eBay Data Analyst Interview Guide

eBay's Payment Compliance Data Analyst role sits inside the financial crimes and AML org, not the product analytics team most candidates picture when they hear "eBay DA." From coaching sessions we've run, candidates who prep only for marketplace funnel questions get blindsided when the interview pivots to data lineage through risk calculation engines, ETL validation for case management systems, and sanctions monitoring pipelines. If you want this role, you need to think like someone who audits the numbers before anyone else touches them.

eBay Data Analyst Role

You'll work at the intersection of eBay's payments infrastructure and its compliance obligations, writing SQL against large warehouse tables that track transaction risk scores, AML alerts, and seller/buyer dispute outcomes. Tableau dashboards you build won't just show conversion funnels; they'll feed decision boards used by risk and compliance leadership to monitor policy enforcement and flag suspicious activity patterns. Success after year one means owning a metric suite (say, the listing quality scorecard or a sanctions monitoring dashboard) that compliance stakeholders trust enough to present in regulatory exams without re-validating every number.

A Typical Week

The analysis block is the largest chunk, but the infrastructure and writing slices are what make this role distinct from a typical product DA seat. You're not just querying clean tables. You're tracing ETL lineage docs when upstream join keys don't match, filing tickets for stale snapshot tables, and documenting every field's source and transformation logic in Confluence so the next analyst doesn't have to reverse-engineer it. That documentation habit isn't busywork; it's what compliance audits demand.

Projects & Impact Areas

Trust and safety analytics anchors the work. You'll build dashboards tracking buyer/seller dispute rates, fraud detection metrics, and AML alert volumes, then present actionable findings to directors who need to decide whether a policy change warrants an A/B test or an immediate rollout. Seller tools measurement runs alongside this: when eBay launches new features to save sellers time and boost profits, you're quantifying adoption funnels and measuring whether promoted listings actually lift seller ROI or just shift organic traffic. The compliance thread ties everything together, because even "product-flavored" analyses (like return rates by category) eventually feed into risk models and regulatory reporting.

Skills & What's Expected

Data architecture and pipeline understanding is the skill most candidates underweight in prep. The role explicitly requires you to document end-to-end data flow, recommend architecture improvements, and implement data quality controls. Meanwhile, machine learning is listed as "nice to have" in the job description, so redirecting ML study hours toward schema design and lineage tracing is a better bet. Business acumen in payments and AML/sanctions compliance also scores high, meaning you need to speak the language of risk teams, not just product managers.

Levels & Career Growth

The gap between P3 and P4 comes down to metric ownership. P3 analysts execute well-scoped analyses and maintain dashboards; P4 analysts define what the team should measure and push back when a metric definition doesn't hold up under scrutiny. That distinction matters especially in compliance analytics, where a poorly defined KPI can create regulatory exposure. At P5 and above, you're setting measurement frameworks across multiple teams and mentoring others on analytical rigor.

Work Culture

The analytics org values getting the metric definition right before shipping a dashboard. That's a real cultural signal, not a platitude, and it shows up in how much interview weight goes to data quality. The pace is steady rather than startup-frantic, which is refreshing if you've been burned by "ship it now, fix the numbers later" environments. Some employee reviews note that strategic shifts can slow things down mid-quarter, but for compliance-adjacent work, deliberate beats fast.

eBay Data Analyst Compensation

eBay's RSUs vest over a multi-year schedule (from what candidates report, four years with periodic vesting is the norm). Ask your recruiter to confirm the exact cliff length, vesting cadence, and whether refresh grants exist for your level, because these details vary by offer and aren't published. The equity component won't surprise you with wild swings, but it also won't carry your comp the way it might at a pre-IPO startup.

