Level of Detail for Dummies · Tableau Tuesdays

Level of Detail Expressions

How to make Tableau compute a number at any level you want — INCLUDE, EXCLUDE, and FIXED explained from scratch.

Mary Kinya · June 14, 2026

Calculations Intermediate Tableau Tuesdays

TABLEAU TUESDAYS

BEGINNER SERIES

Level Of Detail,
explained.

#DataFam #KenyaTUG #Tableau

tableauuser groupsNairobi

THE PROBLEM

Your chart can only
answer one question at a time.

You build a chart grouped by Region.
Tableau totals everything by Region since that's all it knows.

Then your boss asks:
"What's the average spend per customer?"

That's a different question. At a different level.
Tableau can't answer it from a Region chart. Not without help.

LOD = Level of Detail

A way to tell Tableau: "Calculate this at a different level than what's on the chart." That's the whole idea.

THINK OF IT THIS WAY

Like a cashier who only
sees the daily total.

A cashier prints one receipt at the end of the day, the grand total.

You ask: "How much did customer Jane spend?"
The cashier says: "I only have the total."

An LOD expression is like giving that cashier a separate notebook that tracks each customer, even while the till still shows the daily total.

😕 Without LOD

Tableau only calculates at your chart level. You're stuck with the wrong number.

✅ With LOD

You tell Tableau exactly which level to use — separately from the chart.

OUR DATA

USED IN EVERY EXAMPLE THAT FOLLOWS

6 orders · 4 customers · 2 regions

Region	Customer	Order	Sales
East	Alice	#1	150
East	Alice	#2	150
East	Bob	#1	100
West	Carol	#1	400
West	Dan	#1	100
West	Dan	#2	100

EAST TOTAL

400

Alice 300 · Bob 100

WEST TOTAL

600

Carol 400 · Dan 200

⚠ Alice has 2 orders, Bob has 1. This causes the wrong average — as you'll see next.

THE ONE RULE

Every LOD question
has two levels inside it.

One is your chart — What Tableau is already showing.
One is the missing piece — What you actually need.

Find both, then ask where the missing one sits.

INCLUDE

Missing level is more detailed than your chart

EXCLUDE

Missing level is less detailed than your chart

FIXED

Value must stay locked no matter what the chart shows

INCLUDE

WHEN YOU NEED SOMETHING MORE DETAILED

"Average spend per customer,
shown by region"

Your chart is at Region level.

The missing piece is at Customer level which is more detailed. So you INCLUDE Customer.

ORDER
most detailed

CUSTOMER

REGION
your chart

How to remember it

Customer is more detailed (finer) than Region. You're pulling in something your chart doesn't have. You're including a finer level.

INCLUDE IN ACTION

Why the obvious answer is wrong.

Region	Customer	Order	Sales
East	Alice	#1	150
East	Alice	#2	150
East	Bob	#1	100
West	Carol	#1	400
West	Dan	#1	100
West	Dan	#2	100

✕ Without LOD · AVG(Sales) by Region

East shows 133. That's (150+150+100)÷3 = avg of 3 orders. Not customers. Alice's extra order skews the result.

✅ With INCLUDE · average of customers

Sums per customer: Alice=300, Bob=100. Then (300+100)÷2 = 200. East=200, West=300. Correct.

Read as: "For each Customer, sum their Sales. Then average those, by Region."

INCLUDE — THE NUMBERS

133 vs 200. One is a lie.

✕ ORDERS AVERAGED

Order #1

150

Order #2

150

Order #3

100

133

3 orders, wrong denominator

✅ CUSTOMERS AVERAGED

Alice

300

Bob

100

200

2 customers, right denominator

The 67-point gap exists only because Alice has 2 orders.

INCLUDE forces Tableau to group by customer first, so the order count stops mattering.

AVG( { INCLUDE [Customer] : SUM([Sales]) } )

EXCLUDE

WHEN YOU NEED SOMETHING LESS DETAILED

"Each customer's share
of their region's total"

Your chart is at Customer level.

To calculate a %, you need the Region total, which is broader.
So you EXCLUDE Customer from the denominator.

ORDER
most detailed

CUSTOMER
your chart

REGION
← need this

Think of it as zooming out

Your chart is at Customer level. The regional total is one level above. EXCLUDE temporarily zooms out to get that number then brings it back for the division.

EXCLUDE IN ACTION

Each customer's share of their region.

Region	Customer	Order	Sales
East	Alice	#1	150
East	Alice	#2	150
East	Bob	#1	100
West	Carol	#1	400
West	Dan	#1	100
West	Dan	#2	100

✕ Without EXCLUDE

SUM([Sales])÷SUM([Sales]) = 100% for everyone. Dividing a number by itself — there's no way to reach the regional total from a customer row.

