Moving a monolithic AI-insights platform to a tiered, localizable, repeatably-priced product — built to scale across global markets without exploding operational complexity.
📊 Full deliverables index · 📈 ARR Model · 🧠 Decision Science
Prepared by: Product Strategy Taskforce · Subject: Ai Palette — AI-powered consumer-insights & product-innovation platform (Singapore, ~$13.5M raised; customers incl. Nestlé, Danone, Kellogg's, Cargill, Olam).
Status / disclaimer: Conceptual strategy from secondary research + clearly-labeled assumptions. Ai Palette is private — no audited financials or SEC filings exist, so every figure here is an estimate anchored to public benchmarks (audit trail →).
MarketPulse is a multi-market expansion and monetization strategy case study for Ai Palette, a Singapore-based AI consumer-insights and product-innovation platform. The repo answers four questions with quantified, source-traceable frameworks: what to standardize vs. localize, how to package features into pricing tiers, how to price by market, and which go-to-market motion to use — backed by a Markov Decision Process (MDP) for market sequencing, a SHAP-style explainability layer for scoring, and a bottoms-up ARR model (≈$9.2M Year 3 / $28M Year 5). It includes an executive presentation, a commercial-architecture workbook, a consulting strategy memo, an interactive dashboard, and a clickable product wireframe — all built from public benchmarks, not Ai Palette's internal data. Try it live: the entire case is deployed as an interactive web app at market-pulse-nine-rho.vercel.app — no install required.
▶ Live demo: market-pulse-nine-rho.vercel.app
The whole strategy is also a deployed, interactive product — a Next.js web app (TypeScript, Tailwind CSS, Recharts) hosted on Vercel. Nothing to install:
| Section | What you can do |
|---|---|
| Dashboard | Explore market-attractiveness scores, the TAM → SAM → SOM funnel, the competitive-positioning 2×2, the localize-vs-standardize split, tiered pricing, regional willingness-to-pay, and the 3-wave roadmap. |
| ARR Model | Drag the assumptions on a live bottoms-up ARR calculator and watch every chart (build, new-vs-expansion, tornado sensitivity) recompute instantly. Base case ≈ $9.2M Year 3 / $28M Year 5. |
| Decision Science | Run an interactive Markov Decision Process (MDP) market-entry sequencer and re-weight a SHAP-style score explainer in the browser. |
| Strategy | Read the full written case — 11 chapters with a chapter navigator. |
| Deliverables | View or download the executive deck, commercial-architecture workbook, ARR model, strategy memo, and one-page cheat sheet. |
Run it locally with cd web && npm install && npm run dev.
What problem does MarketPulse solve? It prevents the two failure modes of market expansion: over-localizing a product (engineering bloat, slow rollout) and under-localizing it (lost deals, poor product-market fit). It does this by explicitly scoring what must be localized vs. kept global, market by market.
What is Ai Palette? Ai Palette is an AI-powered consumer-insights and product-innovation platform headquartered in Singapore, with ~$13.5M raised and customers including Nestlé, Danone, Kellogg's, Cargill, and Olam. It is the real-world subject this case study is built around.
Is this Ai Palette's actual internal strategy? No. Ai Palette is privately held with no public financials, so every number in this repo is an estimate anchored to public benchmarks and clearly labeled as such — see the assumptions and data sources audit trail.
What pricing model does MarketPulse recommend? A hybrid model: a platform subscription as the anchor, usage/credit-based billing (e.g., AI concept-generation runs, markets, categories), and seats as an expansion lever — three tiers (Essentials, Growth, Enterprise) mapped to buyer maturity and deal size.
What is the recommended market entry sequence? Wave 1: deepen Southeast Asia and India. Wave 2: enter the United States (highest willingness-to-pay). Wave 3: EU and Middle East. The sequence is generated by a scored rubric and an MDP solver, not a qualitative guess.
What technical methods does this use?
A Markov Decision Process (MDP) for optimal market-entry sequencing under uncertainty, a SHAP-style explainability layer to justify each market's score, and a bottoms-up ARR model with sensitivity (tornado) analysis — see decision-science/ and arr-model/.
Who is this repo for? Product strategists, GTM/RevOps teams, and consultants studying multi-market SaaS expansion, tiered pricing design, or localization-vs-standardization tradeoffs — and anyone evaluating this as a portfolio case study.
Is there a live demo of MarketPulse? Yes. A fully interactive web app is deployed on Vercel at market-pulse-nine-rho.vercel.app. It includes the solution dashboard, a live ARR model you can drive with sliders, the MDP + SHAP decision-science tools, the full written strategy, and downloadable deliverables — no sign-up or install required.
What is MarketPulse built with?
The interactive web app is built with Next.js (App Router), TypeScript, Tailwind CSS, and Recharts, and is deployed on Vercel. The standalone ARR model is a Python/Streamlit app; the MDP solver and SHAP-style explainer are plain Python (NumPy + Matplotlib); and the underlying data lives in machine-readable CSVs under frameworks/.
