Research methodology
Krafie analyzes publicly available web and trend information to help you explore product ideas. Sales, revenue, and demand estimates are research guides, not verified Etsy results or a promise of future performance.
Read the evidence in context
Krafie uses publicly available information from sources such as Google search, Google Trends, Pinterest Trends, and other web sources, together with calculations and AI-assisted analysis. Not every result uses every source. A trend on Google or Pinterest reflects activity on that source; it is not a measurement of searches or purchases on Etsy. Naming a source does not imply a partnership or API connection with that provider.
Sources differ in coverage, geography, and time period. Missing or older inputs, seasonality, and changes in a product's price can affect the result. If a source or date is unavailable, use the result as a starting point for further research.
Monthly sales use favorites activity, cart counts, listing badges, and recent public purchase signals when available. Revenue multiplies that sales estimate by the current listing price, which may be the lowest variant price. It is a gross projection before refunds, fees, and costs, not the seller's actual receipts or profit. Keyword scores are not official Etsy search volumes or conversion rates.
Use the research alongside product quality, intellectual-property checks, pricing, customer fit, seasonality, production cost, and your own marketplace experience.
AI can help interpret signals or suggest product ideas and listing text, but it can also generate unsupported details. Check any claim against its underlying evidence. Etsy account access does not verify an AI-generated answer.
Fresh 24-hour purchase counts carry the most weight; Popular now is the strongest badge signal. We compare favorites, cart activity, and badge-based baselines, using bounded adjustments when signals occur together. A fresh purchase count starts at 70% weight and can reach 85% when supporting activity reinforces a higher sales pace. A low count can lower the estimate. Ads do not add sales, and dated signals lose weight and expire after 30 days.
Carts are current interest, not daily purchases. Counts such as “20+” remain minimum observations, and missing data is not treated as zero purchases. The model uses heuristic weights that have not been calibrated against sellers' private order histories. A single 24-hour snapshot does not establish a month of sales, and the monthly projection is not a measured total or guaranteed minimum.
When at least three recent collection checks span 48 hours, a short history can refine the current estimate by at most 10%. We compare independently dated activity and favorite growth over actual elapsed time. Repeated purchase messages are not added together, missing counts are not zeroes, and “20+” counts do not establish a decline. Current purchase signals retain priority; older history cannot refresh them. Without enough history, the current-signal calculation stays unchanged.
Shops rank by how many distinct listings have fresh success signals in our collected data: a positive purchase message checked within 24 hours, or a Popular now or Bestseller badge checked within seven days. Each listing counts once, even if it has several signals. Older listings still qualify when their signals are fresh. Missing signals do not prove a listing is unsuccessful.
Ties favor more listings with recent purchases, then Popular now, then Bestseller, then supporting cart activity. Ads, favorites, prices, and total listing volume do not decide the ranking. Sales and revenue averages cover only the qualifying listings and remain estimates. This compares the listings we have checked, not every listing or shop on Etsy. Partial collection results are identified when the request limit is reached.
Every keyword uses the same formula. Buyer demand summarizes recent purchase, Popular now, cart, badge, and favorite evidence from matching listings, with each shop weighted equally. Purchase signals carry the most weight; Popular now is the strongest badge signal. Confidence describes the supporting evidence separately. Missing evidence stays unavailable. This score does not measure keyword search volume.
Ads activity summarizes how often ads were seen among recently checked matching listings, with each shop weighted equally. It stays unavailable when there is too little fresh evidence. A check without an ad sighting does not prove that a campaign is inactive. The score does not reveal targeted search terms, spend, ad performance, or sales. Advertising contributes to competitive pressure, not a sales bonus.
Saturation compares listing and shop prevalence with other qualified phrases in the same product category. Broad wording, concentrated sellers, established competitors, and observed ads add competitive pressure. Small comparison pools receive less relative weight. The score has a conservative baseline because a small sample cannot establish an almost-empty market. It is not Etsy-wide keyword difficulty or a count of competing search results.
Opportunity balances demand with room to compete using their harmonic mean, discounts broad wording by up to 30%, and uses a bounded adjustment for relative position among supported phrases. Comparison pools are calculated before search, category, long-tail, and result-limit filters. Unknown and zero demand cannot become positive through ranking, and favorites alone cannot produce a strong score. A keyword needs its own listing support; it cannot borrow a parent category’s demand. These are heuristic comparison scores, not probabilities of success or measured search volumes.
Titles and verified seller tags provide direct phrase evidence. Focused long-tail phrases can qualify with three listings across two shops; broader phrases need eight listings across two shops. Production methods, file formats, and garment brands alone do not make a phrase a niche. Specific subjects, recipients, and uses add focus; evergreen wording is not penalized for being useful year-round. Product intent and phrase-completeness checks reject obvious fragments. All keywords includes these long-tail phrases. Growth and trend require at least three shared observation dates for the same listings across multiple shops, with sufficient coverage. They compare purchase pace or favorite growth; missing history stays unavailable. Measured trend can adjust base opportunity by up to 5%. Searches, clicks, and click-through rates are never inferred from favorites or carts.
When enabled and available, Pinterest Trends supplies a separate market outlook for matching phrases. The same matching rules apply across the catalog; a broad category or a different seasonal phrase cannot lend its trend to a niche keyword. Current interest may make a bounded adjustment to an already supported opportunity, within the existing evidence limits. Forecast direction describes movement within Pinterest’s predicted series and stays separate from current interest and Etsy buying signals. Missing matches, expired observations, and unavailable forecasts stay unavailable. Pinterest’s normalized interest indexes are not Etsy search volumes or proof of future sales.
The commercial application being prepared covers analysis of a seller's own shop and seller-reviewed listing publishing. Etsy-sourced analytics and AI processing require the applicable authorization from Etsy before they are enabled. External research estimates will remain identified separately from connected-shop records; they are not a claim to access another shop's private sales or buyers.
Data being publicly visible does not by itself grant permission to collect or reuse it. Source access, permitted uses, and attribution need to be assessed for each input. Research references do not grant rights to copy another seller's product, photos, or design.
Etsy integration status
