Google Ads Audiences, Remarketing and Attribution Interview Questions
Google Ads remarketing interview questions: 79 questions with model answers on audiences, remarketing, Customer Match, attribution and experiments, from Basic to Advanced.
- Model answers
- Basic to Advanced
- Search and filter
Remarketing, Customer Match, attribution models, experiments and incrementality. Questions are grouped from Basic to Advanced, each with a model answer. Use the filters or search to find what you need, then take the free Google Ads test to check your score. For official definitions and policies, see the Google Ads Help Center.
What these Google Ads remarketing interview questions cover
These Google Ads remarketing interview questions cover audience segments, remarketing lists, Customer Match and first-party data, optimised targeting, attribution models including data-driven attribution, campaign experiments, lift tests and incrementality.
Interviewers use these questions to check whether you can prove that ads really work, not just report numbers. Work through the audience questions first, then the attribution and experiment questions, and compare models with the free Attribution Model Comparator tool.
All questions, from Basic to Advanced
Basic questions
What is remarketing in Google Ads?
- Showing ads to people who previously interacted with your website, app, YouTube channel or customer lists, across Search (RLSA), Display, YouTube, Demand Gen, PMax signals.
What is RLSA?
- Remarketing Lists for Search Ads: using audience lists in Search campaigns to adjust bids (observation) or restrict targeting to past visitors.
What's the difference between "Targeting" and "Observation" audience settings?
- Targeting: ads show only to people in the audience.
- Observation: no restriction; you can see performance and (with manual bidding) adjust bids; Smart Bidding uses signals anyway.
What is Customer Match?
- Uploading first-party customer data (emails, phones, addresses; hashed) to target/exclude customers or seed lookalikes across Search, Shopping, YouTube, Gmail, Display, PMax signals.
Expert insight: 2026 updates centralize uploads through Data Manager and expanded identifiers in some regions; match rates improve with phone in E.164 + email + names.
Intermediate questions
How does data-driven attribution (DDA) work in Google Ads?
- Uses observed paths of converting and non-converting users to estimate each Google Ads interaction's contribution (Search, Shopping, YouTube, Display) and assign fractional credit.
Expert insight: DDA compares paths of converters vs non-converters (counterfactual approach) to estimate incremental contribution of each touch; it only sees Google Ads touchpoints.
Why did Google remove rule-based models (first click, linear, time decay, position-based)?
- Low usage and DDA's superiority as default; only last click remains as an alternative.
When might last click still be useful?
- Simple accounts, comparing with external last-click systems, very low data; but it undervalues upper-funnel campaigns.
What are Google Ads custom experiments?
- Duplicate a campaign as a trial with changes, split traffic (cookie-based or search-based), run for weeks, and compare with statistical significance; apply or discard.
What types of experiments can you run?
- Custom experiments (bidding, keywords, match types, landing pages), ad variations/assets tests, video experiments (creative A/B with brand lift), PMax uplift and asset experiments, AI Max one-click experiment, Demand Gen A/B.
Cookie-based vs search-based split – difference?
- Cookie-based: a user is consistently in one arm (better for user-level effects).
- Search-based: each search randomly assigned (faster data, but same user can see both).
How long should an experiment run?
- Typically 4–8 weeks, at least 2 full conversion cycles and enough conversions per arm for significance; avoid ending early.
What are the eligibility requirements for Customer Match?
- Good policy compliance history, good payment history, account history and spend thresholds (Google updated requirements), and compliance with Customer Match policy (data collected with consent).
What are in-market and affinity segments?
- In-market: people actively researching/planning purchases in a category.
- Affinity: people with long-term interests/lifestyles.
What are custom segments?
- Audiences you define by search keywords, URLs or apps – Google finds people with matching interests/behavior.
What is "Your data" segment?
- Remarketing lists (website visitors, app users, YouTube users), Customer Match lists, GA4 audiences.
What's the minimum list size for remarketing?
