Dynamic pricing software for retail catalogs, not hotel rooms

Dynamic pricing software automatically adjusts prices in response to conditions rather than leaving them fixed until somebody remembers to look. This page covers dynamic pricing for ecommerce retail catalogs: physical products, competitor comparison, stock and margin constraints. If you price hotel rooms, short-term rentals, flights or event seats, you want revenue management software instead. Same idea, different math, different category of vendor.

First, the category confusion worth clearing up

Search for dynamic pricing software and most of what ranks is built for hospitality. That is not vendors being unhelpful. It is two genuinely different problems sharing a name, and buyers waste calls before they find out. Here is the split.

The rest of this page is the retail version. If you landed here for lodging, the useful thing we can tell you is what to search for: revenue management system, channel manager integration, and pricing by length of stay. Nothing below will help you, and we would rather say so than keep you reading.

What actually changes a price in an ecommerce catalog

Dynamic pricing sounds like a continuous algorithm. In practice it is a short list of triggers, each with a correct response and a guardrail that stops the response from going wrong. Write the guardrail column first. Teams that write the response column first are the ones who end up explaining a $12.00 sale price to a finance director.

Rule-based and model-based dynamic pricing

Rule-based: deterministic and explainable

A rule is a scope, a strategy and a set of floors. Match the lowest in-stock competitor within this category, never below a 32 percent margin, never below MAP, never move more than 5 percent in a day. It produces the same price from the same inputs every time, which means you can reconstruct why any price is what it is. It needs no sales history, so it works on the day you install it, including on products you have never repriced.

What it cannot do is find the best price. A rule tracks the market. If the whole category is mispriced, a rule keeps you accurately mispriced. That limitation is worth accepting for a year while the measurement discipline gets built.

Model-based: estimates response, costs explainability

A model estimates how units respond to price and proposes the price that maximizes a goal. It can find a better price than a rule, and it can find it on products where your instinct is wrong. It also needs your own price and volume history, it degrades quietly when the market shifts underneath it, and it is much harder to explain to a category manager who thinks the number is wrong. See price optimization software for what the data requirement really looks like.

The arrangement that works in mid-market is layered, not either-or. Rules hold the floors and track the market every refresh. Measurement, whether an elasticity estimate or a designed test, sets the target the rules aim at. A human owns the exceptions. Nothing in that arrangement needs to be autonomous to be automatic.

The trust limits: what dynamic pricing should not do

Every constraint below costs some short-term margin. Each one is still worth keeping, because the downside is not a bad quarter, it is a screenshot that outlives the quarter.

• Do not surge-price consumables and essentials. Water, filters, generators, medical supplies, baby formula. Raising the price of a necessity because demand spiked is the single fastest way to become a story, and in a declared emergency it can also be unlawful.
• Do not personalize prices per shopper. Two people, same product, same minute, different price because of device, location or browsing history. It gets discovered, it is difficult to defend, and it turns a pricing tactic into a consumer-protection question. Segment by market, channel or volume instead, and publish the logic.
• Do not change price between add-to-cart and checkout. A price that moves inside a session reads as a bait and switch even when the rule was innocent. Freeze the price for the life of the cart.
• Do not let a price oscillate visibly. Four changes a day on the same SKU teaches regular shoppers to wait for the dip. Cap changes per week as well as per day.
• Keep advertised prices honest. If a feed lags the storefront, you have an incorrect advertised price and a customer service decision. Push changes to the store, the marketplace listing and the shopping feed on the same cycle.
• Never automate below MAP. If you sell brands with a policy, one automated violation can cost the account. Set the MAP floor as a hard constraint, not a warning.

How to roll out dynamic pricing with a control group

The reason most dynamic pricing programs cannot prove their value is that they went live everywhere at once, so there is nothing to compare against. A control group costs you a few weeks and buys you an answer.

1. 1

   1. Pick a scope you can read

   One or two categories, roughly 100 to 300 SKUs, with enough weekly unit volume that a single week produces a number rather than noise. Skip the long tail for now.

