Themis – automotive market analysis (Automotive Market)

Themis is a proprietary, modular analytical system developed over many years, whose task is to turn dispersed, unstructured market data into a coherent, comparable decision-ready picture. Automotive Market is the application of this platform to the new-car market: every day the system delivers an up-to-date view of prices, discounts and offers from multiple brands, based on actual manufacturer price lists.

The engine behind Themis was not built from scratch for this project. It originates from tools developed for financial analysis and due-diligence projects — starting with the structuring of invoice data for reporting purposes. Over the years it grew into a multi-purpose analytical platform; the first major application was Themis – real-estate and location market analysis, and Automotive Market is another branch of the same architecture, enriched with dedicated custom OCR and brand-specific parsers for automotive documents.

The problem that Automotive Market solves

In the automotive market, prices, discounts and financing terms change dynamically. Price lists arrive as PDFs, information is scattered across manufacturer websites and folders, and comparing offers between brands is time-consuming and error-prone.

What reality looks like without the system

• Market analysis takes days instead of minutes.
• A price change is often noticed only after the fact.
• Discounts and financing are compared manually, model by model.
• There is no history of snapshots, so it is hard to prove the direction of changes.
• Every department works on different data and a different update date.

What Automotive Market changes

The system automatically retrieves and parses price lists from multiple brands — including documents whose structure does not allow simple text extraction, which is why a key element of the module is custom OCR tailored to the specifics of automotive price lists. Data are normalised to a common schema and stored as versioned snapshots, using the same mechanism that Themis employs in the real-estate module. The team receives the same coherent market view: catalogue prices, promotional prices, discounts, financing parameters and offer statuses.

From manual research to a ready analytical feed.

Data pipeline architecture

Layer 1: acquisition of price lists and offers
The system uses web sources and SharePoint. For each brand, retrieval and file-filtering rules are defined (including model-year filters), so that only current price-list documents reach the analysis.

Layer 2: custom OCR, routing and brand-specific parsers
Every PDF is directed to a dedicated brand parser, supported by custom OCR where documents do not allow standard text extraction. This makes it possible to handle different document formats stably and maintain high data-extraction quality even when manufacturers change document layouts.

Layer 3: snapshots and change history
After parsing, data go into unified snapshot tables — the same model on which Themis is based in the real-estate module. The system creates an import history, detects duplicates via SHA-256 and records differences between successive offer publications: new items, price changes and dropped offers.

Layer 4: export and distribution of insights
Results are made available as a ready analytical feed for Excel for analysts, and ultimately also via an API that powers the React dashboard. Data do not end in the database — they continue to work in business decisions.

What Automotive Market calculates and monitors

Vehicle offer standardisation
The system maps to a single model data such as: brand, model, variant, body style, powertrain type, power, gearbox, offer status and validity dates.

Price and discount analysis
The common schema covers catalogue and promotional prices, discount value in PLN and as a percentage, as well as on-road price variants. This makes it possible to compare pricing strategies across brands and segments.

Comparison of financing terms
The system stores financing parameters, including 50/50, 3×33, leasing variants, periods and down payments. This allows assessment of the real attractiveness of an offer, not only the headline price.

Detection of changes between snapshots
Events are identified automatically: new, price_change and dropped. The team receives an immediate answer to what has really changed on the market since the last publication.

Data quality and reliability
Built-in document-duplicate control and import versioning enable auditability. Every record has a source, a snapshot date and a provenance context.

Themis as a platform – not a single project

Automotive Market demonstrates a complete approach to the problem: from the idea and architecture design, through acquisition and custom OCR of unstructured PDF documents, to change analysis and delivery of a ready analytical feed — to Excel or as a React dashboard. It is also proof that the Themis engine is not a one-off solution under a single branding: the same core — document extraction, standardisation, snapshots and change analysis — today serves both the real-estate market and the automotive market, with roots reaching back to financial-analysis and due-diligence projects.

Practical effects of the implementation

Faster reaction to price changes. Changes are detected between snapshots without manual document comparison.

One truth for the whole team. Sales, purchasing and analytics work on the same data model and the same update date.

Less manual work, more decisions. Elimination of dozens of hours per month lost on searching for and manually normalising price lists.

Better brand and model benchmarking. A common schema enables comparisons that were previously time-consuming or impossible.

Readiness for further AI/ML layers. Clean, historical snapshot data form the base for subsequent recommendation and predictive models.

Technology – what stands behind the system

FAQ – Themis Automotive Market