What Metric Drop Questions Actually Test (And Why They Matter for Data Analysts)

Imagine you are three months into your new data analyst role at Paytm. Your manager walks over on a Monday morning and says, “Hey, our daily active users dropped 18% overnight — figure out what happened.” That exact real-world scenario is what interviewers are simulating when they throw a metric drop question at you.

Metric drop questions — sometimes called “metric investigation” or “root cause analysis” questions — are designed to evaluate whether you can think like a product analyst under uncertainty. The interviewer is not just checking if you can write SQL or build dashboards. They want to see your mental framework: how you prioritize hypotheses, how you separate signal from noise, and whether you ask the right clarifying questions before jumping to conclusions.

In the Indian tech job market, these questions are especially prominent at product-led companies. Swiggy might ask you why order completion rate dropped on a Tuesday night. Flipkart might ask why GMV fell during a sale event. Razorpay might ask why payment success rate dipped on a specific gateway. The domain changes, but the framework you need to answer it remains the same.

What makes this question type tricky is that there is rarely one single correct answer. Interviewers are evaluating your process — your ability to break a vague, open-ended problem into a logical, structured investigation — as much as they are evaluating your final hypothesis. A candidate who jumps straight to “it must be a bug” without asking clarifying questions will lose points immediately, even if they happen to guess the right answer.

The good news? This is a completely learnable skill. Once you internalize a solid framework and practice it across different product scenarios, metric drop questions become your opportunity to shine rather than something to dread.

Interview Questions This Topic Generates at Top Companies

Metric drop questions show up in data analyst and product analyst interviews at virtually every top tech company hiring in India right now — from Meesho and PhonePe to Google and Amazon. Below are some of the most commonly asked variations you should absolutely be prepared to answer with a clear, structured framework. Practice each one out loud before your next interview round.

1. “Our app’s Day 7 retention dropped by 15% last week compared to the previous week — walk me through how you would investigate this.” (Asked at Swiggy, Meesho, and similar growth-stage companies)
2. “Flipkart’s add-to-cart rate fell sharply during the Big Billion Days sale. How would you identify the root cause, and what data would you pull first?”
3. “Paytm’s transaction success rate on UPI payments dropped from 94% to 87% overnight. How would you diagnose whether this is an internal issue or an external one?”

The SQL Skills You Need to Investigate a Metric Drop in Real Life

When a metric drops, the first thing analysts do is segment the data — by time, by platform, by geography, by user cohort — to isolate where exactly the drop is happening. Here is a practical SQL query pattern used to detect metric drops by day and platform, the kind of query you might write in a real investigation at a company like Razorpay or Flipkart. Understanding this query will also help you talk confidently about your analytical approach during interviews.

-- Investigating a drop in order completion rate by platform and date
-- Compare current week vs previous week to isolate the drop

WITH weekly_metrics AS (
    SELECT
        DATE(created_at)                          AS event_date,
        platform,                                 -- 'android', 'ios', 'web'
        COUNT(order_id)                           AS total_orders,
        COUNT(CASE WHEN status = 'completed' 
                   THEN order_id END)             AS completed_orders,
        ROUND(
            COUNT(CASE WHEN status = 'completed' 
                       THEN order_id END) * 100.0
            / NULLIF(COUNT(order_id), 0), 2
        )                                         AS completion_rate_pct
    FROM orders
    WHERE created_at >= CURRENT_DATE - INTERVAL '14 days'
    GROUP BY DATE(created_at), platform
),

week_labels AS (
    SELECT
        *,
        CASE 
            WHEN event_date >= CURRENT_DATE - INTERVAL '7 days' 
            THEN 'current_week'
            ELSE 'previous_week'
        END AS week_label
    FROM weekly_metrics
)

SELECT
    platform,
    week_label,
    ROUND(AVG(completion_rate_pct), 2)  AS avg_completion_rate,
    SUM(total_orders)                   AS total_orders,
    SUM(completed_orders)               AS total_completed
FROM week_labels
GROUP BY platform, week_label
ORDER BY platform, week_label;

-- If one platform shows a sharp drop (e.g., Android: 91% → 76%),
-- that narrows your investigation to an app release, OS update,
-- or device-specific issue — classic root cause segmentation.