- Generally 100 active users for Display/YouTube and 1,000 for Search (RLSA) to serve; lists must also meet eligibility.
How should audiences be used with Smart Bidding in Search?
- Add audiences in Observation mode to gain insights; Smart Bidding already uses audience signals; avoid Targeting mode unless you want to restrict (e.g. RLSA-only for broad expensive keywords).
What is optimized targeting?
- Google expands beyond your selected audiences to find people likely to convert (Display, Demand Gen, Video); turn off for strict remarketing.
How do you build an effective audience strategy for PMax?
- Signals: Customer Match (best customers), website converters, custom segments from top converting search terms and competitor URLs, in-market segments relevant; one signal per asset group theme.
What are GA4 predictive audiences?
- Audiences of users likely to purchase/churn within 7 days or predicted top spenders, available when GA4 has enough data; usable in Ads.
How do you exclude existing customers in Search?
- Customer Match list as exclusion (Targeting/exclusions), or use new customer acquisition settings; note match rates aren't 100%.
What are the privacy restrictions for audiences?
- No targeting based on sensitive categories (health, religion, sexual orientation, financial hardship); personalized advertising policy; housing/employment/credit restrictions in some countries; Consent Mode for EEA/UK.
Advanced questions
What's the problem with "before/after" testing in Google Ads?
- Seasonality, competitor changes, tracking changes, learning periods confound results; experiments with control arms are more reliable.
What is Conversion Lift in Google Ads and when to use it?
- Randomized holdout studies measuring incremental conversions from YouTube/Demand Gen and (2026) Search/PMax for eligible advertisers; use when deciding major budget allocation.
Expert insight: 2026 self-serve lift for Search/PMax requires minimum budget/conversion thresholds (e.g. 1,000+ conversions per Google's guidance) – plan budgets accordingly.
What are geo experiments and how are they run for Google Ads?
- Turn ads on/off or change spend in matched regions; compare outcomes with synthetic control; Meridian GeoX and open-source tools support analysis; works across channels/offline sales.
Brand search incrementality: how would you test it?
- Pause brand ads in some geos/time periods (or holdout experiment), measure total brand conversions (paid + organic); if most shift to organic, brand ads are less incremental; consider competitor presence.
What is MMM and how does Google Meridian help?
- Marketing mix modeling estimates channel contributions from aggregated time-series data; Meridian is Google's open-source Bayesian MMM with reach/frequency and calibration from experiments.
How do you combine platform attribution, experiments and MMM?
- Platform for daily optimization, experiments for causal calibration, MMM for budget allocation across channels; reconcile with business KPIs (MER, CAC).
What does "statistically significant" mean in Google Ads experiments?
- The experiment shows a confidence level (e.g. 95%) that the difference isn't random; Google shows confidence intervals for metrics.
Why can an experiment show "not significant" even if trial looks better?
- Insufficient sample size/duration, small effect size, high variance; extend duration or increase traffic split.
What's the minimum traffic split to choose?
- 50/50 for fastest significance; lower splits (e.g. 20%) reduce risk but take longer.
How do you evaluate lead-gen experiments when lead quality lags?
- Import offline outcomes, wait for qualified data, compare cost per qualified lead; or use proxy metrics with validated correlation.
Explain "cross-device conversions" in Google Ads.
- Conversions where click on one device and conversion on another, measured via signed-in users and modeling; included in conversions column.
What is the "Paths" and "Assisted conversions" reporting?
- Attribution reports show conversion paths, top paths, assisted conversions, path length and time lag – useful to see upper-funnel contribution.
How does view-through conversion counting differ across campaign types?
- Display/Demand Gen/Video can record view-through/engaged-view conversions; Search has no view-throughs; check if VTCs are included in "Conversions" (often only in all conversions).
Your PMax claims 300 conversions; Search claims 400; GA4 shows 500 total for Google paid. Why sums differ?
- Each campaign uses DDA within Google Ads paths; GA4 uses cross-channel attribution; overlapping interactions; modeled conversions; reporting date differences.