2. 2

   2. Split into treatment and control

   Randomize within strata of velocity, margin band and price band, so the two halves are genuinely comparable. Do not put the interesting products in treatment and the boring ones in control, which is the split everybody reaches for and the one that guarantees an unreadable result.

3. 3

   3. Write the guardrails before the strategy

   Margin floor, cost floor, MAP floor, maximum change per day, maximum changes per week, and an exclusion list. Agree them with finance in writing. They are far harder to add after the first surprising price.

4. 4

   4. Run in review-queue mode for two weeks

   Rules generate recommendations, nothing publishes. Read what the rule wanted to do every day. This is where you discover the marketplace seller you should have excluded and the cost record nobody updated since March.

5. 5

   5. Turn it on for treatment only

   Leave the control group alone even in the week it is obviously mispriced. That discomfort is the price of knowing whether any of this worked.

6. 6

   6. Run one full purchase cycle

   Four to eight weeks for most catalogs, longer for considered purchases. Ending a test early because week two looked good is how a random fluctuation becomes company policy.

7. 7

   7. Read gross margin dollars first

   Margin dollars, then units, then revenue, in that order. A rule that grows revenue while shrinking margin is a rule that discounted, which you could have done without software.

8. 8

   8. Check the distribution before expanding

   Look at the per-SKU contribution, not just the total. If most of the gain came from three products, you learned about three products rather than about the rule.

A worked example, with invented numbers for illustration. Treatment is 120 SKUs, control is 120 comparable SKUs, and the test runs six weeks. Treatment produces $61,400 of gross margin, control produces $58,900. The difference is $2,500, or 4.2 percent. That is a result worth expanding, but only after step eight: if $2,100 of the $2,500 traces to three SKUs where a competitor happened to go out of stock, the rule has not been demonstrated yet. Run it another cycle. To sketch the size of the prize before you start, the dynamic pricing ROI calculator does the same arithmetic on your own numbers.

Two failure modes worth naming

The repricing spiral

You write a rule that beats the lowest competitor by one cent. A competitor runs the same rule. Every refresh cycle knocks a cent off both prices, and because the cycles are hours apart the descent takes days rather than minutes, which is exactly why nobody notices. Both of you end up at a floor with the same share you started with, and the category is permanently cheaper. The fixes are unglamorous: match rather than undercut, require a minimum gap before reacting, cap the daily change, and set a margin floor you would actually defend rather than one chosen to avoid ever binding.

Chasing a price nobody can buy

A marketplace listing shows $89.00 on a product you sell at $109.00. The rule cuts you to $89.00. The listing is out of stock, or it is a refurbished unit, or it is a single item from a seller with no inventory behind it. You have cut 18 percent of your price to lose to an offer that never shipped. Three defenses stop it: capture stock status on every competitor listing and exclude out-of-stock offers from the market calculation, require a match confidence threshold so a refurb or a variant never anchors your price, and require a larger move to be supported by more than one source. Coverage and freshness per source live on the data quality dashboard, because a rule is only as good as the price it read.

What our dynamic pricing software does

Rules are built in a visual builder with hard guardrails: margin floor, cost floor, MAP floor and maximum daily change. Every rule has a scope and a strategy, and floors always outrank strategy. Recommended changes go to a review queue for approval, export as CSV, or push directly to the connected store. Every change is stored with its before and after price, the rule that produced it and the competitor listing that triggered it, so any price on your site can be explained a month later.

Underneath, competitor price monitoring supplies the prices, stock status and promotions the rules read, refreshed daily on Starter, twice daily on Growth, four times daily on Scale and hourly on Enterprise. Matching runs automatically on GTIN, UPC or EAN first, then title and attribute embeddings, then image comparison, with a confidence score on every match. Changes push through Shopify, WooCommerce, BigCommerce, Adobe Commerce, Google Merchant Center and Amazon Seller Central, or via the REST API and webhooks. Alerts go to email and Slack.

Two things it does not do, stated plainly. There is no per-shopper personalized pricing, for the reasons above. And there is no autonomous mode that reprices a whole catalog with no floors and no review, because every team that asks for it eventually asks for the change log that proves what happened. Rule design is covered in repricing software and the dynamic pricing guide.