Meridian · Atterna Analyst

The dashboard says +75%. Your query says +54%.

That’s episode 1 of Meridian: eight episodes of analyst work on a practice marketplace’s data. You check each number on the data before the team acts on it, then defend the decision.

about eight hours on the main path
SQL, plus ready-made Python
practice data, real traps

Play episode 1 free with an account →

Episodes 2–8 are in Atterna Analyst: €59 · one-time · early-access price. What’s included

Episode 1 · Record week · Tuesday, 22 September · #promo-review

Step 1 of 3

Vera Lind commercial director

Good morning! The AUTUMN20 promotion ended on Sunday, and the dashboard shows a record: weekly GMV grew 75% compared with a normal week.

Dashboard · weekly GMV+75%promo week vs a normal one

Tomás Ruiz: “Best week of the year! Extend and double it” · Decision: Thursday meeting, 11:00

Step 2 of 3

Vera Lind commercial director

Before anyone calls it a record, let's check: paid orders and GMV for each of the three full weeks — two normal ones and the promo.

Your query

SELECT  w.week_start,  COUNT(*) AS orders,  SUM(o.items_total_eur) AS gmv_eurFROM meridian.weeks AS wJOIN meridian.orders AS o  ON o.created_at >= w.week_start AND o.created_at < w.week_endWHERE o.status = 'paid'GROUP BY w.week_startORDER BY w.week_start

Week	Paid orders	GMV
31 Aug – 6 Sep	2,625	€149,812
7–13 Sep	2,751	€150,503
14–20 Seppromo	3,289	€230,967

You analyst

Vera, here are paid orders and GMV for the three weeks. The promo week against an average normal one is +54%. I don't count cancelled orders; they aren't part of GMV.

Vera Lind commercial director

Hmm. But the dashboard says +75%. Where does the difference come from?

Step 3 of 3

Dashboard query (Luka)

SELECT date_trunc('week', o.created_at) AS week_start,       COUNT(DISTINCT o.order_id) AS orders,       SUM(o.items_total_eur) AS gmv_eurFROM meridian.orders AS oJOIN meridian.order_items AS i ON i.order_id = o.order_idWHERE o.status = 'paid'GROUP BY 1ORDER BY 1

Week	Dashboard GMV	Correct GMV
31 Aug – 6 Sep	€323,431	€149,812
7–13 Sep	€313,943	€150,503
14–20 Seppromo	€557,752	€230,967

You analyst

Luka, I think I found it. After the JOIN with line items the order total repeats on every row, and SUM adds it as many times as the order has products. Attached is a week-by-week comparison: how the dashboard counts and how it really is.

Luka Kim senior analyst

So that's it — annoying. In the promo people bought more items at a time, 1.9 line items on average against 1.7, so the dashboard overstated the promo week most of all. I'll fix it now. Good thing this came up before Thursday and not at the meeting itself.

+54%Promo week: +54% vs an average normal week. The dashboard shows +75%.

Be sure of the number before you say it.

Next in episode 1: “A chart for the meeting”, “Reconciling with finance”, “What the platform really earned” and your recommendation for Thursday.

Play episode 1 free with an account →

Eight episodes

Each one is a working week: a colleague brings a number, you check it on the data and make the call. Each opens when you finish the one before. In total, about eight hours on the main path.

1
Record week
A record promo week on the dashboard. Was the record real, and did the week make the platform money?
Free with an account
2
The fast-delivery badge
An A/B test says a badge lifted conversion. Were the groups split fairly, and is the lift more than noise?
Atterna Analyst · €59
3
Who stays after the first purchase
Retention looks like it collapsed. Who really comes back, and who only came for the discount?
Atterna Analyst · €59
4
The seller who drags things down
A list of shops to switch off by rating. Is the rule fair to small shops, and can you defend it?
Atterna Analyst · €59
5
What one delivery costs
Free delivery from €25, or a fee for everyone? What delivering one basket really costs.
Atterna Analyst · €59
6
A forecast for the sale
How many couriers to book for each day of the spring sale, checked against last year first.
Atterna Analyst · €59
7
The price of a discount
End the commission discount for big shops? Both sides quote the same test; you read it properly.
Atterna Analyst · €59
8
Red Monday
Weekly GMV is down 15% and three people blame three causes. Find the step that broke before 12:00.
Atterna Analyst · €59

Real screens

From episode 2, “The fast-delivery badge”: the product team says a badge lifted conversion by 17%.

The task: a colleague’s request, why it matters, what to hand in and the table to end up with. The numbers the decision rests on stay in view.Swipe sideways, or tap to open full size

Your query runs on the episode’s data. Split by app version, one group is missing people: app 5.2.Swipe sideways, or tap to open full size

Accepted: the check reruns your query on a second dataset, the dashboard number changes from 50/50 to 48.6%, and the team replies.Swipe sideways, or tap to open full size

Is it for you?

For people who already answer work questions with data. New to SQL? SQL is free: all 10 days of Start with a free account. You pay only for practice: case rounds (€19) and Meridian (€59).

What’s real and what’s practice

The marketplace, colleagues and records are fictional practice data; the SQL, the checks and the kinds of traps are real. Before you start: How to check numbers an AI gives you.

Atterna Analyst · Meridian

Eight working weeks, about eight hours on the main path

All Meridian episodes, all cases and everything in Foundations, for one payment.

8 Meridian episodes
Metric choice
A/B tests
Cohorts
Forecasts
Memos the team acts on
All cases
Foundations included

€59Atterna Analyst
one-time · early-access price

Get Analyst

One-time payment, no subscription · Full refund within 14 days

Or play episode 1 first, free with an account Or Atterna Foundations · €19