How do you measure the value of upper-funnel YouTube for a lead-gen business?
- Brand lift, search lift, conversion lift, geo tests, CRM "first touch" fields, DDA assisted conversions.
What biases exist in Google Ads conversion reporting?
- Self-attribution (Ads credits only Ads touchpoints), retargeting/brand capture of existing demand, modeled conversions, view-through credit.
What is incrementality vs ROAS?
- ROAS = attributed revenue ÷ cost; incrementality = additional revenue caused by ads; high ROAS campaigns (brand/retargeting) can have low incrementality.
How would you design an incrementality test for PMax in the UAE?
- Geo split by emirates (e.g. Dubai+Sharjah vs Abu Dhabi+others) after checking comparable baselines, or Google's conversion lift if eligible; 4–6 weeks; measure total online sales from backend.
What sample size considerations exist for lead-gen tests with 100 leads/month?
- Very small; detecting differences < 30–50% is hard; test big changes, run longer, use proxy metrics (form starts) carefully.
What's "regression to the mean" in PPC decisions?
- Keywords/ads with extreme performance in small samples tend to normalize; avoid reacting to one week of data.
Explain "time lag" report and its use.
- Shows days between first click and conversion; informs conversion window settings and when to evaluate recent data.
How do you report "true" CAC across Google channels?
- Backend new customers attributed (with UTM/GCLID linkage), incremental adjustments from lift tests, blended CAC = total Google spend ÷ new customers (sanity check).
What's the role of UTMs when auto-tagging is on?
- GA4 uses GCLID for Google Ads; UTMs still helpful for CRM and other analytics; avoid conflicting manual UTMs that override (set "allow manual tagging to override" carefully).
What does "conversion modeling" mean for experiments?
- Modeled conversions apply to both arms similarly; still, ensure tracking consistent across arms.
How do you test landing pages in Google Ads properly?
- Custom experiment with different final URLs (split traffic) or CRO tool A/B with consistent traffic; measure CVR and lead quality.
What's a "holdout" in remarketing and why use one?
- Exclude a random portion of remarketing audience to measure incremental conversions of remarketing.
How do you present experiment results to a CEO?
- Hypothesis, setup, duration, results with confidence intervals, business impact (extra profit/leads), decision and next steps.
Explain "Meridian GeoX" (2026).
- Google's geo-experiment tooling integrated with Meridian for designing and analyzing geo lift tests to calibrate MMM; availability expanded globally in 2026.
What's the difference between incremental ROAS and marginal ROAS?
- iROAS: incremental revenue per cost for the whole campaign vs no ads.
- Marginal ROAS: return of the next unit of spend; used for budget increases.
How do you calculate marginal ROAS from Performance Planner or tests?
- Compare value at two budget levels: (value2 − value1) ÷ (cost2 − cost1); scale while marginal ROAS ≥ break-even.
A/B testing ad copy: why use conversions per impression?
- Captures both CTR and CVR in one metric; best for ad testing within same ad group.
Interview: "How do you know Google Ads is working for a client with offline sales?"
- Offline conversion imports tied to GCLID/EC for leads, closed revenue by campaign, cohort analysis, lift tests (geo), and MER trends vs spend changes.
What is the "Results" tab/experiment results interpretation?
- Shows change % per metric with confidence and interval; look at primary KPI (conversions/value) and cost; apply only if significant positive impact on goal.
What's the risk of running multiple experiments simultaneously on overlapping campaigns?
- Interaction effects confound results; schedule tests sequentially or isolate campaigns.
How do you attribute YouTube-driven brand searches (2026 changes)?
- Google introduced crediting YouTube/Demand Gen for subsequent branded searches (with brand mapping and windows) as a reporting metric/goal category – use as supplementary evidence, validate with lift.
How do you improve Customer Match match rates?
- Include multiple identifiers (email, phone with country code +971, names, country, zip), normalize data, upload regularly via Data Manager; 2026 options allow more identifiers in some regions.