✅ With EXCLUDE · Alice=75%, Bob=25%

EXCLUDE drops Customer from the denominator, rolling up to Region. Alice=300÷400=75%. Bob=100÷400=25%. Adds to 100%. ✓

FIXED

WHEN THE VALUE MUST STAY LOCKED

"How many customers spent
over 250 in total, by region?"

The customer total must be calculated at Customer level and stay there, even when Customer leaves the chart.

INCLUDE and EXCLUDE both react to the chart.

FIXED ignores the chart entirely.

Like writing in permanent marker

FIXED calculates its value before Tableau draws the chart. Change the chart, add filters, remove dimensions — that value doesn't move. It's locked.

FIXED IN ACTION

Without FIXED, the chart breaks the count.

Customer	FIXED Total	Over 250?	Region
Alice	300	✓ Yes	East
Bob	100	✗ No	East
Carol	400	✓ Yes	West
Dan	200	✗ No	West

✕ Without FIXED

Remove Customer from view → SUM becomes region total. East=400 → flagged. West=600 → flagged. Regions pass the test, not customers. Wrong.

✅ With FIXED → East: 1, West: 1

Customer totals locked before chart draws. Alice ✓, Carol ✓. Bob and Dan don't qualify. Correct count survives the view change.

COUNTD( IF { FIXED [Customer]:SUM([Sales])}>250 THEN [Customer] END )

THE CHEAT SHEET

Two steps. Every time.

Define what level is on your chart, then what level is missing.

Where does the missing level sit?

More detailed → INCLUDE

Less detailed → EXCLUDE

Must ignore the chart → FIXED

TEST YOURSELF

Read the question. Which LOD?

"Average order value per customer, shown by region"

INCLUDE

"Each product's share of its category total"

EXCLUDE

"Customers whose total spend ever exceeded 500"

FIXED

"Max sales per order, summarised by salesperson"

INCLUDE

"Each region's contribution to the company total"

EXCLUDE

Spotted the pattern? It's always about where the missing level sits.

THAT'S THE WHOLE GAME

Two levels?
Reach for a LOD.

Next time a formula gives you the wrong number, stop and ask:

"What are the two levels hiding in this question?"

tableauuser groupsNairobi

1 /

Mary Kinya Tableau Tuesdays presenter · Kenya TUG

Mary leads Tableau Tuesdays, the weekly community sessions where members break down tricky Tableau concepts together. LOD expressions are her personal favourite feature.

Tableau Tuesdays

Aggregation in Tableau vs SQL

One concept. Two tools. Completely different behavior — and why it matters for every number on your dashboard.

Tonny Kithinji · June 23, 2026

SQL Aggregation Intermediate

TABLEAU TUESDAYS

Aggregation in
Tableau vs SQL

One Concept. Two tools. Completely different behavior

#TableauTuesday

tableauuser groupsNairobi

THE FOUNDATION

What is an aggregation?

Aggregation collapses many rows into one summary.
Examples: SUM, COUNT, AVG, MAX, MIN

RAW DATA (ROW-LEVEL)

3 transactions

Nairobi · KSh 5,000

Mombasa · KSh 3,200

Kisumu · KSh 7,500

AFTER AGGREGATION

1 summary row

Total revenue

KSh 15,700

Aggregation is fundamentally about granularity — what level of detail do you want to see?

HOW SQL AGGREGATES: GROUP BY

In SQL, every aggregation is explicit — You write every rule.

SELECT
    region,
    SUM(sales) AS total_sales,
    COUNT(DISTINCT customer_id) AS customers,
    AVG(sales_amount) AS avg_sales
FROM customer_details
WHERE sale_date >= '2026-01-01'
GROUP BY 1
ORDER BY 2 DESC

region	total_sales	customers	avg_sale
Nairobi	182,400	47	3,881
Mombasa	94,200	28	3,364
Kisumu	61,750	19	3,250

HOW TABLEAU AGGREGATES: AUTOMATICALLY

Tableau aggregates implicitly.
- It scans the dimensions on your Viz canvas (Rows, Columns, Detail) and builds the SQL GROUP BY statement automatically behind the scenes.

YOU DRAG...

Region → Rows
Sales → Columns

That's it. Tableau figures out the rest.

TABLEAU DOES...

SELECT region,
SUM(sales)
FROM data
GROUP BY region

By default, Tableau assumes measures should be totaled.
- When you drag a measure to a shelf, it instantly becomes SUM(Sales).
- Right-click the pill to change to AVG, COUNT, etc.
Tableau protects you from millions of rows.

