Dynamic pricing in retail: rules dressed up, or actual decisions?

A merch director sits through a dynamic pricing retail platform demo. The screen shows competitor scraping, an elegant rule builder, a price-change feed updated every fifteen minutes. The vendor calls it "AI-driven."

The phrase "continuous optimization" appears three times in twenty slides. At the end of the demo, the director asks the only question that matters: what is the system actually optimizing for, right now, on this SKU?. The honest answer, almost every time, is: whatever the rule you wrote down says.

That answer is the entire problem. Over the last five years, "dynamic pricing" has become marketing speak for "rule-based pricing with a slicker UI." The platforms branded as dynamic let you write more rules, faster, with prettier dashboards. That's pricing automation — which is genuinely useful — but it isn't pricing decisions. And the two have almost nothing in common architecturally, even though they're sold under the same label.

This matters because retail margin has nowhere left to hide. Average net margins sit around 3.2% on the Deloitte Top 250, and pricing cycles are compressing from weekly to hourly in some categories. In that context, the gap between a rule that fires and a decision that gets weighed is the difference between protecting margin and burning it. Most leaders evaluating "dynamic pricing" tools today are buying the second category while paying for the first.

What rule-based pricing actually is

Strip the vocabulary back. A rule-based pricing platform takes inputs you choose, runs them through conditions you write, and outputs a price change. The conditions can be sophisticated — multi-tier, segmented by store cluster, layered with calendar exceptions. The execution can be fast — sub-hour propagation to web, app and POS. But the logic is exactly what a merchandiser or pricing analyst would have written in an Excel macro, just running on better infrastructure.

The canonical rule looks like this. If competitor X drops below threshold Y on this SKU, match within Z% — unless margin would fall below floor F, in which case hold. That sentence is the whole pricing strategy for that SKU until someone changes it.

The platform's contribution is speed and scale: it can apply that rule to 50,000 SKUs across 200 stores in minutes, where a human team would need a week. That's real value. It just isn't dynamic pricing.

Rule-based pricing has three honest virtues. It's predictable — the pricing team can replay any decision and explain it in one sentence. It's auditable — the rule is the audit trail.

And it's controllable — turning it off, narrowing its scope, or overriding a specific case takes seconds. These are not trivial. For categories where pricing strategy genuinely is a small set of stable rules (regulated products, contractual pricing, long-tail commodities), a rule engine is the right tool. That control matters most for markdown automation in regulated sale periods, where the law constrains when and how prices can move.

The trap is using it where it isn't. Most of fashion, most of consumer electronics, most of fast-moving consumer goods cannot be honestly compressed into rules. In those categories price elasticity moves with weather, social signal, competitor inventory position, and what the customer paid last Tuesday.

When a rule engine is pointed at those categories, the result isn't pricing optimization. It's pricing reflexes. The system reacts faster than a human team would, but it reacts to the same narrow set of signals, in the same way, every time.

What decision-based pricing requires that rules can't deliver

True dynamic pricing isn't a faster rule loop. It's a different object architecturally. At each pricing tick, the system must weigh — simultaneously, not sequentially — at least seven factors that no single if-then chain can capture cleanly:

These factors trade against each other. Matching the competitor protects share but breaches the brand price-perception bound. Holding price protects margin but extends stock cover into the danger zone. Pushing the markdown earlier in some stores frees shelf space but signals weakness across the network if customers cross-shop.

No rule chain expresses this honestly, because the weights between factors aren't constant. They shift with season phase, with category, with how much of the assortment is already in the same state.

A decision-based platform models this as an optimization problem at every tick. Given these constraints and these weights, what price on this SKU at this store gets us closest to the multi-objective target? The answer is recomputed continuously — not because the system is showing off, but because the weights themselves move with the data.

That recomputation is what "dynamic" actually means. Not the speed of execution. The continuous re-evaluation of the trade-off.

You can tell the difference with one question to the vendor. *Pick a SKU. A competitor drops 5%.

Our stock cover on that SKU is at the 90th percentile of the assortment. The SKU is a brand-protected tier-1 product, two weeks into a new launch. What does the system do, and why?*

A rule-based platform either has a rule for that exact configuration (it doesn't), or it falls back on the closest rule (which is wrong). A decision-based platform recomputes the trade-off and shows you the weights it used. The difference is visible in thirty seconds.

Why the rebranding happened

"Dynamic pricing" became the label because rules sold poorly. Telling a merch director "we'll help you write rules faster" doesn't land in a boardroom that's already heard the BI promise and watched it underdeliver. Telling them "we use AI to dynamically optimize prices" lands.

