← 004

Novo Nordisk, and the number that was true twice 40 slides · Discover, Define, Develop, Deliver

01 / 40

Case file 004 · Novo Nordisk · on Astrato · reconstructed

Novo Nordisk, and the number that was true twice

Shad Abbaslou · Product Design Lead · Hands-on across AI, Agents and Workflows

18 min read · 9 min skimmed · 40 slides

A reconstruction. The engagement was real and delivered on Astrato; the company figures are public, the data in the product is fictional, and no outcome is claimed.

The group reporting cockpit on a desk: four measures on four bases in one table, a bar chart of group sales on each basis, and beside them the measure tiles and the generated SQL

Case file 004 · Novo Nordisk · on Astrato · reconstructed

A reporting layer for a company whose sales both rose and fell in the same quarter, where the honest answer depends on a basis nobody was showing.

Outcome · up front

The honest version of a number is a slower one

Every figure in that band is public, taken from Novo Nordisk's own reporting, and none of them is a result of this work. I have not put an adoption number or a time-saved number on this page, because I cannot show you the measurement and a number you cannot check does not belong on a site whose argument is that every number carries its source.

DKK 309bnFY2025 group net salesPUBLIC · ANNUAL REPORT 2025
67%of revenue on one moleculeOZEMPIC + WEGOVY, BOTH SEMAGLUTIDE
DKK 60.1bncapex, about 19% of salesPUBLIC · PP&E PURCHASES, 2025
14fill-and-finish sitesUP FROM 11 AFTER THE 2024 TRANSACTION

The brief, in three lines

  • Whata reporting layer for Novo Nordisk, built on Astrato
  • Forfinance and supply planning
  • Problemone quarter’s sales were up and down at once, depending on a basis no screen showed

THE PRODUCT, NOT A PICTURE

A walkthrough, 18 seconds, recorded from the rebuild. The cockpit, step by step: the basis switched and back, the bridge, the segments, and the same screen seen as one segment’s controller. Each step is captioned as it happens, and the transcript has the same words.

Phase 1 of 4 · widenDiscover

What was actually wrong, and for whom: the evidence before any answer.

Two brands. One molecule. Two…FIG. 004 · ONE MOLECULE, TWO…That concentration is the reason…Sales grew 32%. Sales fell 4%."Which number do I put in the…FIG. 005 · FOUR TRUE ANSWERSThe people who already knew which…What a reconstruction can honestly…

The company

Two brands. One molecule. Two thirds of the company.

Novo Nordisk sells about DKK 309bn a year. Ozempic and Wegovy are DKK 206bn of that, roughly two thirds, and they are the same active molecule, semaglutide, in different presentations for different indications. One API, one supply chain, one patent estate, carrying most of a company that employs around 69,000 people and sells in about 170 countries.

The semantic layer canvas: five validated source systems joined once into the model, each read only
The brief, in the product: five sources, read only, joined once.THE PRODUCT, AS IT RUNS

The company

FY2025 GROUP NET SALES, DKK 309,064m, DRAWN TO SCALE OZEMPIC WEGOVY EVERYTHING ELSE DKK 127,089m DKK 79,106m DKK 102,869m BOTH OF THESE ARE SEMAGLUTIDE 67% OF GROUP REVENUE, ONE MOLECULE 1 active ingredient behind two thirds of the revenue, with US exclusivity ending in 2032
Diabetes care, the century-old franchise, is now flat in kroner and price-pressured. Obesity carries the growth. Both run on the same molecule, which is why a question about supply and a question about revenue are the same question here.WHY ONE SUPPLY CHAIN IS THE WHOLE BUSINESS

The company

That concentration is the reason this file exists. When one molecule carries two thirds of the revenue, the questions a finance team asks and the questions a supply team asks stop being different questions. "Can we make more" and "what will we book" are the same sentence read from two ends, and they were being answered in two different systems that did not agree.

The basis problem

Sales grew 32%. Sales fell 4%.

