In May 2016, Uber’s head of economic research, Keith Chen, told NPR’s Hidden Brain something that should make every product manager building a restaurant ordering system sit up: more riders accept a 2.1x surge multiplier than a round 2.0x one. Same order of magnitude. Roughly the same price. Different number of decimal places. Different outcome.
For ride-hailing, that’s a curiosity. For restaurant technology — where self-service kiosks, ordering apps, and digital menu boards now drive a meaningful share of QSR revenue through automated upselling — it’s closer to an unclaimed lever. Every kiosk vendor building a suggestive-selling engine is already competing for the same two or three seconds of attention on the “add fries?” screen. Few are asking whether the number displayed on that screen is doing as much psychological work as it could.
This piece lays out the research behind kiosk upsell pricing psychology — why precise numbers read as more legitimate than round ones — where that logic hits a hard limit, and what it means for how POS and kiosk vendors design, price, and pitch their upsell and cross-sell engines.
Uber introduced surge pricing to solve a real supply problem: when ride requests spike, prices rise to pull more drivers onto the road and ration scarce rides among the riders who value them most. Economically, that’s textbook supply and demand. Psychologically, it’s a minefield.
Chen’s research found that demand doesn’t fall off smoothly as the surge multiplier increases — it drops in a lurch every time the number crosses a round boundary. The jump from a 1.9x to a 2.0x multiplier suppressed ride requests more than six times as much as the equivalent jump from 1.8x to 1.9x, even though the dollar difference between those two jumps was often nearly identical. What changed wasn’t the price. It was the shape of the number.
Chen’s own explanation, given directly to NPR, is that a round multiplier like 2.0 reads as a decision a person made — an arbitrary doubling, as if “somebody just made that up” — while a number like 2.1 implies a system computed it from real inputs. Riders don’t need to understand the algorithm to trust it more; the mere appearance of precision does the reassuring.
This tracks with an older idea in behavioral economics: Kahneman, Knetsch, and Thaler’s “dual entitlement” principle, which holds that customers accept a price increase as fair only when it’s visibly tied to a real cost or condition. A price hike that looks like a decision made at the customer, with no visible justification, reads as exploitation. A price that looks like the output of a calculation — even a black-box one — borrows some of that justification by implication. Precision is a proxy for legitimacy. It doesn’t prove a price is fair. It just makes the price feel less arbitrary, which for most purchasing decisions is enough.
The 2.1x effect isn’t a one-off quirk of ride-hailing — it’s a specific case of a broader pattern researchers call the numerical precision effect, and it turns up in some surprising places.
In 2024, a study published in the Journal of Consumer Psychology tested something that looks, on paper, like it shouldn’t work: researchers compared a precise 7.7% discount against a rounded-up 8% discount on identical products. By simple maths, 8% is the better deal. Shoppers didn’t see it that way — the 7.7% discount produced higher purchase intent, because the precise figure read as a real, time-boxed calculation rather than a marketing round-up. Harvard Business Review’s coverage of the research noted that precise discount framing lifted purchase intent by as much as 21% across the nine experiments the researchers ran.
The same pattern shows up in real estate. A Cornell study examining more than 27,000 home sales across South Florida and Long Island found that homes listed at a precise price — roughly £414,000 (about $553,505), say, instead of a round £411,000 (about $550,000) — sold for about 0.73% more on average. Buyers perceived the precise listing as evidence the seller had actually calculated the home’s value, and priced their offers closer to the ask as a result. A related Columbia Business School negotiation study found the mechanism working in reverse: counteroffers stated in precise dollar amounts were read as a sign the other party was better informed, causing negotiating partners to concede more ground than they did against round counteroffers.
There’s an important asterisk here, laid out well in a California Management Review synthesis of the research: precision and roundness aren’t universally good or bad — they suit different kinds of purchase decisions. Precise numbers prompt analytical thinking and work best for functional, utilitarian choices, where a customer wants to feel they’re making a calculated decision. Round numbers evoke emotional processing and tend to perform better for experiential, premium, or aspirational purchases, where the feeling of the number matters more than its exactness. A £20.00 dessert can feel more indulgent than a £19.87 one. A £2.13 fry upsell can feel more considered than a flat £2.00. Same underlying psychology, opposite recommended treatment — depending entirely on what job the price is doing.
None of this would be worth a POS vendor’s attention if upselling at self-service kiosks were a marginal channel. It isn’t — it’s arguably the highest-leverage lever currently available in QSR technology.
Across multiple vendor case studies, kiosk orders consistently run 15–30% higher in average ticket size than orders taken by a human cashier. McDonald’s has reported that kiosk customers spend roughly £0.75–£1.50 more per transaction than counter customers (originally reported as $1–2), a gap that compounds into meaningful revenue across thousands of daily transactions per location. Crazy Bowls & Wraps, working with Elo’s kiosk hardware and upsell software, recorded a 38% increase in average order value. PDQ Chicken reported a 25% increase in average ticket size after rolling out strategic, data-driven kiosk upselling. A KFC franchise operator in Bulgaria, working with Ordering Stack, saw a 3% sales increase in just three weeks after adding an upsizing prompt to its kiosk flow.