The vendors who actually built decision platforms are the same vendors who watched competitors win deals with the same language wrapped around a rule engine. So the language migrated. Today, almost every pricing tool calls itself dynamic, which means the word has stopped carrying information.

This isn't pure cynicism. Some rule-based platforms genuinely added forecasting models that suggest rule parameters — "based on last season, set your match threshold at 4% not 5%." That's a meaningful improvement over a blank rule editor, and vendors are entitled to talk about it.

But suggesting rule parameters is still rule-based pricing. The rule, once set, is what executes. The forecast informs the human who writes the rule; it doesn't enter the decision loop.

The most important buyer question is what kind of logic is actually running at execution time? In today's market, that is the hardest question to get answered cleanly. Vendors use the same vocabulary regardless of which side of the line they sit on. And the proof points (case studies, ROI claims, "AI-powered" badges) don't distinguish the two architectures at all.

The test for a CFO and a merch director

Before any pricing-platform evaluation, two questions cut through the demo theater.

First question, for the merch director: "Can I write out the pricing logic this platform applies to my top-100 SKUs as an if-then chain a junior analyst could follow?" If yes, the platform is automating rules you already have — useful, but you're paying a dynamic pricing retail platform premium for a rule engine. If no — if the logic depends on a continuously updated optimization that no human could replay by hand — you're looking at decisions, and the evaluation criteria shift.

Second question, for the CFO: "What changes about our pricing strategy if we adopt this platform?" A rule-based platform changes execution speed and consistency. Strategy stays where it was — in the heads of the pricing team and the spreadsheets they maintain.

A decision-based platform changes the boundary of what the team controls. The team owns the guardrails — margin floors, brand tiers, vendor protections. The system owns the trade-off inside those guardrails. The team's role moves from setting rules to setting policy. That's a different operating model, not a tooling upgrade.

Both can be the right answer. The mistake is choosing one while believing you bought the other.

The deeper structural point

The rules-vs-decisions split isn't unique to pricing. We've made the same argument about BI tools: visibility was sold as steering, but visibility doesn't steer. We've made it about forecasting: a prediction isn't a decision. We've made it about workflow automation: moving paperwork faster isn't moving decisions faster. And we make it about the whole revenue picture in what is yield management in retail: price alone is one lever of four.

Pricing is the next category where the same line gets drawn. And like the others, the line isn't between "good" and "bad" tools. It's between tools that automate the logic you already have and tools that change what the logic is. Both have a place. They don't have the same place.

The Solya position

This article isn't a pitch for Solya pricing — it's a buyer's-side mental model, and the model works whether you eventually buy from us or not. Solya is built on the decision-based side of the line, by design: the pricing surface plugs into the same decision engine that handles markdown, replenishment. And inter-store reallocation, so trade-offs aren't isolated per system. That architecture has its own evaluation criteria, its own integration cost, its own change-management profile — none of which look like buying a rule engine.

If you're currently evaluating a dynamic pricing retail platform, the question we'd suggest you start with is the merch director's question above. Not because the answer determines whether Solya is right for you, but because the answer determines whether the platforms on your shortlist are even comparable to each other. Today, in most evaluations we see, they aren't. And the buyer pays a decision-platform price for a rule engine, or worse, dismisses a decision platform because it doesn't behave like the rule engine they were comparing it to.

The question to put on the next vendor call

Before the next demo, write down — in plain English — the pricing logic you'd want the platform to execute on your three most strategically sensitive SKUs. If you can write it down, you don't need dynamic pricing. You need rule automation, and you should buy the cheapest reliable rule engine on the market.

If you can't write it down — if the logic depends on weights that shift with the data — ask every shortlisted vendor the merch director's question. Watch what happens. The answers will sort the market for you, faster than any RFP ever will.

For a definitional anchor that frames where dynamic pricing came from and the two architectures most pages collapse into one, start with what is dynamic pricing in retail?.

---

Want a buyer's-side read on your current pricing stack?

At Solya, we offer retail leadership teams a 30-minute diagnostic focused on the rules-vs-decisions line. We look at what your current pricing platform actually does, where the boundary between automation and decision sits in your stack, and what changes if you cross it. No vendor takedown, no generic pitch — a structured conversation grounded in your assortment and your constraints.

You'll walk away with:

• A clear map of where your current pricing logic is rule-based vs decision-based
• The categories in your assortment where the rule-vs-decision distinction is most expensive
• The evaluation criteria that actually separate the two architectures, beyond the marketing labels