Dynamic Pricing: What Ecommerce Can Learn From Steam and Sony

What dynamic pricing actually means

Dynamic pricing is a broad term, and most of the confusion around it comes from collapsing three different things into one word. It is worth separating them before we look at what Steam and Sony are doing, because the risk profile is completely different at each level.

Rule-based dynamic pricing changes the price by formula: demand, inventory, time of day, competitor prices. Airlines and hotels have done this for decades, and nobody is outraged that a Tuesday flight is cheaper than a Friday one. The price is the same for everyone who meets the same conditions.

Personalized promotion keeps the price identical for everyone but changes which offer each shopper is shown. One person sees a bundle, another sees free shipping, a third sees nothing because they were going to buy anyway. The shelf price does not move; the merchandising does.

Personalized pricing is the controversial one: the same item, at the same moment, costs more for you than for the person next to you, based on what the store thinks you will pay. This is the version that makes headlines with the word “outrage” in them.

Keep those three straight, because Sony and Steam each picked a different one, and the choice is the whole story.

What Sony is doing: personalizing the price

Since late 2025, Sony has reportedly been running A/B pricing tests on the PlayStation Store. The experiment was spotted and tracked by the price tracker PSPrices rather than announced by Sony, so every number here should be read as a reported estimate, not a company figure. Sony has not publicly confirmed the program at all.

What the trackers describe is straightforward and, to a lot of players, unsettling: the same game, in the same region, on the same kind of device, showing a different price to different accounts. Shoppers appear to be sorted into a control group that sees the standard price and test groups that see a personalized discount, with the split reportedly driven by signals such as past purchasing behavior and store activity.

The reported scale grew quickly, from a few dozen games in around 30 regions to well over 100 titles across roughly 70 regions, spanning Europe, Latin America, Asia, the Middle East, and Africa. Reports disagree on whether the largest markets were included or deliberately held back for regulatory reasons, and on the exact discount spread, which has been described anywhere from the mid-single digits to the high twenties in percent. First-party flagships like God of War Ragnarök, Marvel’s Spider-Man 2, Helldivers 2, and The Last of Us Part II were reportedly in the mix alongside major third-party publishers.

This is personalized pricing in its purest, most contested form. The number on the tag is different for you than for someone else, and the thing deciding the difference is a model’s guess about your behavior. It is also exactly the move that generated a wave of angry coverage the moment players noticed it.

What Steam is doing: personalizing the discovery of discounts

Valve announced a change that sounds similar but is mechanically very different. Starting in early 2027, Steam is retiring its manual, calendar-based Discounts & Events section, the familiar Weekend Deals, Midweek Deals, and Daily Deals that everyone saw at once. Valve has already stopped scheduling new curated placements there.

What replaces it is a personalized, algorithmic feed of discounted games. Which deals get surfaced to you will be driven by your library, wishlist, and browsing behavior, so two players open the store and see a different set of games on sale.

Valve’s stated justification is a conversion story, and the numbers it shared are the part ecommerce operators should underline. In testing, the personalized approach let Steam surface roughly ten times more games per day to shoppers, with measurable lifts in store-page visits, wishlist additions, and cart additions. Developers, meanwhile, are being promised new tools to run their own discounts and events now that the central calendar is going away.

Put plainly: Steam found that the calendar was the bottleneck. A human scheduling the same deals for everyone could only show a tiny slice of the catalog. An algorithm matching discounts to interest showed an order of magnitude more, and people bought more as a result, without anyone paying a different price.

The two lessons, side by side

Put the two approaches next to each other and you get a clean decision framework for your own store.

Both companies are chasing the same underlying truth: flat, one-size-fits-all pricing and calendar-driven sitewide sales leave money and relevance on the table. Static pricing assumes every visitor is identical. They are not, in intent, in price sensitivity, or in what they are even looking for. The old model treated a loyal repeat customer, a first-time bargain hunter, and someone who was going to buy at full price in thirty seconds as the same person, and priced all three the same way.

But the two paths carry wildly different risk. Sony’s path, charging different people different amounts for the identical item, is the version that earns you a public-relations fire and, in some jurisdictions, a regulatory review. Steam’s path, showing the right offer to the right person at a price that is the same for everyone, captures most of the upside with almost none of the downside. For the overwhelming majority of ecommerce businesses, Steam’s model is the one to copy first.

Why this matters for ecommerce now, not in 2027

It would be easy to file this under “interesting gaming news.” That would be a mistake. Two of the most pricing-sophisticated storefronts on the planet, with every incentive to protect customer trust, decided independently and at the same time that static pricing was costing them. When the biggest players move together, the techniques they validate trickle down to everyone else’s customers’ expectations within a couple of years.

