What Free Costs The Person Taking It

The seventy-ninth article was about a number that could not be pinned down. This one is about a finding that is well established, and about the four conditions under which acting on it will cost you money.

The Verdict, Stated First

Five claims, in descending order of confidence.

One. The effect is real and the experiment is elegant. A design that holds the price difference constant and moves only the zero, in a top marketing journal, with the paper hosted openly by two of its authors' institutions.

Two. The boundary conditions are documented and numerous. A 2022 paper names three prior exceptions before adding a fourth of its own, which is unusually well-mapped for a finding this popular.

Three. The commercially important exception is incidental costs. Where taking up an offer requires time, travel or effort, the recipient bears that whatever you charge, and a zero price can reduce demand.

Four. That exception describes most professional services. On our own arithmetic a free two-hour meeting is more expensive to the client than a paid half-hour call once their time is worth more than $33.33 an hour.

Five. Free shipping thresholds have an exact break-even and most are set below it. On our own arithmetic the required uplift is the shipping cost divided by the gross margin, which at 20 percent margin is $40 on an $8 shipping cost.

Our Grades For These Claims

Applying the scheme from the first article in this series.

Grade A for the 2007 findings, from an abstract obtained verbatim from the publisher and design details from copies of the paper hosted by two of its authors' own universities.

Grade A for the 2022 boundary condition, from body text obtained from the publisher.

Grade B for the three earlier exceptions, which reach us through that 2022 paper's characterisation of them rather than from the papers themselves.

Grade C for the magnitude of the original effect, which reaches us as "dramatically more" and "dramatically fewer" rather than as numbers.

Grade A for our own arithmetic, on entirely invented commercial parameters.

Our position: a well-established effect with unusually well-documented limits, where the limits are the commercially useful part and are almost never mentioned.

A Note On Method

Everything here is verified to August 2026.

We obtained the 2007 paper's abstract verbatim from the publisher[1], and method and discussion passages from copies hosted by two of its authors' own universities[2][3]. We did not obtain its results tables, so we report no percentages from it.

We obtained body text of the 2022 paper from the publisher[4], including its characterisation of three earlier exceptions. We did not obtain its findings, only its setup and the mechanism it proposes.

We did not obtain any of the papers describing the earlier exceptions, and report them through that characterisation[4].

The result of the original experiment reaches us as a bibliographic service's one-sentence summary[5], which gives direction and no magnitude.

All arithmetic is ours and every commercial parameter in it is invented.

This article discusses research on pricing. It is not pricing, marketing or commercial advice.

The Claim

What is being proposed, in the abstract's own words.

"When faced with a choice of selecting one of several available products (or possibly buying nothing), according to standard theoretical perspectives, people will choose the option with the highest cost-benefit difference. However, we propose that decisions about free (zero price) products differ, in that people do not simply subtract costs from benefits but instead they perceive the benefits associated with free products as higher."[1]

Four observations, ours.

The claim is specifically about zero and not about cheapness. A one cent price and a zero price are nearly identical economically and the proposal is that they are not psychologically.

The mechanism proposed is on the benefit side rather than the cost side. Not that free removes a barrier, which would be unremarkable, but that free inflates the perceived value of the thing itself.

That distinction matters for what follows. If free merely removed friction, adding other friction would not undo it, and the boundary conditions below suggest it does.

And the abstract states the design in its final sentence, which is where the paper's quality shows.

One framing point before the evidence, since this article sits at the eightieth of a hundred and the pattern is now visible. The last several articles have found famous findings whose numbers had drifted or whose conditions were narrower than advertised. This one is different in a specific way.

Here the finding is solid, the design is the best this series has examined, and the literature has documented its own exceptions explicitly. The problem is not that anyone got it wrong. It is that the exceptions are where the commercial value sits and nobody repeats them.

The 2007 Paper

The source.

Shampanier, K., Mazar, N., and Ariely, D. (2007), Zero as a Special Price: The True Value of Free Products, Marketing Science, 26(6), 742–757, November, DOI 10.1287/mksc.1060.0254[1].

Three observations, ours.

The paper is openly available from two of its three authors' institutions, one a business school and one a university engineering school, which is better access than most of what this series covers.

It appeared in Marketing Science, a quantitative journal, rather than a consumer psychology outlet, which places it among modelling work and explains the formal framing of the claim.

And the third author is the same one whose work the sixty-ninth and seventy-second articles both touched. Three appearances across this series, and in this case the underlying paper is the most accessible of the three.

