PARMA Presale Pipeline
Preparation cycle 15→7 days, estimation accuracy 60%→85-90%, AI automates 40% of documentation, conversion ≥70%
What doesn't work
Commercial proposal preparation was manual and inconsistent across departments. Average cycle for large proposals — up to 15 business days. Effort estimation accuracy — ~60%. No unified artifact repository or templates, estimates depended on subjective experience, architects and analysts unevenly loaded. Manual document review and approval.
Architectural approach
Standardized presale pipeline: formalized artifact sets per proposal type, clear stage sequence (request → analysis → design → estimation → defense), automated checklists, internal "presale portal" with templates. AI for text preparation, structures, diagrams, and effort analysis — reduces 40% of standard documentation sections.
What made it hard
60% estimation accuracy meant every second project launched with the wrong budget — but presale specialists were confident in their estimates. Introducing AI for documentation generation met skepticism: 'machines don't understand client context.' Standardizing artifacts across different departments (from backend to InfoSec) — each one insisted their presale was unique.
My role & contribution
CTO
Author of the presale section in the production strategy. Designed the presale pipeline: formalized 5 problem areas, identified 20+ root causes, developed 30+ KPIs. Introduced multi-variant proposals and mandatory artifact sets.
How it looks
How it works
5 problem areas → 20+ root causes → 30+ KPIs. Multi-variant proposals (Economy/Standard/Premium) with timeline, scope, cost details. Mandatory artifacts: presale plan, architecture sketch, effort calculation (standard task database), draft implementation plan, consolidated package. CRM integration, OTIF monitoring, account manager training. Architecture council for major proposals.
Why this way
Pipeline model with AI instead of individual specialist "artistry"
Improve individual pre-sale stages (e.g., only proposal templates)
Point improvements: fix one stage but problems flow to the next. Pipeline approach: entire cycle standardized, every stage controlled, AI replaces routine, results depend on the system, not individual people.
Reproducible process: 15→7 days, 60%→85-90% accuracy. Every presale — complete artifact package
Results
- 01
- Preparation cycle: 15 → 7 business days
- 02
- Effort estimation accuracy: 60% → 85-90%
- 03
- AI automation: 40% of standard sections
- 04
- Artifact completeness: 100% for all closed proposals
- 05
- Proposal-to-contract conversion: ≥70%
- 06
- 30+ KPIs for process measurement
Impact on business
Proposal preparation time halved (15→7 days). Estimation accuracy improves from 60% to 85-90%, reducing execution losses. AI eliminates 40% of documentation routine. ≥70% conversion — every proposal systematically developed. Reduced architect workload through templates and automation. Presale becomes a managed project pipeline, not a point of uncertainty.
Algorithms & patterns
Technologies
- CRM
- AI Assistants
- LLM
- KPI Dashboard
- Jira
- Confluence
- Presale Portal