Wedge

Target Shopify-native D2C brands with $10M–$50M revenue, focusing strictly on marketing attribution and cohort LTV analysis. This niche utilizes highly standardized database schemas, allowing for zero-setup deployments and immediate proof of value. Expand horizontally into inventory and supply chain forecasting, then target adjacent verticals like mid-market B2B SaaS.

Timing

LLMs equipped with Python sandboxes and text-to-SQL capabilities now reliably translate natural language into complex multi-table joins, execute the code, and return mathematically accurate visualizations without human intervention.

Why This ICP

Mid-market e-commerce brands generate high volumes of structured data across standard platforms like Shopify, Meta, and Klaviyo, but operate on tight margins that preclude hiring full-time data analysts.

Size Of Prize

~40,000 mid-market e-commerce and SaaS companies in the US × ~$20,000/yr allocated to fractional data analysts or premium BI seats = ~$800M annual prize.

Gap Narrative

Mid-market operators lack dedicated data teams but face complex attribution and inventory questions across siloed platforms. Current BI tools demand SQL literacy or rigid dashboard configuration, leaving business leaders unable to ask conversational questions and receive immediate, synthesized answers.

Defensibility

The core text-to-SQL engine is fundamentally a commodity. Defensibility stems entirely from localized workflow lock-in and context accumulation; as the system learns a company's custom metric definitions and embeds itself in daily chat channels, replacing it requires retraining a new system on the company's specific business logic.

Why This Thesis

The Service-as-Software model delivers the actual output, such as a written insight, a chart, or a spreadsheet, matching the exact workflow of requesting an analysis via Slack rather than forcing the buyer to learn a new software interface.

Hardest Part

Guaranteeing deterministic accuracy in calculations and data transformations generated by probabilistic models. The system must never hallucinate a metric or misalign a time series across disparate internal data structures.

Min Viable Scope

Automate month-end budget versus actuals reporting for B2B software companies using standardized CSV exports. Deliberately exclude predictive forecasting, scenario planning, and direct ERP API integrations.

Cold Start Problem

The system lacks context on idiosyncratic internal data schemas and undocumented business logic used by specific finance teams. Break this by requiring early design partners to provide historical raw data dumps alongside their finished reports to train the mapping logic.

Time To First Value

2 to 4 weeks of onboarding to map internal data schemas and validate the first automated variance report

Data Moat Available

true

Technical Difficulty

High

SAM

~$800M-1.2B targeting US and European middle-market private equity firms

SOM

~$20-50M realistic 3-year capture through direct sales to early-adopter deal teams

TAM

~15k global private equity and alternative investment firms × ~$200k/yr baseline junior analyst or outsourced research spend ≈ $3B

Growth Rate

~12-18%/yr, driven by rising junior finance talent compensation and immense pressure to accelerate deal screening and diligence cycles

Paid Comparable Spend

~$150k-250k/yr in fully-burdened compensation per junior investment professional, or ~$80k-120k/yr for retained offshore financial research teams

Kill Thresholds

• CAC exceeds $15,000 after 90 days
• Week-four active user retention drops below 20%
• Human correction rate remains above 30% on generated financial models
• Average contract value falls below $25,000 annualized

Leading Metrics

• Time to generate initial diligence teardown
• Target companies processed per week per firm
• Client acceptance rate of unedited output
• Human-in-the-loop escalation percentage

What Proves Right

Private equity deal teams deploy the service to screen at least five target companies per week without requesting manual rework. Annual contracts close at the $40,000 to $60,000 price point, directly replacing offshore research retainers. User cohorts achieve a 60% week-four retention rate for generating preliminary financial models and market maps.

What Proves Wrong

Deal teams abandon the output because they cannot trust the underlying data lineage, reverting to manual Excel models. Principals refuse to pay headcount-replacement rates, treating the service as a standard software tool capped at lower SaaS pricing. Escalation to internal human review exceeds 40% of generated reports, destroying the unit economics.