The Design

What was done, from the paper itself.

"Three hundred ninety-eight subjects took part in the experiment. We use a Hershey's as the low-value product and a Lindt truffle (hereafter, 'Lindt') as the high-value product. The experiment includes a free condition (0&14), a cost condition (1&15), and a second free condition (0&10). In the 0&14 and 0&10 conditions, the price of Hershey's is 0 cents, and the prices of Lindt are 14 cents and 10 cents, respectively. In the 1&15 condition, the price of Hershey's is 1 cent, and the price of Lindt is 15 cents."[2]

Four observations, ours.

398 subjects across three conditions, in a real transaction with real chocolate, is a respectable field study rather than a questionnaire.

The products are chosen to make the trade-off real. A Lindt truffle is worth more than a Hershey's, so a participant paying a few cents more for the better one is behaving sensibly.

Three conditions rather than two is what makes this careful. The 0&10 condition tests whether the effect depends on the size of the gap, not just on the presence of zero.

And the whole design turns on the comparison between the first two, which the next section sets out.

Holding The Difference Constant

The feature that makes this a good experiment. Ours.

Four observations.

In the 0 and 14 condition, upgrading from the Hershey's to the Lindt costs 14 cents. In the 1 and 15 condition, the same upgrade costs 14 cents.

The economic decision is identical. The trade-off a participant faces, in exchange-rate terms, has not moved at all between those two conditions.

All that changed is one cent on both prices, which turns the cheaper option from nearly free into actually free.

So any difference in behaviour between the conditions cannot be attributed to the price gap, to affordability, or to the relative merit of the products. It is attributable to the zero and to nothing else, which is what a well-designed manipulation looks like.

One objection is available and the design anticipates it. A one cent price requires handling a coin and a zero price does not, so the conditions differ in transaction friction as well as in the zero.

That is a real confound in a table-based chocolate sale and it is small. The 0 and 10 condition helps here too: it shares the zero and the no-coin property with the 0 and 14 condition while differing in the gap, so comparing the two free conditions isolates the gap from the friction.

The Result

What happened, at the strength our sourcing allows.

A bibliographic service's summary of the paper records: "In contrast with a standard cost-benefit perspective, in the zero-price condition, dramatically more participants choose the cheaper option, whereas dramatically fewer participants choose the more expensive option."[5]

The paper's own discussion of the field experiment notes that "the results are similar to the hypothetical choices in Experiment 1"[2], so the real-money result matched the earlier hypothetical one.

We did not obtain the results tables and report no percentages.

Four observations, ours.

"Dramatically" is a word rather than a number, and this series has spent enough articles on magnitudes to be uncomfortable reporting one without a figure.

The direction is clear and is what makes the finding matter: share moved from the better product to the free one, against a trade-off that had not changed.

The hypothetical-to-real correspondence is worth noting, because the seventy-fifth article's central criticism of another literature was its reliance on hypothetical choices. Here the authors ran both and report agreement.

And we would want the numbers. The strength of the design deserves a magnitude, and our sourcing did not produce one, which is a gap in this article rather than in the paper.

The Authors' Own Alternatives

What they considered besides their own explanation, from the paper's discussion.

They record: "a free Hershey's involves benefits and no costs, while a Lindt for any positive price involves both benefits and costs. It is possible that options that have only benefits create more positive affect compared with options that involve both benefits and costs."[3]

And they raise a second: "Alternatively, much like the disutility of paying while consuming (paying for a vacation while experiencing it: Prelec and Loewenstein 1998), it is possible that options that involve", at which point our reproduction is cut off.[3]

Three observations, ours.

Both alternatives are affect-based rather than computational, and both are compatible with the headline finding while implying different boundary conditions.

The "only benefits and no costs" framing is the one the later boundary literature attacks, because a free offer with an incidental cost is not an option with only benefits.

And the authors raising these themselves is worth crediting. They did not present a single mechanism as settled, which is more than most retellings of their work manage.

Four Documented Exceptions

The part this article exists for, and it is unusually well mapped.

A 2022 paper in a marketing journal introduces the question: "The present research investigates whether there are conditions in which a zero price is less effective than a nonzero price at penetrating the market. If a zero price results in lower demand than a nonzero price, this would constitute a boomerang effect, that is, although companies intend to drive demand up by reducing the price to zero, the strategy produces the opposite effect."[4]

And it names the earlier work: "Three papers provide exceptions to the general rule that zero pricing boosts consumer demand."[4]

Three observations, ours.