In the first quarter of 2026 Novo Nordisk reported sales up 32% at constant exchange rates and adjusted sales down 4% at constant exchange rates. Neither figure is wrong and neither is spin. They are the same sales measured on two bases, and the gap between them is a DKK 26.8bn provision reversal with no cash attached to it.

"Which number do I put in the deck?" THE QUESTION THE WHOLE PROJECT IS AN ANSWER TO

The basis problem

ONE QUARTER OF SALES, READ FOUR WAYS, ALL OF THEM CORRECT 01 REPORTED, CONSTANT FX +32% the headline, carrying a provision reversal 02 REPORTED, KRONER +7% the same thing after the currency it is booked in 03 ADJUSTED, KRONER −1% strip the reversal and the sign changes 04 ADJUSTED, CONSTANT FX −4% how the business actually traded in the quarter TWO SWITCHES: REPORTED AGAINST ADJUSTED, AND KRONER AGAINST CONSTANT CURRENCY NEITHER WAS ON THE SCREEN A FIGURE WITHOUT ITS BASIS IS NOT A FIGURE, IT IS AN OPINION
Over half of Novo Nordisk's revenue is booked in dollars and reported in kroner, so the currency switch moves every growth rate by several points on its own. The reversal switch moves some of them across zero. A dashboard that picks one and says nothing has not simplified the problem, it has hidden it.FOUR READINGS, TWO SWITCHES, ONE QUANTITY

Research

The people who already knew which number was right

The analysts in finance and supply planning were not confused by the basis problem. They had solved it privately, each in their own spreadsheet, and the cost of that was not error so much as duplication and delay.

The bridge chart: one figure walked from reported to constant FX to adjusted, step by step
The bridge between two bases, as the readers asked for it.THE PRODUCT, AS IT RUNS

Research

What a reconstruction can honestly say about research is limited, and I would rather say the limited true thing than a detailed invented one. The pattern across this kind of engagement is consistent and it held here: the people closest to the data had already built the correct logic by hand, and they distrusted any central dashboard that could not show its working. The question they asked of every new screen was not "is this pretty" but "where did this come from, and on what basis".

Phase 2 of 4 · narrowDefine

The problem narrowed to the one worth solving, and the rules any answer had to keep.

The instinct in a room like that…So the rule the screens enforce is…Five systems that cannot change…FIG. 002 · THE KPI TREEYou cannot fix the systems. You…

The basis problem

The instinct in a room like that is to pick a default and move on. I argued against it, and the argument is the whole design. A default basis is a silent editorial decision taken by whoever built the dashboard, applied to everyone who reads it, and invisible at the moment it matters most. The company had introduced adjusted measures for 2026 precisely because the reported figures had stopped describing trading. Choosing one on a reader's behalf throws that away.

The basis problem

So the rule the screens enforce is narrow and absolute: no measure renders without the basis it was computed on. The basis is not a filter in a panel somewhere. It sits against the number, in the same visual breath, and it is the one control on the screen that restates everything at once.

The blueprint

Five systems that cannot change, drawn as one

THE ESTATE, END TO END: FIVE VALIDATED SYSTEMS, ONE FX TABLE, ONE SEMANTIC LAYER, ONE GATE THE SOURCES FIVE VALIDATED SYSTEMS, GXP SAP ERP, MES, LIMS, QMS, Historian Changing one means revalidating it Read only; nothing writes back THE FX TABLE PERIOD AND BASE RATES The currency basis lives here Constant rates against period rates Twelve thousand rows, not validated THE MODEL THE SEMANTIC LAYER v_group_sales, basis logic once Both segment shapes kept Compiles to SQL on the warehouse THE COCKPIT MATRIX · BRIDGE · CAPITAL Every figure carries its basis One act restates the screen Off-basis tiles marked as such THE READER FINANCE AND SUPPLY PLANNING Under their own warehouse role Every number opens to its SQL The audit log says who changed what THE BASIS reported or adjusted, constant or period FX EVERY FIGURE ON THE COCKPIT OPENS TO THE QUERY THAT MADE IT THE LOGIC SITS ABOVE THE SYSTEMS BECAUSE IT CANNOT SIT INSIDE THEM
Nothing to the left of the model can be touched. The basis is a gate rather than a setting because a figure that has not passed it does not exist on the cockpit.THE ESTATE BEFORE THE SCREENS