It also connects directly to a shift we have been tracking in how people shop: the rise of AI-assisted buying. When a shopper asks an assistant to find the best price on something, or when an AI shopping agent is doing the buying on their behalf, flat pricing and blanket sales become even blunter instruments. The store that can respond to context, inventory, demand, and the individual, cleanly and transparently, is the one that wins the transaction. We wrote about how AI is reshaping the measurement side of this in how AI search changes marketing attribution, and the pricing side is the next shoe to drop.

Five things your store should take from this

You do not need to be Valve or Sony to act on any of this. The same mechanics scale down to a store doing five figures a month.

1. Replace the calendar with a signal engine. A sitewide “20% off this weekend” is the retail equivalent of Steam’s old deals calendar. It discounts everyone, including the people who would have paid full price, and it trains your customers to wait for the next sale. Shift budget toward offers triggered by behavior: cart value, browse history, wishlist and favorites activity, time since last purchase, and churn risk. The discount goes to the shopper who needs a nudge, not the one already reaching for their card.

2. Personalize the offer surfaced, not the price. Start exactly where Steam is. Same price for everyone, but the right product and the right promotion shown to the right segment. This is lower-risk, legally boring, and usually where the fastest conversion gains live. It is also the part most stores already have the data to do and simply never built.

3. Treat pricing as an experiment, not a decree. Sony’s entire program is an A/B test with a control group. The lesson is not “charge more.” It is “stop guessing.” Every pricing and promotion change should have a holdout group and a measured lift, so you know whether the move made money or just moved numbers around. Most stores change prices on a hunch and never learn whether the hunch was right. This is the same discipline we bring to SEO and content: measure the lift, not the activity.

4. Mine the data you already have. Wishlists, abandoned carts, repeat-purchase cadence, and browse depth are the exact signals both companies are using. Most stores collect every one of these and act on none of them. Before you buy any new tool, map what your platform is already recording. The raw material for a personalized-promotion engine is almost always sitting unused in your existing analytics and order history.

5. If you touch the price itself, do it in daylight. Sony’s real mistake was not testing prices. It was the silence. Regional pricing, loyalty tiers, first-purchase offers, and dynamic markdowns on aging inventory are all defensible when they are transparent and rule-based. Opaque “you specifically pay more” pricing is the thing that burns trust and draws regulators. The direction of travel is clearly toward dynamic pricing; the winners will be the ones who do it without making a single customer feel cheated.

The trust line, drawn clearly

There is a simple test for whether a pricing move is on the safe side of the line. Ask: if a customer found out exactly how this price was set, would they feel the system was fair? A regional price, a loyalty discount, a first-order coupon, a clearance markdown on last season’s stock, all pass. A customer who learns the rule nods and moves on. A price that is higher because a model decided this particular shopper would tolerate it fails instantly. The customer who learns that rule feels punished for being loyal or for being in a hurry.

That test is why we steer clients toward the Steam model and toward rule-based pricing, and away from opaque personalized pricing, even though the latter can show a short-term margin bump. The bump is real and the backlash is also real, and the backlash compounds. Trust is the one asset in ecommerce you cannot A/B test your way back to once you have spent it.

What this looks like as a service

Sony and Steam have armies of engineers and data scientists to build this. Most ecommerce businesses do not, which is exactly the gap worth closing, and it is why we built a dedicated practice around it. SEMPITE builds dynamic pricing and personalized-promotion engines for ecommerce businesses that want the Steam-and-Sony playbook without a team of fifty.

A typical engagement starts with a signal audit, mapping the behavioral data you are already collecting and the gaps worth instrumenting. From there we build the personalized-promotion layer first, the low-risk, high-return starting point that surfaces the right offer to the right segment at one price for all. Where it fits your market and your margins, we add rules-based dynamic pricing: regional pricing, loyalty and first-purchase offers, and demand- and inventory-aware markdowns, all transparent and defensible. An experimentation framework with control groups and clean lift measurement is baked into every change, and we set the trust guardrails that keep you on the right side of the line Sony walked up to. It pairs naturally with our eCommerce and marketplace operations and agentic-commerce readiness work.

We also publish the receipts before we sell anything. In our study We Fact-Checked 25 Viral AI-Shopping Stats, eleven of them failed verification, and we say so in public. The same standard applies here: we sell what actually moves revenue and protects trust, and we measure it, rather than promising a magic pricing number.

The bottom line

The storefronts with the most pricing power on the planet just told you where ecommerce is heading. Flat pricing and blanket sales are on the way out, replaced by systems that respond to context and to the individual shopper. The opportunity is not to copy Sony’s risk. It is to capture Steam’s upside, deliberately and transparently, before your competitors do. If you want to see what a personalized-pricing engine would look like for your catalog, start with a Brand Audit or run a free visibility check first.