A literature that has explicitly catalogued its own exceptions is in better health than most this series covers. Somebody counted them.

The phrase "the general rule" concedes the effect while bounding it, which is the correct framing and the one popular accounts skip.

And the exceptions are mechanistically different from each other, which the next three sections set out, so they apply in different situations rather than being one caveat stated three ways.

Value Discounting

The first exception, reported through the 2022 paper's characterisation.

It records that two earlier papers "tested a value-discounting hypothesis; consumers infer that a discounted (or free) product has a low production cost, so consumers lower their WTP for the product once the promotion ends."[4]

We did not obtain those papers and report this description.

Four observations, ours.

The cost is deferred rather than immediate. Free raises demand now and lowers willingness to pay afterwards, which no single-period measurement would detect.

The inference described is reasonable rather than irrational. A customer seeing something given away concludes it is cheap to make, and often they are right.

That makes it the most dangerous of the four for a service business. A free consultation invites the inference that consultations are cheap to provide, and the price you eventually quote has to overcome it.

And it is the one exception this series has already covered from another direction. The sixty-fourth article's dual entitlement finding says customers hold views about what a price should reflect, and a giveaway supplies evidence about that.

One caution on the timing of that cost, because it determines whether a firm ever notices. The demand rise is immediate and the willingness-to-pay fall is later, so any measurement taken during a promotion records the benefit and not the cost.

Which means the natural way to evaluate a free offer is guaranteed to flatter it. Uptake during the free period is the easy number and the wrong one, and the figure that matters is what those customers pay afterwards.

Comparison Inhibition

The second exception, and it is the most surprising.

The 2022 paper records: "A noteworthy study by Mao (2016) showed that a zero price (but not a low, nonzero price) inhibits comparisons against the regular price, thus reducing the attractiveness of the offer."[4]

We did not obtain it and report this description.

Four observations, ours.

The mechanism is that a zero price stops the customer computing the saving. A discount from forty dollars to ten is visibly worth thirty; a drop to zero is not compared to anything.

If that holds, the effect works against the promotion's whole purpose. The reason to discount is to make the saving salient, and going to zero reportedly makes it less so.

It also produces a specific and testable prescription. Stating the value being given away should restore the comparison, which is why "a free consultation, normally two hundred dollars" is a different offer from "a free consultation."

And we would flag that as our own inference. The paper is described as showing the inhibition, not as testing the remedy, and we did not read it.

Two features of this exception make it the most useful of the four for a small firm, ours. The remedy costs nothing: stating a value is a sentence, not a change to the offer.

And it does not require choosing between free and paid. You can keep the zero and restore the comparison, which is not true of the incidental-cost exception, where the only remedy is to demand less of the customer.

The Boomerang Paper

The third and fourth exceptions, from the paper that catalogued the others.

Fan, X., Cai, F. C., and Bodenhausen, G. V. (2022), The boomerang effect of zero pricing: when and why a zero price is less effective than a low price for enhancing consumer demand, Journal of the Academy of Marketing Science, 50(3), 521–537, May, DOI 10.1007/s11747-022-00842-1[4][6].

Its proposal: "We propose that the effect of zero (vs. low, nonzero) pricing on consumer demand depends on the incidental costs: nonmonetary costs that are inherently involved in acquiring or using the product (e.g., the time required to commute to the store and/or wait in line). Consumers bear the same incidental costs regardless of the product's monetary price."[4]

We obtained the paper's setup and mechanism and not its findings, so we report what it proposes and tests rather than what it concluded.

Incidental Costs

Why that is the exception a service business should care about. Ours.

Four observations.

The final sentence is the whole argument in one line. Consumers bear the same incidental costs regardless of the monetary price, so dropping the price to zero does not reduce what the offer actually costs them.

Which means the zero has a diminishing relationship to the total. Where the monetary price was most of the cost, removing it removes most of the cost. Where the time was most of the cost, removing the price changes very little.

The examples given are commuting and queuing, and the professional-services version is obvious once stated. An hour of a client's attention is an incidental cost you cannot waive.

And a chocolate bar at a table has almost no incidental cost, which is precisely why the original experiment worked so cleanly. The design that best isolates the effect is also the design least like a service business.

The Dragging-Down Effect

A companion finding from the same authors, which we can name and not evaluate.