The blueprint

WHAT THE COCKPIT IS HELD TO: DESIGN TARGETS, NOT MEASURED OUTCOMES, BECAUSE NONE WERE MEASURED A NUMBER THAT CARRIES ITS BASIS 0 figures allowed on the cockpit without a basis stamp; the outcome section says why no adoption number is claimed TARGET · HELD BY THE SUITE BELOW 1 act TO RESTATE THE WHOLE SCREEN one basis change restates every tile, every chart and the generated SQL TARGET · SPEC NN-01, NN-02 100% OF OFF-BASIS TILES MARKED a per-chart override exists, because sometimes one is needed, and it is visibly marked TARGET · SPEC NN-04 2 SEGMENT SHAPES KEPT history before and after the 1 January 2025 restructure both drawn, with the seam TARGET · THE MODEL SECTION 7 TABLES ON THE MODEL, READ ONLY five validated systems, the FX table and the semantic layer, with nothing writing back TARGET · SPEC NN-05 A TARGET IS A SENTENCE THE PRODUCT MUST SATISFY; A RESULT WOULD NEED A MEASUREMENT I CANNOT SHOW
Targets, not results. The outcome section explains why no adoption number is claimed; what can be claimed is that the rebuilt cockpit satisfies these four sentences every time the site is released.HELD TO, NOT CLAIMED

Modelling across

You cannot fix the systems. You model across them.

WHERE THE LOGIC WENT, GIVEN WHERE IT COULD NOT GO ERP orders, financials MES batch records LIMS test results QMS deviations HISTORIAN sensor readings VALIDATED VALIDATED VALIDATED VALIDATED VALIDATED THE SEMANTIC LAYER joins across all five, and holds the basis logic in one place so every screen inherits the same definition THE SCREENS READ ONLY, DOWNWARD. NOTHING WRITES BACK INTO A VALIDATED SYSTEM.
This is also why a warehouse-native tool was the right one. Astrato stores nothing and extracts nothing: every chart compiles to SQL that runs on the client's own warehouse, under the reader's own warehouse role. The governance is the warehouse's, not the dashboard's, which in a regulated estate is the only arrangement that survives an audit question.THE LOGIC SITS ABOVE, BECAUSE IT CANNOT SIT INSIDE

Phase 3 of 4 · widenDevelop

The answer drawn, tried and argued over, including the parts that were thrown away.

The first version let you set…Nine seconds, recorded from the…Each chart, and what it has to sayThe temptation is to quietly map…A manufacturer spending a fifth of…FIG. 007 · PART SPENT, PART BUILTFrom a validated record to a…The hardest conversation was not…

The basis problem

The first version let you set basis per chart. It demoed well and it was wrong. Two tiles side by side on different bases is a chart that lies by adjacency. Basis went global, and the per-chart override became an explicit, labelled exception that marks the tile.

The cockpit on the adjusted basis: the same quarter's headline with its sign changed, every tile restated
The basis switched: the headline restates, sign included.THE PRODUCT, AS IT RUNS

The basis problem

Nine seconds, recorded from the rebuild. The basis goes from reported to adjusted and group sales turn from +32.1% to −4.2%; to kroner, −1.1%; back to reported, +7.4%; and to constant FX, +32.1% again. Each time every figure, the chart and the generated SQL restate, and the cards whose sign changed say so.

Each chart, and what it has to say

Each chart, and what it has to say

A chart is chosen by the relationship it has to show, in the categories of the Financial Times’ Visual Vocabulary. In the rebuild every mark reads at 3:1 against its card or carries its value in print, so no number depends on telling two colours apart.

