CogniSpan®
Product note · Rev 4.2

A price is a hypothesis. We give it evidence.

CogniSpan runs three systems in sequence — observe the market, model your own demand, then move the price inside limits you control. Here is exactly what each one does.

MOD-01

Competitor Scraper

Hourly price truth from every marketplace you compete on.

Variant-level matching resolves size, colour and capacity so a 32x32 selvedge denim is never compared against a 30x30. Every observation is timestamped and kept for 36 months.

  • Refresh1h (Growth)
  • Match rate97.2%
  • SourcesShopify, Amazon, Walmart
  • MAP alertsReal-time
MOD-02

Elasticity Engine

A demand curve per SKU, not a category rule of thumb.

A hierarchical Bayesian model pools thin-data SKUs with their category, separates promo lift from base demand, and reports elasticity ε with a 90% credible interval instead of a single number.

  • ModelHierarchical log-log
  • Min history6 weeks
  • Outputε + 90% CI
  • RefitNightly
MOD-03

Smart Repricer

Price moves that defend margin, inside limits you set.

The optimizer searches for the contribution-margin peak, not the volume peak — then clamps the result to your floor, ceiling, MAP and maximum daily move before anything reaches your storefront.

  • ObjectiveContribution margin
  • Max move±7% / 24h
  • FloorsCost, MAP, custom
  • Write-backShopify, CSV, API
Appendix B

Model assumptions

Written down, because a pricing model you cannot interrogate is a pricing model you should not trust.

Functional form
Log-log demand, ln(units) = α + ε·ln(price) + seasonality + promo
Pooling
Partial pooling of ε toward the category mean, weighted by observation count
Promo handling
Discount periods enter as a separate term so promo lift never inflates ε
Inventory
SKUs below 14 days of cover are excluded from upward-move suggestions
Competitors
Competitor average enters as a covariate, not as the target price
Uncertainty
Suggestions below 0.35 confidence are surfaced as 'hold', not as a move
Scope

What CogniSpan will not do

  • ✕

    We do not match the lowest price on the market. That is a spreadsheet, not a model.

  • ✕

    We do not surface a single point estimate without its credible interval.

  • ✕

    We do not touch prices below MAP or below landed cost, under any configuration.

  • ✕

    We do not need shopper-level data, and we will not ingest it if you offer it.

Onboarding sequence

  1. 01~8 min

    Connect

    Point CogniSpan at your storefront. We import SKUs, costs and 90 days of order lines. No shopper PII ever leaves your store.

  2. 02~2 min

    Observe

    Add up to ten competitor domains. The scraper starts building the price panel immediately and flags MAP breaches from hour one.

  3. 03Overnight

    Model

    Elasticity is fit per SKU against your own demand history, pooled by category where data is thin. Confidence is reported, never hidden.

  4. 04Daily, 10 min

    Decide

    Review the suggested price, the expected margin delta and the credible interval. Apply one SKU, a filter, or the whole catalog.

Join the waitlistSee pricing