AI Data Transformation Partnership
[Client] is becoming an Entrepreneur Operating System. That starts with a data foundation that lets you see every customer clearly and use AI to help them run their business.
Mammoth Growth will centralize all core data into Snowflake, model it using dbt with a medallion architecture, embed governed metric definitions via MetricFlow, and produce a canonical data foundation with separated entities, resolved customer identity, and application-ready views.
Built This Before
5.6M customer records unified for Client X. Identity architecture from scratch for Client Y. Multi-jurisdiction compliance for Client W.
Design for Durability
One well-designed silver layer answers six or seven business questions at the gold layer. Functionally decomposed, highly extensible.
Semantic Layer Standard
Every model ships with governed metric definitions via dbt MetricFlow. Revenue, LTV, churn — defined once, served consistently.
AI-Powered Delivery
Our agentic framework compresses timelines 4-5x vs. traditional engagements. Data profiling, model scaffolding, test generation, and documentation are produced as natural byproducts of the development process.
Quality-Gated SDLC
Every model passes through planning, automated code review, and validation against 56 dbt-specific rules before it ships. Medallion tier placement, naming conventions, test coverage — enforced automatically.
Parallel Agent Execution
Multiple AI agents build independent models simultaneously. A 5-model feature build that takes a single engineer a week takes our system minutes — with tests, docs, and metric definitions included.
Persistent Context
Every task is tracked in a bundle — requirements, architecture decisions, plans, and validation results. Context does not reset between sessions or team members.
Compound Learning
A dedicated internal AI development practice — building and refining the contexts, skills, agents, and quality rules that power our delivery. It compounds with every engagement.
Client X — Unified Customer Intelligence 90 days
Client Y — Identity Architecture from Scratch 25K+ SKUs
- Customer identity fused to ordersA customer is not a first-class entity — it is an order with a name attached. [Client] cannot answer "how many customers do we have" today.
- Revenue metrics differ by toolLTV, churn, and revenue pacing are calculated differently in every system. No single source of truth.
- Compliance data is disconnectedRegulatory Authority records, filing deadlines, and jurisdictional requirements are not connected to the customer or company profile.
- 20 years of legacy architectureThe canonical data model has not been separated from the transactional schema. Edge cases compound.
- No foundation for AI featuresEight engineering squads are building agentic workflows, but there is no canonical, governed data layer for them to consume.
Before
- Customer = an order with a name
- Revenue calculated differently per tool
- Compliance disconnected from profile
- One person with 3 entities = 3 customers
- No AI-ready data layer
After
- Customer as first-class entity
- Governed metrics via MetricFlow
- Compliance intelligence per jurisdiction
- Multi-entity customer visibility
- Canonical data for AI agents via MCP
The Core Foundation is the sequential critical path — each milestone builds on the one before it. The Accelerators are discrete projects that extend the foundation, prioritizable in any order after Milestone 2.
| RFP Outcome | Delivered By | Type |
|---|---|---|
| #1 Canonical Data Model | Embedded across Core Foundation | Core |
| #2 Unified Customer Identity | Milestones 2, 3, 4 + Full Journey | Core Accel |
| #3 Revenue Pacing & Financial Visibility | Milestones 2, 3, 4 | Core |
| #4 Proactive Compliance Intelligence | Milestone 4 + Full Journey | Core Accel |
| #5 360-Degree Customer View | Milestones 2–4 + Business Systems | Core Accel |
| #6 AI-Ready Data Foundation | Canonical design + AI Enablement | Accel |
Core Foundation
Sequential critical path. 4 milestones building on each other. Infrastructure, identity resolution, company mapping, jurisdiction mapping.
Accelerators
Discrete projects. Full customer journey, business systems activation, AI product enablement. Prioritizable post-Milestone 2.
13 weeks — a 13-week engagement. Each milestone builds on the last. Expand any milestone to see its deliverables.
M1: Infrastructure Live, Data Loaded
Snowflake, Fivetran, and dbt initialized. Core Legacy Database data landed. Bronze layer operational. Data profiling complete.
