MODEL V4 · METHODOLOGY
How MarketPulse turns inputs into a decision.
No hidden “AI confidence” number. V4 separates user inputs, live web signals, category benchmarks and modeled assumptions before running 2,500 reproducible, correlated scenarios.
Start with unit economics
Landed cost = unit cost + shipping & duties
Base contribution = price − landed cost − platform fee − return reserve − CAC
Break-even CAC = price − landed cost − platform fee − return reserve
The base case is shown directly; it is not the final score.
Read live web demand signals
After product, market and category are entered, the system checks current search suggestions for the target region. Suggestion breadth, commercial-intent density and related-query variety inform early demand and competition estimates.
Search suggestions show what people are currently looking for. They are not exact monthly volume or sales, and MarketPulse does not present them as marketplace backend data. When live lookup fails, a conservative category prior is used and confidence is reduced.
Handle unknown supply inputs explicitly
When MOQ or lead time is unknown, the model uses category ranges and marks those values as inferred. The action plan then prioritizes obtaining two supplier quotes and rerunning the report with the more conservative quote.
These defaults help a first-time user complete an early screen; they do not replace a written supplier quote.
Stress-test 2,500 correlated scenarios
V4 no longer moves every input independently. A weak-market shock can lower realized price while raising CAC and returns; a supply or currency shock can raise product and freight costs together. Competition, category volatility and lead time change the size of those shocks.
The report surfaces P10 downside, P50 typical and P90 upside contribution, plus the share of profitable simulations. Identical inputs use the same deterministic seed, so results remain reproducible.
Use context-sensitive weights
Weights change with the decision. Regulated products put more weight on risk; unknown demand increases demand weight; high competition increases differentiation weight; long lead time or inferred supply data increases operations weight. The exact weights are shown in every report.
Expose sensitivities, limits and an action plan
V4 separately shocks price, CAC, product cost, freight and returns, ranks the largest contribution risks, and calculates minimum viable price, maximum landed cost and maximum tolerable return rate. Recommendations then use the product name, target market, weakest dimensions and live related queries to set numeric pass-or-stop rules.
What the model does not know yet
Until marketplace APIs are connected, MarketPulse does not claim exact sales, monthly search volume, ad conversion or competitor inventory. Live search signals improve early screening but do not replace real orders.
This is a decision-support model, not a sales guarantee. Taxes, warehousing, currency movements and jurisdiction-specific compliance may require separate review.