify frames the recommendation as a knowledgeable suggestion rather than a data-surveillance output.
2. Core Web Vitals became official Google ranking signals in:
Correct. The Page Experience update in May 2021 introduced LCP, CLS, and FID (replaced by INP in March 2024) as ranking signals.
Core Web Vitals were announced in 2020 but became ranking signals via the Page Experience update in May 2021.
3. Meta's Robyn MMM tool uses which optimization algorithm to automatically select the best model from thousands of candidates?
Correct. Robyn uses Ridge regression (for regularization against multicollinearity) combined with Nevergrad evolutionary optimization (to search the space of possible adstock and saturation curve parameters efficiently).
Robyn uses Ridge regression with Nevergrad's evolutionary optimization. Google Meridian uses Hamiltonian Monte Carlo for Bayesian inference — they are different tools with different approaches.
4. GDPR Article 22 on automated decision-making requires companies to provide what to affected individuals?
✓ Correct — Correct — Article 22 requires meaningful explanation and human review rights, not full technical disclosure.
Not correct — Article 22 requires interpretable explanation of key factors and the right to human review, not technical code disclosure.
5. In a "U-shaped" or position-based attribution model, how is credit distributed?
Correct. U-shaped attribution recognizes both the awareness-generating first touch and the conversion-driving last touch as high-value moments.
U-shaped attribution gives 40% to first touch, 40% to last touch, and 20% to middle touchpoints.
6. Netflix's internal 2012 segmentation analysis found that what the survey had called "drama viewers" actually contained approximately how many distinct taste clusters?
Correct. Over 2,000 distinct taste clusters — defined by behavioral signals, not demographics.
The documented figure is 2,000+ clusters identified by machine learning on behavioral data.
7. What was the approximate revenue headwind Meta cited from Apple's iOS 14.5 ATT update in their 2021 earnings reporting?
Correct. Meta cited a $10 billion revenue headwind for 2021 from ATT-related signal loss, reflecting how central individual-level cross-app attribution was to their advertising value proposition.
Meta reported a $10 billion revenue headwind in 2021 from ATT-related signal loss.
8. Ahrefs' content decay monitoring feature, launched in 2023, primarily helps SEO teams by:
Correct. Content decay monitoring transforms a reactive task (noticing traffic dropped) into proactive triage — AI flags declining pages before losses compound.
Ahrefs' decay feature monitors existing content performance and surfaces refresh candidates. It doesn't auto-rewrite content or predict future topic demand.
9. What is an "adstock" parameter in MMM, and why does it matter for budget decisions?
Correct. Adstock decay rates determine how long an ad's effect persists. Channels with high adstock (like TV) show benefits long after spending stops. Misspecifying adstock leads to under- or over-valuing channels in MMM results.
Adstock is the carryover effect — how advertising effects decay (or persist) over time after the initial impression. Accurately modeling adstock is essential for attributing credit to channels correctly across time periods.
10. California's BOT Disclosure Law applies to which specific category of communications?
✓ Correct — Correct — it is specifically about undisclosed bot-to-consumer commercial communications.
Not correct — the law covers commercial bot communications that don't disclose their automated nature to California residents.
11. Content-based filtering's primary limitation compared to collaborative filtering is:
Correct. Content-based filtering creates a filter bubble — if you liked thriller novels, you keep seeing thriller novels. It's good at relevance, weak at discovery. Collaborative filtering can surface things you didn't know you'd like.
Not quite. Content-based filtering's core limitation is over-specialization: it finds more of what you already like but can't surface surprising discoveries. Collaborative filtering ("users like you also liked") enables serendipitous recommendations.
12. What was The Washington Post's Heliograf AI system specifically designed to cover in 2016?
Correct. Heliograf handled structured, templated content — freeing human journalists for narrative and investigative work requiring judgment.
Heliograf covered structured data stories — sports results, election scorecards. Investigative journalism remained exclusively human.
13. Google's "(not provided)" problem, introduced when SSL encryption became standard in 2013, affected SEO measurement by:
Correct. SSL encryption passes keyword data as "(not provided)" in analytics, severing the connection between search queries and on-site behavior. GSC provides partial keyword data but cannot be joined to session-level analytics data.
"(not provided)" specifically refers to keyword data being removed from analytics sessions — not rankings visibility, GSC data, or backlinks.
14. In Starbucks' "occasion-based micro-segmentation" system, when is a customer's segment membership recalculated?
Correct. Real-time recalculation with each transaction enables immediate response to behavioral shifts like migrating from weekday morning to Saturday afternoon purchasing patterns.
Starbucks recalculates with each transaction — this real-time approach is what makes micro-segmentation actionable rather than historical.
15. In the AI content production tier model, which tier describes content where AI drafts and humans refine?
Correct. Tier 2 AI-Assisted is the dominant enterprise model: AI drafts, humans set strategy and refine. It captures most velocity benefit while maintaining editorial standards.
Tier 2 AI-Assisted is where AI produces first drafts that human editors refine. Tier 1 is fully automated; Tier 3 has humans writing with AI handling research and optimization.
16. The Gymshark Black Friday 2021 Smart Bidding failure occurred because:
Correct.
Gymshark's issue was that Target ROAS — optimising for incremental revenue — deprioritised branded keywords it considered "low-incremental," reducing visibility exactly when it mattered most.
17. Shopify's crawl budget problem in 2018 was caused by faceted navigation creating millions of URL variants. What is the standard technical fix for this issue?
Correct. Canonical tags pointing filter URLs to the canonical category page, combined with URL parameter configuration in Google Search Console, prevents crawl budget waste without blocking legitimate product pages.
Noindexing the product catalog would destroy organic visibility. Canonicalization and parameter handling are the correct solutions.
18. Apple's SKAdNetwork (SKAN) introduces "crowd anonymity thresholds" in SKAN 4.0. What do these thresholds do?
Correct. Crowd anonymity thresholds mean that if your niche campaign only generates a few conversions in a segment, SKAN returns no data — preventing inference about individuals from small-group statistics. This is mathematically necessary for differential privacy to work.
Crowd anonymity thresholds suppress reporting when conversion counts are too small, because small groups are more susceptible to re-identification even with noise added. This is a core privacy mechanism in SKAN 4.0.
19. Target's response to the pregnancy prediction controversy was to:
Correct. Target's pragmatic solution — diluting the signal with noise — preserved the personalization's commercial value while reducing the perceived surveillance quality that had triggered the backlash.
Not quite. Target's solution was elegant and revealing: they kept the personalization but camouflaged it by mixing in random, unrelated offers, so the pregnancy-targeted items didn't stand out as obviously inferred.
20. After Apple's ATT launch, iOS IDFA opt-in rates settled at approximately:
Correct.
Opt-in rates for IDFA access settled around 25–30% across most app categories — meaning roughly 70% of iOS users became unmeasured by third-party tracking systems relying on IDFA.