Methodology
Market signals answer different questions. This page explains what the numbers represent, which limits matter, and how much weight to put on scores and forecasts when testing a product idea.
Understanding the evidence
MeshNomics draws on public 3D model repositories and marketplace signals, refreshed regularly. Each signal is shown with its capture date so you can judge how current it is.
Marketplace purchase indicators: published monthly purchase floors and their trends describe the captured products. A bought-in-past-month badge is a lower bound, not an exact sales count or total market size. Prices are offers, not receipts.
Third-party sales estimates: listing sales figures are provider estimates, not verified transactions. Check the stated period, capture date and sample coverage; an undated estimate cannot establish a current monthly rate. Keyword estimates describe search activity, not purchases; check their units, period and coverage.
Channel fees: dated channel-fee assumptions are estimates for planning. Check the displayed dates and applicable channel rules before using a rate; a captured schedule is not a guarantee that today’s fees are unchanged.
Model engagement: design metadata and download activity measure model interest, not purchases of printed products. Model engagement does not establish demand for a finished product.
Use the product research guide to compare these signals without treating them as interchangeable. The existing market and concept research screens expose the evidence behind a shortlist.
Sample data: where a feature ships ahead of its data source, it runs on a sample set so the workflow can be used end to end. Those screens are labelled as samples.
Design permissions: check the original creator, license and commercial-use conditions before making anything for sale. Unknown permission is not permission; marketplace rules can impose additional requirements.
Interpreting scores
Opportunity scores help prioritize product ideas for further research. A high score is not an expected sales count, a profit forecast or proof that a product will succeed.
Compare the underlying demand and competition evidence, including its date, coverage and units. Sparse or missing evidence limits a comparison; missing data does not mean zero demand.
Production costs and machine comparisons depend on the settings and assumptions you supply. Check material, labour, failure allowances and dated channel-fee assumptions against your own process. Geometry-based estimates are not slicer results or measured production runs.
Sales forecasting
Sales forecasts are estimates over a stated time horizon. Read the range alongside the assumptions, sample coverage and capture dates. A forecast is not a record of completed transactions or a promise of future sales, and a narrow range does not remove uncertainty in the underlying evidence.
A displayed 80% interval describes the model’s estimated uncertainty; it does not guarantee that 80% of future outcomes will fall inside it. Actual performance depends on evidence quality and changing market conditions. Test a small batch and compare your own results before scaling.
Confidence tiers
Confidence labels describe the strength of the available evidence, not the probability that your business will succeed. Low confidence calls for extra validation; even high confidence does not turn estimates or model engagement into verified purchases.
Our standing rule: when the data is insufficient we say so rather than emitting a confident number. Signals are aggregated market estimates, not guarantees of sales.