A bibliographic service records a related paper by the same team: "Four studies, across a range of domains, find a dragging-down effect in which consumers purchase fewer units of a product when a discount applies to more units."[5]

We did not obtain it and report this summary.

Three observations, ours.

If it holds, it is directly contrary to the intuition behind volume discounting. A larger required quantity for the discount reduced purchases, which is the opposite of what a threshold is meant to do.

It bears on the free shipping arithmetic below, and we would want it read alongside. A threshold that customers must reach to earn a benefit may reduce the average order rather than raise it, on this description.

And this is one summary sentence about four studies we did not read. We flag it because it points somewhere important and we cannot evaluate it, which is the honest position.

A Tension We Cannot Resolve

Two findings in the same paper that appear to pull against each other, which we raise rather than paper over. Ours.

Four observations.

The incidental cost mechanism says nonmonetary costs undermine the zero, because the customer bears them whatever you charge.

The pseudo-free finding says positive reactions extend to offers requiring a survey, personal information or watching an advertisement, all of which are nonmonetary costs the customer bears.

Both appear in the same paper's opening pages, so the authors plainly do not regard them as contradictory, and the resolution is presumably about which costs are salient at the moment of choosing. A survey you complete afterwards is different from a queue you can see.

But we did not obtain the findings and cannot tell you. We flag the tension because a reader applying this to their own offer needs to know which of the two their situation resembles, and we can pose that question and not answer it.

Why This Experiment Is Worth Copying

A methodological note, because the design is better than most in this series and the lesson transfers. Ours.

Four observations.

The manipulation changes one cent on two prices and holds everything a rational actor would care about constant. That is the whole art of it.

Compare the seventy-fifth article's contested finding, where the challengers and the original authors disagreed about whether replications had reproduced the conditions. Here there is nothing to disagree about, because the two conditions differ in exactly one specified way.

The third condition does further work most designs skip. Testing a second free condition with a different gap separates the zero from the size of the discount, which a two-condition study could not.

And the transferable lesson for anyone running their own test is the same principle. Change one thing, hold the trade-off constant, and add a third arm that rules out the obvious alternative, which is a more useful takeaway than the finding itself for a business that wants to test something.

What Actually Survives

Our reading, stated directly.

Five statements.

The effect is real and well demonstrated. A design holding the trade-off constant and moving only the zero, in a field study with 398 participants, replicating an earlier hypothetical result.

The mechanism is proposed rather than settled, and the original authors raised two alternatives to their own account in the paper itself.

Four exceptions are documented in the literature, mechanistically different from one another, and catalogued explicitly by a 2022 paper.

The commercially important one is incidental costs: what a customer spends in time and effort does not change when you drop the price to zero.

And the magnitude of the original effect does not reach this article as a number, only as the word dramatically.

What A Free Offer Costs Its Taker

Turning the incidental cost point into arithmetic. Our own calculation, entirely invented figures.

Compare a paid 30-minute call at $50 against a free two-hour meeting, from the client's side, valuing their own time:

At $50 an hour: the paid call costs them $75, the free meeting $100.

At $100: $100 against $200. At $150: $125 against $300. At $200: $150 against $400. At $300: $200 against $600.

Four observations.

The free option is the more expensive one at every rate shown, and the gap widens as the client's time becomes more valuable.

That is not a paradox and it does not require any behavioural finding. It is arithmetic on the total cost of accepting an offer, and only the monetary component was set to zero.

The reason it feels surprising is that only one of the two costs appears on your website. The client's two hours are real and invisible to you.

And the direction of the error is worth stating. A firm designing a free offer optimises the number it can see, and lengthens the thing it cannot.

Two consequences follow that most firms have backwards. A longer free consultation feels more generous and is more expensive to the recipient, so generosity and cost move in opposite directions here.

And the customers who decline are not the ones you would assume. The people whose time is most valuable face the highest cost, which means a long free offer selects against exactly the clients most able to pay.

Thirty-Three Dollars An Hour

Where the two lines cross. Ours, same invented figures.

Setting the two totals equal gives a client hourly value of $33.33.

Four observations.

Above $33.33 an hour, the free two-hour meeting costs the client more than the paid half-hour call. Below it, the free version is genuinely cheaper for them.

That threshold is low. Essentially every client a professional services firm would want is above it, which means the offer is more expensive to exactly the people it is aimed at.