Your strongest negotiation move at P3 and P4 is bringing a competing offer from another marketplace or e-commerce company. Base salary, sign-on bonus, and RSU grant size all have some flex within band, while bonus targets are level-based and rarely budge. Some eBay DA roles appear under titles like "Data Science Analyst" or "Payment Compliance Data Analyst" with different comp bands, so confirm the P-level during your recruiter screen to make sure you're negotiating against the right range. Benefits (401k match, wellness perks at jobs.ebayinc.com/us/en/our-benefits) add real value beyond the offer letter numbers.

eBay Data Analyst Interview Process

Communication gaps between rounds are the hidden timeline killer. Recruiters sometimes go quiet for a week or more after a stage, and candidates who don't send a short "confirming next steps" email risk drifting into a six-week process instead of four. The most common rejection pattern, from what candidates report, is imprecise metric definitions. eBay interviewers will push you to specify how you'd handle auction-specific edge cases like zero-bid items, Best Offer negotiations, and relisted listings when defining something like seller conversion. Vague answers signal you'd struggle with eBay's real problem: five competing definitions of the same metric across teams built over three decades.

Rounds are spread across multiple weeks rather than packed into a single onsite day. The behavioral round may circle back to decisions you made in earlier technical rounds, so treat every answer as something you'll need to defend again later. If you're strong in SQL but shaky on two-sided experiment design (buyer-side vs. seller-side interference in eBay's promoted listings tests, for example), that gap will surface and no single stellar round will paper over it.

eBay Data Analyst Interview Questions

1WITH first_impression AS (
2  SELECT
3    se.user_id,
4    DATE(se.event_ts) AS event_date,
5    MIN(se.event_ts) AS first_impression_ts
6  FROM search_events se
7  WHERE se.event_type = 'impression'
8  GROUP BY
9    se.user_id,
10    DATE(se.event_ts)
11), purchaser_user_day AS (
12  SELECT
13    fi.event_date,
14    fi.user_id
15  FROM first_impression fi
16  WHERE EXISTS (
17    SELECT 1
18    FROM orders o
19    WHERE o.buyer_id = fi.user_id
20      AND o.order_ts >= fi.first_impression_ts
21      AND o.order_ts < fi.first_impression_ts + INTERVAL '24' HOUR
22  )
23)
24SELECT
25  fi.event_date,
26  COUNT(*) AS unique_searchers,
27  COUNT(pud.user_id) AS purchasers,
28  CAST(COUNT(pud.user_id) AS DECIMAL(18,6)) / NULLIF(COUNT(*), 0) AS conversion_rate
29FROM first_impression fi
30LEFT JOIN purchaser_user_day pud
31  ON pud.event_date = fi.event_date
32 AND pud.user_id = fi.user_id
33GROUP BY fi.event_date
34ORDER BY fi.event_date;

1WITH month_bounds AS (
2  SELECT DISTINCT
3    DATE_TRUNC('month', o.order_ts) AS month_start,
4    (DATE_TRUNC('month', o.order_ts) + INTERVAL '1' MONTH - INTERVAL '1' SECOND) AS month_end
5  FROM orders o
6), last_event_asof_eom AS (
7  SELECT
8    mb.month_start,
9    mb.month_end,
10    le.listing_id,
11    le.seller_id,
12    le.event_type,
13    ROW_NUMBER() OVER (
14      PARTITION BY mb.month_start, le.listing_id
15      ORDER BY le.event_ts DESC
16    ) AS rn
17  FROM month_bounds mb
18  JOIN listing_events le
19    ON le.event_ts <= mb.month_end
20), listing_state_eom AS (
21  SELECT
22    month_start,
23    listing_id,
24    seller_id,
25    event_type
26  FROM last_event_asof_eom
27  WHERE rn = 1
28), active_listings_eom AS (
29  SELECT
30    month_start,
31    seller_id,
32    COUNT(*) AS active_listings_eom
33  FROM listing_state_eom
34  WHERE event_type <> 'ended'
35  GROUP BY month_start, seller_id
36), gmv_month AS (
37  SELECT
38    DATE_TRUNC('month', o.order_ts) AS month_start,
39    ls.seller_id,
40    SUM(o.gmv) AS gmv
41  FROM orders o
42  JOIN listing_state_eom ls
43    ON ls.listing_id = o.listing_id
44   AND ls.month_start = DATE_TRUNC('month', o.order_ts)
45  WHERE o.order_ts >= ls.month_start
46    AND o.order_ts < ls.month_start + INTERVAL '1' MONTH
47  GROUP BY
48    DATE_TRUNC('month', o.order_ts),
49    ls.seller_id
50)
51SELECT
52  COALESCE(a.month_start, g.month_start) AS month_start,
53  COALESCE(a.seller_id, g.seller_id) AS seller_id,
54  COALESCE(a.active_listings_eom, 0) AS active_listings_eom,
55  COALESCE(g.gmv, 0) AS gmv
56FROM active_listings_eom a
57FULL OUTER JOIN gmv_month g
58  ON g.month_start = a.month_start
59 AND g.seller_id = a.seller_id
60ORDER BY month_start, seller_id;

