Concept demo · built on Swissmed brands

A market-intelligence engine for Swissmed

Not another dashboard. A system that turns your sell-in, sell-out and market data into answers — and an AI analyst that explains why the numbers moved and what to do, every claim cited back to the figures.

Prepared forSwissmed · Azerbaijan
ScopePharma commercial intelligence
StatusWorking prototype + build plan
Swissmed Market IntelligenceFigures illustrative — real Swissmed brands, simulated values
The problem today

The data exists. The intelligence doesn't.

Swissmed's commercial picture lives in a sprawling Excel workbook that takes an analyst days to wrangle — and still can't answer the question that matters: why?

What the real file looks like

!
636 columns, two-row headersMonths × quarters × full-year × YTD, each split TGT/ACT across 2022–24.
!
1,218 broken cells & nested subtotals#DIV/0! errors and “TOTAL” rows that silently double-count a naive sum.
!
Sell-out only, no geographyOne distributor's view. No market share, no regional breakdown, no “why”.
!
Manual, fragile, slowEvery refresh is hand-built. One mis-summed column and the number is wrong.
$61.7M

what a naive sum of the real workbook reports

~2×

over-counted — because subtotals are mixed into detail rows. This is the #1 correctness risk, and we've already mapped it.

The problem todayWorkbook stats verified from Swissmed's Dec'23 file
What we provide

A business-intelligence engine with a brain

Three layers working together: clean your data once, see everything on live dashboards, and ask an AI analyst anything — grounded in your own figures.

1 · Clean data layer

Upload your Excel as-is. The engine parses the 636-column mess, strips subtotals and errors, and resolves dirty SKU strings into brands, dosages and units — with a human review step. The number is right, every time.

2 · Live dashboards

Sell-in vs sell-out, market share, geography, channel stock and the brand-growth waterfall — all the views in your requirements doc, in packs and value, across every time grain.

3 · AI analyst

The differentiator. Ask “why did Nakom decline?” in plain language and get a cited answer. It also surfaces risks and opportunities before you ask — grounded in your data, never invented.

What we provideMaps 1:1 to Swissmed's requirements brief
The engine

Six modules, one source of truth

Every view your team asked for — filtered live by year, area, business unit and period.

Overview

KPI strip + brand-growth waterfall + SI-vs-SO at a glance.

Sell-in / Sell-out

TGT, ACT, vs-target and vs-PY by brand, in packs and value.

Market share

Share by brand and its movement vs prior year, in points.

Geography

Azerbaijan → Baku → regions → districts drill-down.

Buy-ins & stock

Channel inventory, months of cover and auto-detected buy-ins.

Data sources

Every feed with freshness & coverage — what's live, what's next.

The engineLive in the prototype — click through after this deck
What it surfaces

FY 2023 at a glance — and the story behind it

The engine doesn't just show the headline. It connects sell-in, sell-out and stock into one narrative.

$29.8M
Sell-out value · +14% PY, −3.9% vs target
$32.4M
Sell-in value · +19.6% PY
18.4%
Market share · +1.2pp PY
2.1 mo
Channel stock · overstock risk

What drove growth

Ferrum Lek carried the year — +$1.9M sell-out and +2.1pp share, strongest in Sumgait & Sheki-Zaqatala. Dermazin and Diclac added another $1.3M. Real OTC demand, not just stocking.

What to watch

Nakom is the one to worry about — the only declining line, off $0.4M and −1.4pp share, concentrated in Baku. Meanwhile sell-in ran $2.6M ahead of sell-out — stock is building.

What it surfacesIllustrative figures on real Swissmed brands
The differentiator

An analyst that explains, not just displays

Ask in plain language. Every number traces back to a source — if it can't be grounded, it says so rather than inventing a figure.