The formula generalises and is worth carrying. The free option wins only while the extra time it demands, valued at the client's rate, is less than the fee you waived.

And it gives a design instruction that costs nothing. Shorten the free thing rather than lengthening it, because on this arithmetic duration is the price you are actually charging.

Two refinements to that instruction, ours. Travel counts and is often the larger half. A twenty-minute call and a twenty-minute meeting across town are different offers, and only one of them is short.

And scheduling friction is part of the price too. An offer requiring three emails to arrange has charged the client for those emails, and the fee you waived was probably smaller.

The Free Consultation

The first application. Ours, untested, and not marketing advice.

Four points.

A free consultation faces all four documented exceptions at once, which is unusual. It carries a large incidental cost, invites the inference that advice is cheap, offers no comparison to a stated value, and is the kind of offer whose quantity requirement can drag down uptake.

The incidental cost is the binding one on our arithmetic. Two hours of a client's time is worth more than the fee most firms would charge for the same conversation, so the offer is not the concession it appears to be.

The comparison-inhibition exception suggests a cheap fix. Naming what the session is worth restores the comparison a bare zero reportedly suppresses, and it costs nothing to state.

And the value-discounting exception suggests a harder one. The inference that free advice is cheap advice does not go away when you send an invoice later, and a firm giving away its core service is teaching a lesson about that service.

The Free Shipping Threshold

The second application, and it has an exact break-even. Our own arithmetic, invented figures: an average order of $45, shipping costing $8, a threshold at $60.

The margin earned on the extra $15, against the $8 of shipping given up:

At a 60 percent margin: $9.00 earned against $8.00 forgone, net plus $1.00.

At 50: $7.50, net minus $0.50. At 40: $6.00, net minus $2.00. At 30: $4.50, net minus $3.50. At 20: $3.00, net minus $5.00.

Four observations.

At that threshold the scheme only pays above a 53 percent gross margin, and loses money on every crossing order below it.

The failure is invisible in the obvious metric. Average order value rises, which is what everyone measures, and the margin falls.

The threshold in the example is a third above the average order, which is a common rule of thumb and is not derived from anything.

And the correct threshold is derivable in one line, which the next section does.

Where The Threshold Belongs

The rule. Ours, and it is exact given the definitions.

Break-even uplift equals the shipping cost divided by the gross margin. On an $8 shipping cost:

At a 60 percent margin: $13.33 of uplift, so a threshold of $58.33 on a $45 average order.

At 50: $16.00, threshold $61.00. At 40: $20.00, threshold $65.00. At 30: $26.67, threshold $71.67. At 20: $40.00, threshold $85.00.

Four observations.

A 20 percent margin business needs its threshold $40 above its average order merely to break even on the orders that cross it.

The rule is the same shape as the seventy-third article's credit threshold and the seventy-eighth article's stamp card. Everything divides by the margin, and a low-margin business has almost no room for any of these devices.

It also explains why free shipping is universal in high-margin categories and rare in low-margin ones, without anybody needing to know the research.

And a threshold set that high raises the question the dragging-down finding poses. A requirement large enough to pay for itself may be large enough to discourage the order entirely, and we cannot tell you where that turns.

Which leaves a low-margin business in a genuinely awkward position, and we would rather say so than offer a workaround that does not exist. The threshold that pays is too high to be attractive, and the threshold that is attractive does not pay.

The honest options are all unglamorous. Raise prices to absorb the shipping, charge for shipping and say why, or negotiate the shipping cost down, which is the observable input the seventy-sixth article said to optimise. None of them is a behavioural technique and one of them is usually available.

And Only If The Uplift Is New

The same caveat as the stamp card. Ours.

Four observations.

A customer who adds an item to cross the threshold may be buying something they would have bought next month, in which case nothing has been created.

At a 40 percent margin on the example above, the scheme nets minus $2.00 if all the uplift is new, and minus $6.50 if only a quarter is.

So unlike the stamp card, where the downside was capped at the cost of a free coffee, a badly set shipping threshold loses money on every order and loses more when the uplift is not incremental.

And the diagnostic is the same one. Compare total spend per customer per year, not order value, because order value is exactly the number this device is designed to move.

One further asymmetry between the two devices is worth naming, because it changes which is the safer bet. The stamp card only pays out when the customer completes it, so abandoned cards cost nothing and the giveaway is self-limiting.

