stocura watches two years of demand, finds the seasonal pattern in each SKU, and projects the next three months. No spreadsheets, no manual smoothing.
We import up to 24 months of order data from Shopify and Cin7. SKUs with less history fall back to category-level patterns.
Each SKU is fit independently — your candle business and your garden tools don’t share a season. We surface the pattern visually so you can sanity-check it.
Three-month forecasts feed the Reorder Queue. Month-to-date pacing tells you when reality is diverging from the forecast — early.
Not flat averages. Each SKU has its own season, surfaced as a 24-month chart you can read in three seconds.
Pacing percentage tells you how the month is tracking against forecast — so you can react before the month ends.
Toggle “show last year’s forecast vs actuals” to see how the model performed before trusting it forward.
Tag past promotions and we’ll exclude them from the baseline, so a Black Friday spike doesn’t become next year’s expectation.
Six weeks of data is enough to start. Newer SKUs borrow patterns from their product family while their own signal builds.
Your judgement matters. Bump a forecast up or down per-SKU when you know something the model doesn’t.
A seasonality template at staging. Empirical-Bayes blend with your own data. Nightly self-correction.
Pick Christmas, Father's Day, Mother's Day, or Evergreen — your forecast starts with a prior, not a blank slate. As your own data accumulates, the template fades to zero by month 12. Every night, Stocura re-checks each SKU's actual pattern and auto-applies a corrected template if yours has drifted. Every adjustment is audit-logged.