Deliverables 8
- Provisioned and configured Snowflake environment
- Initialized dbt project with CI/CD pipeline and coding standards
- Configured Fivetran pipelines for all supported connectors
- Data ingested into Snowflake
- Bronze layer models for core data domains
- Silver layer models in progress
- Data profiling report
- Validated and refined canonical model design
M2: Transactions Converted to Customers
Person, Account, Order, and Subscription as independent entities. Deterministic identity resolution. A customer becomes a first-class entity.
Deliverables 8
- Completion of Silver layer for canonical models
- Person, Account, Order, Subscription as independent entities
- Legacy primary key mapping resolved
- Deterministic identity resolution model
- Customer segmentation: acquisition, product mix, service tier
- Revenue pacing by product, channel, state
- Historical transaction LTV by customer
- Semantic layer via dbt MetricFlow
M3: Customers Mapped to Companies
Company as its own entity. Person-to-Company relationships modeled. Multi-company customers visible for the first time.
Deliverables 3
- Gold layer: Company as independent entity
- Person-to-Company relationship model (one-to-many)
- Multi-entity segmentation and LTV
M4: Companies Mapped to Jurisdictions
Jurisdiction as its own entity. Regulatory Authority matching. Compliance rules codified. Customer lifecycle staging assigned.
Deliverables 4
- Gold layer: Jurisdiction as independent entity
- Company-to-Regulatory Authority entity matching model
- Customer lifecycle staging model
- Complete canonical data foundation documentation
Phase Close: Hardening & Knowledge Transfer
Production optimization, knowledge transfer sessions, architecture decision records, operational runbooks, and Q3 roadmap.
Deliverables 5
- Production-hardened models with performance optimization
- Knowledge transfer sessions
- Architecture decision records (ADRs)
- Operational runbooks
- Q3 roadmap recommendation
Full Customer Journey
Upper funnel to in-app journey. Anonymous-to-known identity resolution. Marketing spend connected to LTV.
Business Systems Activated
Customer 360 pushed to HubSpot, Amplitude via reverse ETL. Platinum application-ready views for platform engineering.
AI Product Enablement
Canonical data exposed via MCP for AI agent consumption. Integration points for [Client]'s LLM Gateway architecture.
At a glance: 13-week engagement. Accelerators can begin as early as Milestone 2 completion.
Two architects in parallel across workstreams. The people in discovery are the people building your models. All consultants 8+ years experience.
| Role | Name | Responsibility | Allocation |
|---|---|---|---|
| Executive | Executive Sponsor | Strategic alignment, engagement oversight | As needed |
| Lead | Engagement Lead | Delivery accountability, stakeholder management | Half to Full |
| Architect | Lead Architect | Architecture, canonical model, MetricFlow, MCP, quality | Half-time |
| Architect | Senior Architect | Architecture, identity model, MetricFlow, MCP, quality | Half-time |
| Sr. AE | Senior AE 1 | dbt modeling, entity buildout, identity resolution | Half to Full |
| Sr. AE | Senior AE 2 | dbt modeling, revenue, compliance, jurisdictional | Half to Full |
Included
Full canonical data foundation build
Entity separation & identity resolution
dbt MetricFlow semantic layer
CI/CD pipeline & testing framework
Knowledge transfer & documentation
Architecture decision records
Q3 roadmap recommendation
Client Responsibility
Snowflake compute & storage
dbt Cloud licensing
Fivetran subscription & connectors
Attribution.app licensing
Amplitude licensing
Regulatory Authority data procurement
Git repo provisioning
An estimated 1,920 hours of concentrated, senior-level effort — parallel atomic working teams compressing a year of output into a single calendar quarter.
Scope boundaries — what's not included
- Application layer development (Legacy Database, Laravel/PHP, Vue.js)
- NetSuite implementation or integration
- HubSpot configuration or migration
- Dashboard or report creation
- Regulatory Authority data acquisition
- Rules Engine maintenance or replacement
- Predictive LTV or churn modeling
- Real-time streaming, ODS, or caching layer
Mammoth's scope is the Snowflake warehouse layer: ingest (Fivetran), model (dbt), govern (MetricFlow), and make data available for downstream consumption.
This proposal is confidential and intended solely for [Client]'s evaluation of Mammoth Growth as a data transformation partner.