Why Insurance Companies Hire More Data Analysts in 2026

-- Simplified tables used in the examples
CREATE TABLE policies (
  policy_id   INT PRIMARY KEY,
  customer_id INT,
  product     VARCHAR(30),   -- e.g. 'Motor', 'Health'
  start_date  DATE,
  premium     DECIMAL(12,2)
);
CREATE TABLE claims (
  claim_id     INT PRIMARY KEY,
  policy_id    INT REFERENCES policies(policy_id),
  claim_date   DATE,
  claim_amount DECIMAL(12,2),
  status       VARCHAR(20)   -- 'Approved', 'Rejected', 'Pending'
);

Typical questions an insurance analyst answers:
1. Which product has the highest claim cost per policy?
2. Are claim amounts rising month by month?
3. How long does an average claim take to settle?
4. Which claims look unusual and need review?
5. Which customers are due for renewal and at risk?

-- Simplified loss ratio and claim frequency by product
SELECT p.product,
       COUNT(DISTINCT p.policy_id)                         AS policies,
       COUNT(c.claim_id)                                   AS claims,
       ROUND(COUNT(c.claim_id) * 1.0
             / COUNT(DISTINCT p.policy_id), 3)             AS claim_frequency,
       ROUND(SUM(COALESCE(c.claim_amount, 0))
             / SUM(DISTINCT p.premium), 3)                 AS approx_loss_ratio
FROM policies p
LEFT JOIN claims c
       ON c.policy_id = p.policy_id AND c.status = 'Approved'
GROUP BY p.product;

-- SUM(DISTINCT premium) breaks if two policies share a premium.
-- Aggregate each table first, then join:
WITH prem AS (
  SELECT product, COUNT(*) AS policies, SUM(premium) AS premium
  FROM policies GROUP BY product
), clm AS (
  SELECT p.product, COUNT(*) AS claims, SUM(c.claim_amount) AS paid
  FROM claims c JOIN policies p ON p.policy_id = c.policy_id
  WHERE c.status = 'Approved' GROUP BY p.product
)
SELECT prem.product, prem.policies,
       COALESCE(clm.claims, 0) AS claims,
       ROUND(COALESCE(clm.paid, 0) / prem.premium, 3) AS approx_loss_ratio
FROM prem LEFT JOIN clm ON clm.product = prem.product;

-- Flag claims for human review (rules, not accusations)
SELECT c.claim_id, c.policy_id, c.claim_amount, c.claim_date
FROM claims c
JOIN policies p ON p.policy_id = c.policy_id
WHERE c.claim_date < p.start_date + INTERVAL '30' DAY   -- very early claim
   OR c.claim_amount > 3 * (SELECT AVG(claim_amount) FROM claims);
-- Date arithmetic syntax varies: use DATE_ADD in MySQL.

-- Claims pending for more than 15 days
SELECT claim_id, policy_id, claim_date,
       CURRENT_DATE - claim_date AS days_pending
FROM claims
WHERE status = 'Pending'
  AND CURRENT_DATE - claim_date > 15
ORDER BY days_pending DESC;
-- PostgreSQL syntax; MySQL uses DATEDIFF(CURRENT_DATE, claim_date)

Project: Claims performance dashboard
Data: policies.csv, claims.csv (synthetic is fine)
Pages:
1. Overview - policies, claims, approx loss ratio
2. Trend - claims by month and product
3. Operations - pending claims and settlement time
4. Review list - rule-based flags
Write 5 insights in plain English at the end.

Total Premium = SUM ( Policies[premium] )
Approved Claims = CALCULATE ( SUM ( Claims[claim_amount] ), Claims[status] = "Approved" )
Loss Ratio = DIVIDE ( [Approved Claims], [Total Premium] )
Claim Count = COUNTROWS ( Claims )
Policy Count = COUNTROWS ( Policies )
Claim Frequency = DIVIDE ( [Claim Count], [Policy Count] )

Write insights like this (with your own numbers):
- Product A has the highest claim frequency this year.
- Pending claims older than 15 days rose in the last quarter.
- Early claims (within 30 days) form a small but notable share.
Always add: what you would check next.