LDS Meta-Style SQL 30

Before the Technical Screen. 5 single-table-and-basic-join problems that gate entry to a 45–60 min CoderPad screen.

Meta's Technical Screen assumes you can already write a clean 2-table JOIN, a GROUP BY aggregation, and a basic date filter. If any of these feel slow, fix that first — Meta will move past these in the first 5 minutes.

• Most relevant for: every role (universal floor at Meta)
• Skip if: you're senior — jump to Stage 2
• Domain: social-graph + ad-tech (Meta's two business surfaces)

Meta's first real loop round — 45–60 min on CoderPad with a senior IC. Typical report: 2 SQL questions in the first 20 minutes, then a light product discussion.

Patterns tested:

• Multi-table JOINs with IN filters
• GROUP BY + HAVING thresholds
• Conditional aggregation (SUM(CASE WHEN ...))
• First RANK / DENSE_RANK window function
• The signature conditional-rate pattern (e.g., "invalid click rate by campaign type")

Train yourself to write each in 10–12 minutes. Explain while you code — Meta scores fluency, not just correctness.

• Most relevant for: DSA, DSP, DE, DS-Core
• Also mirrors: Google SQL phone screen, Microsoft HackerRank assessment

The dedicated 45-min SQL round inside Meta's Full Loop onsite. A peer IC extends one or two hard questions for 30+ minutes with probing follow-ups:

"Now add a 1-day forgiveness window…" "Now partition this by visibility…" "What if two messages arrive at the same second?"

Patterns tested (the Meta canon):

• Top-N within group — ROW_NUMBER() OVER (PARTITION BY user_id ORDER BY ...)
• Running totals / DAU rolling — SUM(...) OVER (... ROWS BETWEEN ...)
• Moving averages — Meta's 7-day rolling DAU shape
• Anti-joins — "published posts without reactions" / "pages with no likes"
• Self-join symmetric — "mutual follow pairs" / friend-of-friend
• Islands-and-Gaps streak detection — the classic date − ROW_NUMBER() trick (one Medium writeup claims "80% of Meta DE candidates fail this")
• Sessionization with idle timeout — LAG + cumulative-sum to assign session_ids when consecutive events exceed a 30-min gap

This is where Meta separates IC3/IC4 from IC5 candidates — window-function fluency is the gate.

• Most relevant for: DSA, DSP, DE
• Also mirrors: Apple onsite SQL, Netflix sessionization deep-dive, Google L5 SQL

Meta's distinctive 45-min round — the one no other FAANG company has. Probability, statistics, hypothesis testing, A/B math, and metric-drop diagnosis — all under time pressure.

Meta is the A/B testing company. Every IC level (E3 → E6) is tested on experimentation rigor: lift computation, sample-size adequacy, p-value sanity, control-vs-treatment SQL reads. Plus the iconic "DAU dropped 15% — what query do you write first?" scenario.

Patterns tested:

• Cohort retention — DataLemur's canonical Active-User-Retention question
• A/B test reads — variant comparison with significance checks
• Funnel drop-off — percentile + multi-step conversion
• Rolling KPIs — Meta's 7-day-moving-average CTR for trend detection
• Significance sanity — checking the experiment is well-powered before reading the result

If you skip this stage, you skip the round that decides Meta DS offers.

• Most relevant for: DSA, DSP, DS-Core
• Also mirrors: Stripe analytical-reasoning round (causal inference flavor)

The 45-min product-sense round. "No math, no one right answer" — but you'll still write SQL.

The interviewer hands you an under-specified business prompt:

"How would you measure Reels' content health?" "What's the right metric for group quality?" "If you had to score user engagement on one number, what would go into it?"

You're being graded on judgment, scope, and metric design. The SQL is downstream — the harder skill is deciding what to compute.

The 6 problems are deliberately multi-CTE composite-score scorecards:

• Content virality report — engagement/reach composite
• Group health dashboard — 2-CTE classifier
• User influence scorecard — multi-metric weighted
• Trust & Safety report — 5-CTE moderation aggregate
• User engagement dashboard — 6-CTE weighted composite (the capstone)
• Comment-thread depth — recursive CTE

Every explainer ends with a "How an IC6 interviewer would extend this" section — staff-level probing follow-ups.

• Most relevant for: IC5+ candidates at Meta (DSA, DSP); LinkedIn, TikTok, Snap, Pinterest analytics
• Also mirrors: Stripe analytical-reasoning, Google product-sense onsite