FEATURED CASE STUDY 02 · PRICING INTELLIGENCE

BOQ Pricing IntelligenceHistorical pricing, made usable.

An internal tender-pricing decision-support system that turns historical contract and vendor data into searchable, comparable pricing evidence for new BOQ items, individually or in bulk.

Purpose: Tender Pricing Support
Knowledge Base: Historical BOQ Items
Role: Cost Control + System Design
Delivery: Python · Flask · Standalone Windows App

BOQ Pricing Intelligence semantic search showing historical pricing matches, vendors, contracts, rates and match confidence — Match confidence indicates similarity to the requested BOQ item. Commercial pricing judgment remains with the reviewer.

CONTEXT & ORIGIN

Historical prices existed. The problem was finding the right evidence when a new tender arrived.

Contract archives contained years of BOQ descriptions, vendors and agreed rates, but the value of that history depended on being able to retrieve a genuinely comparable item quickly. Different wording, specifications, units and file structures made simple lookup unreliable.

BOQ Pricing Intelligence was built to convert that fragmented contract history into a structured internal pricing knowledge base, then place retrieval, comparison and review inside one practical tendering workflow.

PositioningCost-control knowledge translated into practical tender-pricing intelligence and commercial decision support.

THE PRICING PROBLEM

A historical rate is only useful when the underlying item is genuinely comparable.

The system was designed around the gap between having contract data and being able to use that data safely and quickly during tender pricing.

Fragmented history

Pricing evidence was spread across contracts, vendors, projects and Excel files rather than one reusable reference base.

Description mismatch

The same commercial item can appear under different wording, abbreviations, typos and levels of specification.

Specification risk

Sizes, units and technical parameters matter. A close text match can still be commercially misleading if the specification differs.

Tender scale

Pricing one line is easy to review manually. Pricing an entire client BOQ requires a repeatable workflow for matches, exceptions and exports.

PRICING INTELLIGENCE ARCHITECTURE

Contract history becomes structured evidence for new tender decisions.

The system separates data structure, matching logic and commercial review so that historical prices support a decision rather than replace one.

01 · HISTORICAL DATA

Historical BOQ line items
Contract references
Vendor records
Unit rates
Work types
Project references
Categories & item groups
Signature dates

02 · INTELLIGENT RETRIEVAL

Search & Match

Semantic and keyword retrieval, supported by filters, specification anchors, unit normalization and ranked similarity.

03 · COMMERCIAL REVIEW

Comparable historical matches
Vendor & contract references
Rate range and statistics
Match confidence
Bulk BOQ review
Manual-pricing exceptions

04 · TENDER SUPPORT

Pricing Evidence

A reviewed reference set that helps pricing teams benchmark new BOQ items and identify where manual pricing is still required.

SYSTEM EVIDENCE

From one pricing question to an entire BOQ review.

The application supports both focused item research and batch tender workflows, while keeping the underlying historical evidence visible to the reviewer.

BOQ semantic single-search results with historical vendor rates and match confidence — 01 · SINGLE ITEM SEARCH
**Comparable historical references**Search a new BOQ description, filter the evidence set and review matching vendors, contracts, dates, rates and similarity confidence.Open full-size view ↗

BOQ bulk-search workflow for selecting and reviewing client BOQ items before matching — 02 · CLIENT BOQ INTAKE
**Review before matching**Upload an Excel BOQ, choose the relevant sheet, inspect the parsed items and exclude rows that should not enter the pricing search.Open full-size view ↗

BOQ bulk-search pricing results showing multiple tender items and historical matches — 03 · BULK PRICING REVIEW
**Entire BOQs, not just one search**Run the confirmed BOQ through the matching engine, review top references item by item and separate weak or missing matches for manual pricing.Open full-size view ↗

BOQ Pricing Intelligence dashboard summarizing items, vendors, contracts, trends and dataset concentration — 04 · PRICING KNOWLEDGE BASE
**History becomes visible**Use the accumulated database to inspect item volumes, vendors, contracts, work types, recurring items, price trends and vendor concentration.Open full-size view ↗

MATCHING LOGIC

Similarity helps retrieve evidence. Commercial logic keeps the evidence relevant.

The search engine combines several matching layers because BOQ descriptions are rarely clean enough for one technique to be reliable on its own.

Semantic search

Sentence-transformer embeddings retrieve similar descriptions even when wording differs.

Keyword precision

Character n-gram TF-IDF supports exact codes, part numbers and specification-heavy searches.

Specification anchoring

Numeric and size tokens such as DN, mm, kV, kW, PN and cable sizes influence ranking.

Unit awareness

Normalized units strengthen comparable matches and penalize incompatible ones.

Commercial filters

Work type, project, category, subcategory and item group narrow the evidence set.

Review confidence

Scores describe match similarity, not whether a quoted rate is commercially correct.

HOW IT WORKS

Five steps from contract history to reviewed pricing evidence.

The workflow keeps retrieval and judgment separate: the system finds and organizes evidence, while the pricing reviewer decides how that evidence should be used.

01
Structure
Import and normalize historical contracts, vendors and BOQ items into a reusable pricing knowledge base.
02
Search
Use semantic or keyword search with work-type, category, project and unit filters to retrieve relevant references.
03
Match
Strengthen retrieval with typo handling, numeric and size/spec anchoring, and unit-aware ranking.
04
Review
Compare vendors, contracts, dates, rates and match confidence before accepting a reference.
05
Export
Move reviewed results into Excel outputs, with unmatched or low-confidence items isolated for manual pricing.

WHAT THE SYSTEM ENABLES

Reuse contract history instead of rediscovering pricing evidence tender by tender.

Historical retrievalFind relevant past contract items without depending on exact wording or remembering where a rate was stored.
Benchmarking evidenceCompare rates together with vendor, contract, date, work type and item context before using a reference.
Bulk tender reviewApply the same retrieval logic across an uploaded client BOQ rather than repeating one manual lookup at a time.
Exception visibilitySeparate low-confidence or unmatched items so they remain visible for manual pricing instead of being hidden behind automation.

MY ROLE

Commercial experience shaped the system. Technology made that experience reusable.

The project started from a cost-control and tender-pricing problem rather than a software brief. I designed the data structure, search workflow, matching logic and review outputs around the way historical pricing evidence is actually used during tender preparation.

Professional lens: Cost Control & Tender Pricing
System role: Designer & Developer
Technology: Python · Flask · SQLite · Excel · Semantic Search
Delivery: Internal browser interface + standalone Windows application

CASE STUDY 02