Entering new markets with one undifferentiated product over-localizes everything (engineering bloat) or under-localizes (lost deals). MarketPulse defines exactly what to standardize vs. localize, how to package it into tiers, how to price it per market, and which sales motion delivers it — then wraps that in a repeatable decision engine.
| Path | What it is |
|---|---|
docs/01-context-and-problem.md |
Business context, the expansion blind spot, scope |
docs/02-assumptions-and-data-sources.md |
Every estimate is traceable here — sources + estimation method |
docs/03-phase1-localization-matrix.md |
Phase 1 — what stays global vs. what gets localized |
docs/04-phase1-feature-tiering.md |
Phase 1 — Essentials / Growth / Enterprise tier logic |
docs/05-phase2-pricing-packaging.md |
Phase 2 — pricing model, packaging, willingness-to-pay |
docs/06-phase2-gtm-onboarding.md |
Phase 2 — PLG vs. sales-led motion & onboarding |
docs/07-market-sequencing.md |
Scored market prioritization & 3-wave entry plan |
docs/08-competitive-benchmark.md |
Competitor benchmark, positioning 2×2, market structure, consolidation signal |
docs/09-strategy-memo.md |
Consulting strategy memo (SCQA + Pyramid Principle + frameworks) — markdown source |
docs/10-arr-model-methodology.md |
Bottoms-up ARR model methodology, defaults, base case & sensitivity |
docs/11-decision-science.md |
MDP entry-sequencing + SHAP-style score explainability (the rigorous core) |
workflow/market-entry-automation.md |
The automated market-entry decision engine (rubric + flow) |
docs/12-valueforge-prd.md |
ValueForge PRD (Deliverable B) — product requirements + user flow |
dashboard/index.html |
Interactive solution dashboard — open live ↗ (Chart.js visualizations) |
wireframe/index.html |
ValueForge interactive wireframe (Deliverable B) — open live ↗, clickable brand-manager flow |
arr-model/ |
Interactive ARR model — Streamlit app (streamlit run app.py) with tornado + live MDP |
decision-science/ |
MDP solver + SHAP-style explainer (Python; generates the decision plot & policy chart) |
DELIVERABLES.md |
Master index of every deliverable mapped to the brief |
frameworks/ |
Machine-readable CSVs: localization, tiering, pricing, market scoring, competitor benchmark |
deliverables/ |
The four graded outputs (see below) |
- Deliverable A — Executive Presentation — rich editable PowerPoint (
deliverables/DeliverableA_Executive_Presentation.pptx, 17 slides, embedded data-viz) + PDF: market sequencing, pricing & packaging, standardization↔localization trade-offs, with quantified charts (TAM funnel, competitor 2×2, localization split, market scores, pricing, regional WTP, ARR build, tornado, SHAP, MDP). - Deliverable B — Commercial Architecture (
deliverables/DeliverableB_Commercial_Architecture.xlsx): 8-tab workbook with live formulas mapping tiers × markets × localized features × monetization, plus the decision engine. - Consulting Strategy Memo (
deliverables/MarketPulse_Strategy_Memo.docx): McKinsey/BCG-style narrative — SCQA, Pyramid Principle, 3C, Porter, Ansoff, value chain, GE-McKinsey, risk register, 90-day plan. - Interactive Solution Dashboard (
dashboard/index.html): structured, filterable visualizations — market scores, TAM/SAM/SOM, positioning 2×2, localization split, pricing, GTM, and the 3-wave roadmap. - ARR Model — interactive Streamlit app (
arr-model/app.py, now with a tornado sensitivity chart + a live MDP cold-start slider) + matching Excel model (deliverables/DeliverableC_ARR_Model.xlsx, live formulas). Base case ≈ $9.2M Y3 / $28M Y5 ARR. - ValueForge Product Strategy Deck — editable PowerPoint (
deliverables/DeliverableA_ValueForge_Product_Strategy_Deck.pptx, native charts) + PDF: 9-slide consultant-style deck — problem, target user, differentiation logic & consumer-resonance mapping, with quantified visuals. - ValueForge PRD + Wireframe (
docs/12-valueforge-prd.md+wireframe/index.html): requirements and a clickable brand-manager flow ending in a decision-ready strategy brief. - Judge's Cheat Sheet (
deliverables/MarketPulse_Judges_CheatSheet.pdf): the entire story on one page.
See DELIVERABLES.md for the full index mapped to the brief.
- Localize the thin edge, standardize the engine. Keep the AI/ML core, data pipeline, and concept-generation global; localize only the data sources, language/NLP, taxonomy, compliance, and GTM workflow. ~80% of code stays shared.
- Three tiers — Essentials (self-serve trend discovery), Growth (insights + concepts), Enterprise (end-to-end + localization + workflow integration) — mapped to enterprise maturity.
- Hybrid pricing: platform subscription (anchor) + usage/credits (Concept Genie runs, markets, categories) + seats as an expansion lever. Pure per-seat is fading; hybrid posts the highest growth in 2025 benchmarks.
- Motion follows ACV: PLG-assisted for Essentials (ACV <$10K), sales-led for Enterprise (ACV >$25K, committee buying). Hybrid in between.
- Sequence: Wave 1 deepen SEA + India (home advantage), Wave 2 attack US (highest willingness-to-pay), Wave 3 EU + Middle East. Driven by the scoring rubric, not gut feel.
Live app: market-pulse-nine-rho.vercel.app · Source: github.com/nightraven4545/MarketPulse
git clone https://github.com/nightraven4545/MarketPulse.git
cd MarketPulse
# 1) the interactive web app (Next.js) — the live site
cd web && npm install && npm run dev # → http://localhost:3000
# 2) the ARR model (Python / Streamlit)
cd ../arr-model && pip install -r requirements.txt && streamlit run app.py
# 3) the decision-science engine (MDP + SHAP)
cd ../decision-science && pip install -r requirements.txt
python mdp_sequencing.py && python explain_scores.pyThe web app is deployed on Vercel with the project Root Directory set to web; every push to main redeploys automatically.
See docs/02 for the disclaimer to keep visible to evaluators: numbers are illustrative, defensible estimates — not company-reported figures.