Why is stock piling up?
Two numbers tell it: sell-in hit $32.4M while sell-out was $29.8M — a $2.6M gap that stays in the channel. Most traces to a September buy-in at Distributor A (+$1.4M, sell-out flat), pushing cover to 2.1 months vs the 1.5 target. Recommendation: hold Ferrum Lek sell-in flat through Q1 and let sell-out draw it down.
SI $32.4MSO $29.8MDistributor A
Surfaced proactively — before you ask
Risk

Stock building faster than it sells. SI $2.6M ahead of SO, mostly a Sep buy-in. Slowing Ferrum Lek SI next quarter avoids a write-down.

Watch

Nakom is the brand to worry about. SO −$0.4M YoY, −1.4pp share — the only declining line, weak in Baku.

Growth

Ferrum Lek carried the year. +$1.9M and +2.1pp share, strongest in the regions — OTC pull, not stocking.

The intelligence layerGrounded, cited, deterministic detection — no hallucinated findings
How your data becomes intelligence

From a messy upload to a trusted number

A repeatable pipeline. Re-uploading a file never double-counts — nothing reaches the dashboards until a person confirms the mappings.

01
Upload

Drop the Excel/CSV in. Stored securely, isolated per tenant.

02
Parse

Detect the two-row header, unpivot 636 columns, drop subtotals, coerce errors.

03
Map & review

AI suggests brand/dosage/pack from each dirty SKU; an analyst confirms once.

04
Load

Clean facts into a warehouse — idempotent, fully traceable to source.

05
Intelligence

Dashboards refresh; the analyst re-sweeps for new risks & opportunities.

The pipelineSame pipeline later accepts live distributor / ERP / panel feeds
Data roadmap

From one distributor's invoices to the whole market

Today the file shows sell-out only — about three-quarters of the picture. Each feed we connect widens what the engine can see.

ERP — sell-in / invoicing
Live
Distributor sell-out (×3)
Live
Pharmacy / chain panel
Lagging
Market panel (IQVIA-style)
Lagging
Buy-in / tender log
Manual
Competitor price watch
Lagging
Wholesaler stock feed
Not live
74%

of national sell-out visible today

Connecting the pharmacy panel and a market subscription lifts visibility toward full-market share & competitor view — the part the current Excel simply can't show.

Data roadmapMarket share requires an external panel Swissmed supplies — modelled until then
Delivery plan

Built in phases, value at each step

You see working software early — dashboards on your real data before the analyst, the analyst before connectors.

Phase 0Foundations
Secure multi-tenant base

Accounts, roles, and strict data isolation — your data is provably yours alone.

Phase 1Data → dashboards
Your file in, dashboards out

Upload → clean → review → load; Overview, SI/SO, Geography and the waterfall on real data.

Phase 2The analyst
Grounded AI analyst + proactive insights

Ask anything, get cited answers; nightly sweep surfaces risks and opportunities.

Phase 3Share & docs
Market share, documents & buy-in detection

Add the market panel, ingest reports for the analyst to cite, auto-flag channel loading.

Phase 4Scale
Live connectors & commercialisation

Replace uploads with live feeds; same engine, no rebuild.

Delivery planEach phase has a concrete “done” gate before the next begins
Why this is credible

Engineered for trust, not just for show

The hard parts — the ones that make pharma BI fail — are designed for, not glossed over.

We've already read your real workbook.636 columns, 1,218 broken cells and the nested-subtotal trap are verified facts in our build spec — not assumptions.
The AI can't make numbers up.Risk detection is deterministic code; the AI only explains. Every figure is cited or it isn't shown.
Your data is isolated by construction.Database-level tenant isolation — even the AI runs inside your boundary and cannot read anyone else's data.
Honest about what's real.Modelled market share is labelled as modelled; uploaded data is labelled as uploaded. No blurring the two.
Why this is crediblePrinciples drawn from the technical build spec
Next steps

Let's point the engine at Swissmed's real data.

01Confirm one real totalOne hand-checked FY figure to calibrate the parser against.
02Walk the live prototypeClick through the working demo, on your brands.
03Start Phase 1Your Dec'23 file in, dashboards out — on real data.

A dashboard shows you what happened. Swissmed Intelligence tells you why — and what to do next.

Swissmed Market IntelligenceConcept demo · figures illustrative
1 / 11