How would you use Customer Match for a business setup company?
- Upload existing clients to exclude from acquisition, upsell renewals/accounting, seed lookalikes in Demand Gen, signal PMax with high-value clients.
What's a "combined segment"?
- Audience combining segments with AND/OR/NOT logic (e.g. in-market for real estate AND visited pricing page NOT converted).
How do you segment remarketing by funnel stage?
- Visitors (all), engaged (60s+/2+ pages), product/service viewers, cart/form starters, converters; different messages/bids; exclude converters.
What's the best membership duration for remarketing lists?
- Match sales cycle: e-commerce 7–30 days for carts, 90–180 for visitors; B2B/real estate 180–540 days.
Why might remarketing lists show "Too small to serve" in Search?
- Below 1,000 active users; increase duration, combine lists, or use Observation.
How do you use YouTube engagement audiences?
- Viewers of specific videos, channel subscribers, engaged viewers – target in Demand Gen/Video with deeper funnel content.
Demand Gen lookalike segment settings – what's narrow vs broad?
- Narrow: closest similarity (~2.5% of population), Balanced (~5%), Broad (~10%); suggestion mode (2026 default in many cases) allows expansion.
How does Google identify "new customers" for NCA goals?
- From Customer Match lists, converters tags (past purchasers), and conversion-based customer lists; quality of lists determines accuracy.
What's a "high-value customer" list and how to use it?
- Customers with top LTV; used as signal, for value rules (higher value for similar users), or for new customer acquisition value settings.
How do you protect against audience targeting violating policies in real estate (housing)?
- Avoid sensitive targeting, follow housing/credit restrictions where applicable (some countries restrict age/gender/zip targeting), rely on contextual and intent signals.
What are "life events" segments and how could they help in the UAE?
- Audiences around life milestones (moving, starting business, marriage, graduation) – e.g. "starting a business" for business setup firms, "moving" for real estate/movers.
How do you build a custom segment for business setup in Dubai?
- Keywords: "freezone license cost", "start company in UAE", "trade license Dubai"; URLs: competitor/freezone authority sites; apps: business/banking apps; use in Demand Gen/YouTube and PMax signals.
Why are audience signals less important for mature PMax campaigns?
- Conversion data and real-time signals dominate; signals help initial exploration and asset group direction.
What's audience expansion's impact on reporting?
- Results may come from outside original lists; check segment reports to see which segments drove results.
How do you refresh Customer Match lists automatically?
- Data Manager scheduled connections (CRM, BigQuery, Sheets, Salesforce/HubSpot integrations) or API (Data Manager API) syncs.
Interview: "If cookies disappear for most users, how will you still target and measure?"
- First-party data (Customer Match, CRM, logged-in users), enhanced conversions, consent mode modeling, contextual/custom segments, AI-driven targeting, MMM/lift measurement.
What is "similar audiences" replacement in Search?
- None directly; Smart Bidding and optimized targeting use signals; Customer Match/your data as observation give insights.
How can you use audiences to bid more for high-value users in Smart Bidding?
- Conversion value rules for audiences (e.g. +50% value for "high-intent visitors" or Customer Match VIPs), separate campaigns for RLSA with higher targets.
What's a 2026 change in Customer Match uploads?
- Google is centralizing uploads via Data Manager/API (some legacy methods phased), expanded identifiers (e.g. IP/timestamp in certain regions), changes to hashing/encryption requirements – check current docs.
When would you target remarketing audiences on Search with broad keywords?
- To reach past visitors with broad, expensive, generic terms (e.g. "business setup") only if they've shown interest – RLSA-only campaign with broad keywords keeps costs efficient.
How do you use Customer Match to prevent wasted spend on existing leads?
- Exclude recent leads (uploaded daily) from acquisition campaigns; nurture them with different messaging in remarketing.
More Google Ads question topics
Want an expert to run your Google Ads?
Book a free strategy call and get a clear plan for your campaigns.