How does dynamic pricing software decide on a new price?
Dynamic pricing software takes three inputs and produces one output. The inputs are the competitive signal (what comparable products cost right now), the internal constraint (cost price, margin floor, stock level, promotion calendar) and the objective (revenue, margin or volume). The output is a price, plus a record of why.
The mechanism differs by vendor. Omnia Retail, PriceEdge and Dealavo resolve a rule tree written by a human. 7Learnings resolves a demand forecast produced by a model. Pricefx resolves a governance layer that may or may not require an approver. Which mechanism fits depends less on sophistication and more on who has to defend the price afterwards.
What guardrails should be in place before switching on automated repricing?
Four guardrails prevent almost every expensive repricing incident. A floor price per product stops the engine from selling below cost when a competitor lists an error price. A maximum movement per day stops a feedback loop between two automated repricers from spiralling. A category-level rollout means the first mistake happens across fifty products rather than fifty thousand. And a manual override that logs who overrode what keeps the exceptions visible instead of silent.
The tools differ in how first-class these controls are. Omnia Retail and Pricefx treat guardrails as part of the strategy definition. Cheaper tools tend to implement floors but not movement limits, which is worth checking before automation goes live.
How fast does a price change actually need to publish?
Publication cadence should match how fast the category moves, not how fast the software can go. Consumer electronics during a promotional period can justify several checks and publications per day; Price2Spy supports up to eight checks daily and Dealavo repositions inside marketplace cycles. Furniture, industrial parts and most B2B catalogues move on a weekly rhythm, and repricing hourly there just adds noise and confuses customers who saw a different price yesterday.
The constraint is often the channel rather than the tool. Marketplace APIs rate-limit updates, and a storefront cache can hold an old price long after the engine changed it. Verify the full path from decision to visible price before assuming a cadence.
Who should not implement dynamic pricing?
Three situations argue against it. A catalogue under roughly a thousand SKUs rarely produces enough movement to repay the configuration effort. A market where customers see the price change repeatedly — subscriptions, contracted B2B customers, anything with a published price list — turns dynamic pricing into a trust problem. And a business without reliable cost prices per SKU cannot set a safe floor, which makes automation genuinely dangerous rather than merely unhelpful.
In the first two cases, competitor monitoring without automation is usually the better purchase. Minderest, Dealavo and Price2Spy all serve that need without touching prices.
Where do dynamic pricing rollouts break?
Product matching breaks first. If the engine believes your SKU and a competitor SKU are the same product and they are not, it will confidently set a wrong price. Matching on EAN or GTIN is reliable; matching on product title is not, and every rollout needs a validation pass before automation is enabled.
Ownership breaks second. Pricing sits between commerce, finance and sales, and a rule set that nobody owns drifts into a pile of manual overrides within two quarters. The implementations that survive have one named owner and a scheduled review of the rules.
Data freshness breaks third. A daily crawl against an hourly competitor produces prices that are always one step behind, which looks like a failing algorithm but is actually a failing input.