UNDER THE HOOD

Every SUM is a silent GROUP BY.

Every time you see SUM(Sales) in Tableau, it's silently doing a GROUP BY query — grouped by whatever dimensions are on the current view.

SUM(Budget Margin)

● Sum Average Median Count Count (Distinct) Minimum Maximum Percentile Std. Dev Variance

Changing the aggregation is just telling Tableau how to collapse the rows — the grouping always comes from the view.

COMMON ERRORS

Aggregate vs non-aggregate errors

Tableau: "cannot mix aggregate and non-aggregate"

e.g. SUM([Sales]) / [Sales] — Dividing aggregate by row-level.

You're asking Tableau to look at a summarized group and an individual row simultaneously without clear instructions.
Rule of thumb: Keep your granularities aligned. Aggregate both sides, or leave both unaggregated.

SQL: filtering on aggregates with "WHERE"

WHERE SUM(sales) > 10000 → error.
Use HAVING SUM(sales) > 10000 instead.

Tableau's superpower: LOD expressions

What if you need aggregation at a different grain than the view?
- That's what Level of Detail expressions solve.

SQL EQUIVALENT

Subquery / CTE

WITH cust_totals AS (
SELECT customer_id,
SUM(sales) AS ltv
FROM orders
GROUP BY customer_id
)
SELECT region, AVG(ltv)
FROM cust_totals ...

TABLEAU LOD

FIXED expression

{ FIXED [Customer ID] :
SUM([Sales]) }
// Then AVG() that in
// the view — done.

LOD = Nested aggregation

FIXED — locks the grain regardless of the view.
INCLUDE adds granularity.
EXCLUDE removes granularity.

In SQL, you'd have to write a subquery or CTE for each.

FILTER ORDER MATTERS

How Filters change Aggregation

SQL and Tableau evaluate filters at very different stages. This affects what gets counted.

SQL FILTER ORDER

WHERE runs BEFORE aggregation

TABLEAU FILTER ORDER

FIXED runs before dimension filters!

#1 source of "why is my number wrong?" in Tableau: A Dimension Filter DOES affect INCLUDE/EXCLUDE LODs but DOES NOT affect FIXED LODs.

PERFORMANCE

Aggregate in SQL or Tableau?

Rule of thumb: The closer to the data, the faster.

Do it in SQL when:

You have millions of rows and need a materialized summary table.
Multiple dashboards share the same aggregation.

Do it in Tableau when:

Your dataset is already summarized and reasonable in size.
You need interactive ad-hoc slicing.

💡 Remember: SQL makes aggregation explicit and strict. Tableau makes it invisible and flexible. Master both and you'll always know where your numbers come from.

When you're building a dashboard, do you prefer to aggregate in SQL first or let Tableau handle it?

#TableauTuesday · DATA · COMMUNITY · NAIROBI

tableauuser groupsNairobi

1 /

Tonny Kithinji Tableau Tuesdays presenter · Kenya TUG

Tonny is a data professional in Kenya's energy sector. He first met Tableau in school but fell in love with its power when faced with complex real-world data. His sessions are known for making technical concepts immediately practical.

✍️

Write for us

Got a Tableau tip, career story, or viz breakdown? To publish a post, reach out to us and we will share the template. How? Email us.

Blogs &Insights

Level of Detail Expressions

Level Of Detail,explained.

Your chart can onlyanswer one question at a time.

Like a cashier who onlysees the daily total.

6 orders · 4 customers · 2 regions

Every LOD questionhas two levels inside it.

"Average spend per customer,shown by region"

Why the obvious answer is wrong.

133 vs 200. One is a lie.

"Each customer's shareof their region's total"

Each customer's share of their region.

"How many customers spentover 250 in total, by region?"

Without FIXED, the chart breaks the count.

Two steps. Every time.

Read the question. Which LOD?

Two levels?Reach for a LOD.

Aggregation in Tableau vs SQL

Aggregation inTableau vs SQL

What is an aggregation?

Every SUM is a silent GROUP BY.

Aggregate vs non-aggregate errors

Tableau's superpower: LOD expressions

How Filters change Aggregation

Aggregate in SQL or Tableau?

When you're building a dashboard, do you prefer to aggregate in SQL first or let Tableau handle it?

Write for us

Blogs &
Insights

Level Of Detail,
explained.

Your chart can only
answer one question at a time.

Like a cashier who only
sees the daily total.

Every LOD question
has two levels inside it.

"Average spend per customer,
shown by region"

"Each customer's share
of their region's total"

"How many customers spent
over 250 in total, by region?"

Two levels?
Reach for a LOD.

Aggregation in
Tableau vs SQL