GPT-5

Two-week plan for launching a customer feedback survey for a 20-person software team:

1. Define survey objective and target customer segment.
2. Assign owner and small working group:

Owner: Product Manager
Support: UX, Customer Success, Engineering Lead

3. Draft 8–10 focused questions covering satisfaction, usability, missing features, and open feedback.
4. Review questions with leadership, support, and sales.
5. Choose survey tool and distribution channel.
6. Build the survey and add tracking fields.
7. Test internally with the 20-person team.
8. Fix unclear wording, broken logic, or technical issues.
9. Prepare launch message:

Subject: Help us improve [Product]
Time required: 3 minutes
Deadline: [Date]

10. Launch to selected customers at start of week 2.
11. Send one reminder after 3–4 days.
12. Monitor response rate daily.
13. Close survey at end of week 2.
14. Export results and summarize themes.
15. Share findings and recommended actions with the team.

Risk: Low response rate; reduce it by keeping the survey short and clearly explaining how feedback will be used.

Assume “99% accurate” means:

• Sensitivity: $P(+|D)=0.99$
• Specificity: $P(-|\neg D)=0.99$, so false positive rate $P(+|\neg D)=0.01$
• Prevalence: $P(D)=0.001$

Using Bayes’ rule:

$$ P(D|+) = \frac{P(+|D)P(D)}{P(+|D)P(D)+P(+|\neg D)P(\neg D)} $$

$$ = \frac{0.99 \times 0.001}{(0.99 \times 0.001) + (0.01 \times 0.999)} $$

$$ = \frac{0.00099}{0.00099 + 0.00999} = \frac{0.00099}{0.01098} \approx 0.0902 $$

So the chance they actually have the condition is about 9.0%.

sensitivity = 0.99
specificity = 0.99
prevalence = 1/1000

p = sensitivity * prevalence / (
    sensitivity * prevalence + (1 - specificity) * (1 - prevalence)
)
print(p)  # 0.09016...

Use ROW_NUMBER() to rank each customer’s orders by newest timestamp, then keep rank 1:

SELECT customer_id, order_id, created_at
FROM (
  SELECT
    customer_id,
    order_id,
    created_at,
    ROW_NUMBER() OVER (
      PARTITION BY customer_id
      ORDER BY created_at DESC, order_id DESC
    ) AS rn
  FROM orders
) ranked
WHERE rn = 1;

PARTITION BY customer_id ranks orders separately per customer.

ORDER BY created_at DESC puts the most recent order first.

order_id DESC is a tie-breaker if two orders have the same created_at.

If you want PostgreSQL’s concise DISTINCT ON version:

SELECT DISTINCT ON (customer_id)
  customer_id, order_id, created_at
FROM orders
ORDER BY customer_id, created_at DESC, order_id DESC;