A Dress Rose From $34 to $39—and Demand Climbed One-Third.

In a women’s clothing catalog, raising a dress price from $34 to $39 increased demand by roughly one-third. Raising it again to $44 produced no comparable gain, according to research by Eric Anderson and Duncan Simester. That result changes the pricing decision: the final digit does more than alter the amount a customer pays; it can tell the customer what kind of offer they are seeing. The useful question is not whether every price should end in nine. It is whether the price communicates “bargain,” “premium,” “simple” or “suspect” before the customer has enough information to judge value directly.

The Last Digit Became a Bargain Cue

Anderson and Simester ran three field experiments with mail-order catalogs, changing price endings while real customers made real purchases. Their 2003 paper in Quantitative Marketing and Economics reported higher demand from 9-ending prices in all three studies. The increase was strongest for new items and weaker for products the retailer had sold in previous years.

That boundary matters more than the famous dress result. Customers already familiar with an item have memory, prior prices and experience to guide them. A new item arrives with less evidence. Its price must therefore perform two jobs: state the cost and help the customer classify the offer.

A 9 ending had become familiar retail language for “this is a good deal.” It could raise demand even when using it required raising the price. That cannot be explained by a simple left-digit mistake, because $34 and $39 begin with the same digit. The ending was functioning as a signal.

Customers Borrow Meaning From the Seller

Most buyers do not carry a reliable market price for every dress, software plan, repair service or business tool. Anderson and Simester made that problem central in their Harvard Business Review analysis, “Mind Your Pricing Cues.” When customers lack a strong internal reference, they look to the seller’s presentation for clues: a sale label, a comparison price, a guarantee or a familiar ending.

Marketing works here by reducing interpretation work. The customer is not merely calculating dollars. They are asking whether the offer belongs in the bargain category, whether it appears fairly priced and whether the seller understands the conventions of the market.

That also explains why the effect is contextual. A 9 ending may support a value-oriented retailer, a trial offer or an unfamiliar product whose benefit is difficult to compare. The same ending can weaken a luxury service, a high-trust professional engagement or a deliberately simple subscription. In those settings, a round number may communicate confidence and ease rather than discounting.

The Signal Weakens When the Context Says It Twice

The catalog experiments also found evidence that 9 endings were less effective when a “Sale” cue was already present. The two devices were apparently doing similar work. Once the promotion label had classified the offer as a bargain, the final digit added less information.

This is where marketers misuse psychological pricing. They stack a crossed-out reference price, a countdown, a bright badge, a coupon and a 9 ending onto one offer, then credit whichever element survives the reporting dashboard. The result may produce urgency, but it can also train customers to distrust the regular price or postpone buying until the next promotion.

The ethical line is straightforward. A price cue should help customers interpret a genuine offer, not manufacture a false comparison. A business that cannot explain the reference price, the discount window or the product difference is using confusion as a conversion tool. Short-term response can then hide damage to margin, repeat purchase and brand credibility.

Test the Meaning, Not Just the Number

A small business can test one comparable offer at two price endings while keeping the product, audience, message and timing constant. The useful outcome is contribution margin, not units alone: a higher conversion rate can still lose money, while a modest response at a higher price may improve the economics.

A midsize company can separate new from returning customers and unfamiliar from established products. That is closer to the condition the research identified. An enterprise can add regional or category holdouts, monitor returns and repeat purchase, and prevent different channels from teaching conflicting price positions.

This is where OrionPilot’s connected approach becomes practical. A pricing experiment should keep the strategic hypothesis attached to the creative treatment, channel execution, customer response and final analysis; otherwise one team tests “value signaling” while another adds a sale badge that changes the experiment. Connecting those decisions turns a clever number into evidence the business can actually reuse.

The $39 dress is memorable because the higher price won. Its durable lesson is more disciplined: customers read prices as communication. Choose the ending that fits the meaning of the offer, then measure whether that meaning improved both the sale and the business behind it.