The charts in the rebuild, by what they show
ChartWhat it has to showThe form, and why
One measure, every basisDeviation: four readings of the same quantity either side of zeroColumns from a zero line, signed and coloured by direction; the basis in use solid, the other three quiet, every one with its value printed
Group sales growthChange over time, across a change of definitionQuarterly columns with the segment change drawn as a seam, because a line would join two shapes that do not restate into each other
From one reading to the nextFlow: how one figure becomes another, step by stepA bridge, each step named and labelled, so the provision reversal is a bar with a name rather than a gap
SegmentsDeviation, by segmentHorizontal bars from zero, labelled at their tips
Capital projectsPart-to-whole, at magnitude: spent within budgetBars on one kroner scale, spent solid and what remains outlined; rings would show two schemes at 47% and 67% as comparable when one is three times the other

Modelling across

The temptation is to quietly map the old segments onto the new ones and present an unbroken line. That line would be fiction. The model keeps both shapes and the charts draw the seam: before the restructure the series is drawn on the old basis and labelled as such, and the join is a visible discontinuity rather than a smoothed curve. A reader who wants a continuous trend has to ask for a restated one, and gets told what the restatement assumes.

The model canvas with the query panel open: the compiled SQL for the selected measure with its basis and the FX join in the text
Every figure opens to the query that made it.THE PRODUCT, AS IT RUNS

The capital programme

A manufacturer spending a fifth of revenue on concrete

Novo Nordisk spent DKK 60.1bn on property, plant and equipment in 2025, roughly 19% of sales, on API capacity, aseptic production and packaging. At that scale the capital portfolio stops being a finance footnote and becomes an operating system of its own, with its own reporting needs.

The capital view of the cockpit: the capital measures on the current basis
The capital view on the same bases.THE PRODUCT, AS IT RUNS

The capital programme

TWO NAMED EXPANSIONS, BUDGET AGAINST SPENT AT 31 DEC 2025 KALUNDBORG · API DKK 49.6bn DKK 23.1bn SPENT · ABOUT 47% · FULLY OPERATIONAL DURING 2029 HILLERØD · API DKK 15.9bn DKK 10.7bn SPENT · ABOUT 67% · OPERATIONAL DURING 2028 BARS DRAWN TO THE SAME SCALE
Drawn to a common scale, which the project reporting did not do. Two schemes shown side by side as percentage-complete rings read as comparable progress, when one of them is three times the size of the other and a year further out. Percentage hides magnitude, and magnitude is the decision.WHY PERCENT COMPLETE IS THE WRONG SHAPE

The flow

From a validated record to a number someone signs

WHERE A PERSON DECIDES, AND WHERE THE SYSTEM ONLY CARRIES 01 VALIDATED RECORDbatch, test, deviation 02 WAREHOUSElanded unchanged 03 SEMANTIC LAYERjoined, basis applied 04 THE SCREENbasis shown, always 05 OPEN THE SQLa person checks 06 STATE THE BASISa person signs EVERY QUERY RUNS ON THE CLIENT’S OWN WAREHOUSE, UNDER THE READER’S OWN ROLE. NOTHING IS EXTRACTED AND NOTHING IS STORED. THE TWO MARKED STAGES ARE DELIBERATE FRICTION. THEY ARE WHERE ACCOUNTABILITY ATTACHES.
Most of this chain is plumbing. The two accented stages are the design: a reader can always see the query, and a reader who publishes a figure states the basis they are publishing on, which is recorded beside it.FRICTION WHERE IT BUYS ACCOUNTABILITY

Research

The hardest conversation was not about charts. It was about whether a dashboard should be allowed to show a figure that a reader is not entitled to see the derivation of. I do not think it should. If the derivation is too sensitive to show, the figure is too sensitive to put on a shared screen.

Phase 4 of 4 · narrowDeliver

What the rebuild is held to, what each decision moved, and what I would do differently. No outcome is claimed.