When eBay's experimentation questions ask you to pick guardrails for a seller protection change, they expect you to also reason about whether the exposure logging captured both buyer and seller interactions correctly, pulling pipeline and A/B testing into the same answer. That overlap between data quality and experiment design is where most candidates stall, because practicing them as separate topics leaves you unprepared for prompts like the risk_score lineage question above, where a 12% discrepancy between systems is the actual problem to solve before any statistical analysis matters.

Practice eBay-style marketplace questions across all six areas at datainterview.com/questions.

How to Prepare for eBay Data Analyst Interviews

eBay is pouring resources into vertical experiences for enthusiast buyers (luxury watches, trading cards, auto parts) and AI-driven seller tools designed to save time and boost profits. The company posted $11.1 billion in revenue with 15% year-over-year growth, and the 2026 category trends and global ad campaign signal where analyst attention is heading next. If you're prepping for a loop, those two initiatives are worth studying closely.

The "why eBay" answer that actually works connects eBay's take-rate-plus-ad-revenue model to a specific vertical or seller tool problem you'd want to investigate. Saying you want to measure how the new AI listing tools affect seller adoption funnels in auto parts, for example, shows you grasp that eBay's revenue isn't GMV and that each vertical has its own dynamics. That kind of specificity separates you from vague answers about marketplace scale.

Try a Real Interview Question

1WITH first_exposure AS (
2  SELECT
3    e.user_id,
4    e.variant,
5    MIN(e.exposure_ts) AS first_exposure_ts
6  FROM experiment_exposures e
7  WHERE e.exp_id = 101
8  GROUP BY e.user_id, e.variant
9), eligible_exposure AS (
10  SELECT fe.*
11  FROM first_exposure fe
12  LEFT JOIN user_fraud_flags f
13    ON f.user_id = fe.user_id
14  WHERE f.user_id IS NULL
15    AND fe.first_exposure_ts >= TIMESTAMP '2024-01-01 00:00:00'
16    AND fe.first_exposure_ts <  TIMESTAMP '2024-01-08 00:00:00'
17), purchasers AS (
18  SELECT
19    ee.variant,
20    ee.user_id
21  FROM eligible_exposure ee
22  JOIN orders o
23    ON o.user_id = ee.user_id
24   AND o.order_ts >= ee.first_exposure_ts
25   AND o.order_ts <  ee.first_exposure_ts + INTERVAL '7' DAY
26  GROUP BY ee.variant, ee.user_id
27)
28SELECT
29  ee.variant,
30  COUNT(DISTINCT ee.user_id) AS exposed_users,
31  COUNT(DISTINCT p.user_id) AS purchasers_7d,
32  CAST(COUNT(DISTINCT p.user_id) AS DECIMAL(18,6))
33    / NULLIF(COUNT(DISTINCT ee.user_id), 0) AS conversion_rate_7d
34FROM eligible_exposure ee
35LEFT JOIN purchasers p
36  ON p.variant = ee.variant
37 AND p.user_id = ee.user_id
38GROUP BY ee.variant
39ORDER BY ee.variant;

Marketplace SQL problems tend to involve joins across buyer, seller, and transaction tables where you need to explain your schema assumptions out loud, not just return correct rows. From what candidates report, interviewers care as much about how you reason through ambiguous data (what counts as "active"? how do you handle items that never sold?) as the query itself. Drill similar patterns at datainterview.com/coding.

Test Your Readiness

Run through marketplace-framed product sense and experimentation questions at datainterview.com/questions. For every prompt, force yourself to cover both the buyer metric and the seller metric, because eBay interviewers will ask about whichever side you skip.