The mechanical reason is straightforward: a kiosk never gets busy, never feels awkward suggesting a second dessert, and never skips the upsell because the line is getting long. It asks the question every single time, at the same optimal point in the order flow, to every customer. That consistency is most of why kiosks already outperform human upselling.
What’s underexplored is the wording inside that consistent ask. If a kiosk surfaces an upsell prompt on every transaction, at scale, across thousands of locations, the specific number displayed on that prompt is one of the highest-frequency, lowest-cost variables a vendor can test. A one or two percentage-point lift in acceptance rate on an “add fries” prompt, multiplied across millions of daily kiosk transactions industry-wide, is a genuinely large number — and it costs nothing to change beyond a line of code in the pricing display layer.
Translating the research into a product decision means drawing a clear line between two zones of the menu and treating each differently.
Anchor items stay round. The core menu — the flagship burger, the combo starting price, the “build your own bowl” base — is where customers form their overall impression of value and quality. This is the emotional, premium end of the spectrum, and round pricing performs better here. A kiosk vendor shouldn’t touch anchor pricing based on this framework.
Upsell and cross-sell prompts get precision. The “add fries,” “make it a large,” “add a drink,” and combo-bundle prompts are the functional, in-the-moment, small-stakes decisions the research suggests respond to precise framing — where the Uber logic and the discount-framing research point in the same direction.
Pair precision with visible personalization. Most modern kiosks already surface “recommended for you” or “customers also added” language next to upsell prompts. That framing and precise pricing reinforce the same underlying message — there’s a system behind this suggestion, not a guess — and are worth testing together as well as separately.
Here’s how that distinction might look across a typical kiosk ordering flow:
| Menu moment | Standard framing | Precision-framed variant | Why |
|---|---|---|---|
| Add-on prompt | £2.00 | £2.13 | Implies a calculated price, not a round-number upcharge |
| Combo savings | Save £1.50 | Save £1.47 | Precise savings read as a real calculation, not a marketing round-up |
| Bundle discount | 20% off | 18% off | Mirrors the 7.7%-vs-8% discount-framing research |
| Core menu item | £9.00 | £9.13 | Round pricing signals confidence on anchor items — this direction reverses |
None of these variants have been tested at a kiosk specifically — they’re a direct application of research conducted in ride-hailing, e-commerce discounting, and real estate. That gap is exactly why this belongs on a product roadmap as an experiment, not a rollout.
Precision framing has a ceiling, and the fast food industry has already run the experiment that shows where it is.
In February 2024, Wendy’s CEO Kirk Tanner told investors the company would invest roughly £15 million (reported at the time as $20 million) in digital menu boards to support, among other things, “dynamic pricing” testing beginning as early as 2025. Within days, headlines were comparing it directly to Uber-style surge pricing, #BoycottWendys was trending, and Burger King had launched a “no urge to surge” promotion capitalizing on the backlash. Wendy’s spent the following week publicly clarifying it had never intended to raise prices during busy periods, only to potentially lower them during slow ones — a walk-back serious enough that both CNN and NPR ran explainers on what had actually been said versus what customers heard.
Consumers didn’t wait for a precise multiplier or a well-framed number before reacting. The mere category of “restaurant prices that change based on demand” was enough to trigger the same reflexive distrust that round Uber surges do — because, per the dual entitlement principle again, the fairness question isn’t really about the specific number. It’s about whether the customer believes the price change serves them or serves the seller at their expense. A precisely-framed surge price is still a surge price.
This is also a category regulators are starting to watch. California’s SB 422 and Rhode Island’s SB 2268 both propose limits on self-service kiosk transactions and staffing — a reminder that kiosk pricing and ordering behavior isn’t purely a UX decision anymore in every jurisdiction.
For a POS or kiosk vendor building or refining an upsell engine, the research breaks down into a handful of concrete considerations:
Uber’s surge multiplier and a fast-food kiosk’s upsell button don’t look alike, but they’re solving a strangely similar problem: getting a customer to accept a price at the exact moment they could easily say no. The research says the shape of the number matters more than most pricing teams assume — and self-service ordering, with total control over the presentation layer and no cashier’s discretion in the way, is close to the ideal environment to test it properly. For POS and kiosk vendors, that’s a low-cost, testable question sitting inside a system that already works. That combination doesn’t come along often.
Uber’s surge multiplier and a fast-food kiosk’s upsell button don’t look alike, but they’re solving a strangely similar problem: getting a customer to accept a price at the exact moment they could easily say no. The research says the shape of the number matters more than most pricing teams assume — and self-service ordering, with total control over the presentation layer and no cashier’s discretion in the way, is close to the ideal environment to test it properly. For POS and kiosk vendors, that’s a low-cost, testable question sitting inside a system that already works. That combination doesn’t come along often.
Fill out the form below, and we will be in touch shortly.