The same quarter’s sales could be…An early build let each chart…The 1 January 2025 segment…Capital projects were reported as…Five systems that cannot change…The honest version of a number is…What I would do differently

What was broken · what changed · 01 of 05

Was broken

The same quarter’s sales could be quoted as up 32% or down 4%, and both were correct, because reported and adjusted figures on two currency bases circulated without labels.

What I changed

No measure renders without its basis. The control is global, sits against the figures rather than in a panel, and restates the whole screen in one act.

Why it holds

The company introduced adjusted measures because the reported ones had stopped describing trading. A dashboard that picks one basis silently throws that distinction away.

What was broken · what changed · 02 of 05

Was broken

An early build let each chart carry its own basis, which produced screens where two adjacent tiles measured different things and looked like a comparison.

What I changed

Basis went global. A per-chart override still exists, because sometimes you need one, but it visibly marks the tile as off-basis.

Why it holds

Adjacency is an argument. Two tiles side by side assert comparability whether or not the data supports it, and no caption undoes that.

What was broken · what changed · 03 of 05

Was broken

The 1 January 2025 segment restructure meant history did not restate cleanly, and the easy fix was to map old segments onto new ones and draw one smooth line.

What I changed

The model keeps both shapes and the chart draws the seam. A continuous trend has to be asked for, and arrives with its restatement assumptions stated.

Why it holds

A smoothed line across a definition change is the one chart nobody can audit later, because the discontinuity it hides is exactly what a reviewer would look for.

What was broken · what changed · 04 of 05

Was broken

Capital projects were reported as percentage-complete rings, which made a DKK 49.6bn scheme and a DKK 15.9bn scheme look like comparable progress.

What I changed

Budget and spend drawn to a common scale, with the operational date on the same row.

Why it holds

Percentage hides magnitude, and magnitude is what the decision turns on. Half of fifty billion and two thirds of sixteen are not the same news.

The spec

Five systems that cannot change, drawn as one

NN-01No measure renders without its basis.
NN-02Switching the basis restates the headline, including its sign.
NN-03The generated SQL recompiles for the basis and for the FX join.
NN-04A tile taken off the global basis says so on its face.
NN-05The model shows every source, read only, joined once.
NN-06The cockpit works on a phone, in the dark setting as well as the light one.

Outcome

The honest version of a number is a slower one

Held to the design targets in the blueprint; no outcome was measured
It worked ifResultStatus
No figure appears without its basis0 figures on the cockpit without a basis stampHeld in the rebuild
One act restates the whole screen1 basis change restates every tile, every chart and the generated SQLHeld in the rebuild
An override is never silent100% of off-basis tiles markedHeld in the rebuild
Readers adopt it and save timeNot measured, so not claimedNot measured

Outcome

What I would do differently

What I would tell another designer taking this on: the instinct to simplify is the enemy here. In a company reporting in one currency and earning in another, through a year with a provision reversal large enough to flip the sign of its own headline, the simplification is the error. The job is not to produce one number. It is to make the four honest ones legible, and to make it impossible to quote one without saying which.

  1. Agree the measurement before it ships. This page carries no adoption or time-saved number because I cannot show the measurement. A reporting layer about honest numbers should have shipped with its own.
  2. Put the click in front of readers early. Surfacing the basis costs every reader a click and a moment of thought. I would have tested that trade with finance and supply planning in the first prototype, not argued it afterwards.

THE PRODUCT, NOT A PICTURE

The finished product, running

The group reporting cockpit on a desk: four measures on four bases in one table, a bar chart of group sales on each basis, and beside them the measure tiles and the generated SQL
On a phone: Both basis switches stay at the top. The matrix becomes four cards, every basis a tile to tap, and the inspector opens as a sheet called Measures.
On a phoneThe basis cockpit

Case file 004 · Novo Nordisk · on Astrato · reconstructed

Questions

The honest version of a number is a slower one

A reconstruction. The engagement was real and delivered on Astrato; the company figures are public, the data in the product is fictional, and no outcome is claimed.

The other decks001 · OI·DS002 · Astrato003 · Safe Space005 · JBS

Notes