A shipping threshold pays out on every order that crosses it, including orders that would have crossed it anyway. The customer who was always going to spend eighty dollars now gets free shipping, and that is pure cost with no behaviour change behind it.

Free Trials

The third application. Ours.

Four points.

A free trial is the offer with the highest incidental cost of any in common use. Setting something up, learning it, migrating to it, and deciding about it are hours of work the price does not touch.

On the incidental-cost mechanism, that predicts something specific. Reducing the price of a trial from low to zero should do less than reducing the setup effort at all, and effort is usually the more expensive lever to pull.

The value-discounting exception applies too and in a compounding way. A long free period teaches the customer what the product is worth, and the answer they learn is nothing.

And we would apply the same test as the seventy-eighth article. Could you explain the design to the customer without embarrassment? A trial genuinely intended to let someone evaluate the product passes; one designed so that cancelling is harder than continuing does not.

Pseudo-Free

A category the 2022 paper names, and one worth a business reader's attention.

It records: "Consumers' positive reactions extend to pseudo-free offers, those that are presented as free but require certain concessions (e.g., completing a customer survey, providing personal information, or watching an ad)."[4]

Four observations, ours.

If the positive reaction survives a required concession, then the zero is doing work independent of the true cost, which is a stronger version of the original claim rather than a qualification.

It also sits awkwardly beside the incidental-cost exception, and we would not pretend to resolve it. One says nonmonetary costs undermine the zero and the other says a required concession does not, and separating them is a question about which costs count.

The commercial and ethical reading is the same one. Free in exchange for personal information is a price, and describing it as free is accurate about the money and not about the transaction.

And we would state our own position plainly. An offer whose real price is data should say so, on the disclosure test this series has used since the seventy-second article, and the research on whether people mind is beside that point.

One practical note for a small firm, since this is where the category usually bites. Most Canadian businesses collecting personal information in exchange for something free are subject to privacy obligations, and this article does not address them. That is a question for a lawyer rather than for a behavioural literature, and it is a real one.

What To Charge Instead

Because this article should not end by recommending against every free offer. Ours.

Four observations.

The finding is real, and where the incidental cost is genuinely low, zero does something a low price does not. A free sample handed to someone already standing there is the original experiment's condition.

Where the incidental cost is high, the arithmetic says the useful lever is elsewhere. Shorten it, simplify it, remove the setup, come to them, all of which reduce the cost the customer actually bears.

And a nominal price has one property zero does not, which the four exceptions collectively point at. It preserves a comparison, it signals a value, and it filters for intent, and none of those is available at zero.

So the choice is not between free and expensive. It is between free and nominal, and the literature gives reasons on both sides that depend entirely on how much time your offer demands.

Two things a nominal price does that are worth more than they cost, ours and untested. It filters for intent: a person paying twenty dollars for a session has decided something a free booking does not require them to decide, which is why free bookings are missed more often.

And it establishes that a price exists, which the value-discounting exception says a zero does not. Moving from twenty dollars to a full fee is a different negotiation from moving from nothing to a full fee, and the first has a number in it.

What To Do

Add up what the offer costs the person taking it. On our own arithmetic a free two-hour meeting costs a client more than a paid half-hour call once their time is worth over $33.33 an hour.

Shorten the free thing rather than lengthening it. On the incidental-cost mechanism, duration is the price you are actually charging.

State what the free thing is worth. One documented exception is that a zero price inhibits comparison against the regular price, and naming the value restores it at no cost.

Compute your shipping threshold rather than guessing it. Break-even uplift is the shipping cost divided by your gross margin, which at 20 percent margin is $40 on $8 of shipping.

Watch margin, not average order value. A badly set threshold raises the metric everyone measures and lowers the one that matters.

Ask whether the uplift is new business. A customer buying next month's item today has not increased anything, and the scheme still paid for the shipping.

Expect free advice to teach that advice is cheap. One documented exception is that consumers infer low production cost from a free or discounted price and lower what they will pay later.

Say what the real price is when it is not money. An offer paid for with personal information is free about the money and not about the transaction.

The Limits Of This Analysis

Several caveats matter. This article discusses research on pricing and is not pricing, marketing or commercial advice; the applications are our own reasoning and untested. Everything is verified to August 2026. We did not obtain the 2007 paper's results, only its abstract from the publisher and its method and discussion passages from copies hosted by two of its authors' institutions; the magnitude of the central effect reaches this article as the word "dramatically" and not as a number, which for a series that grades magnitudes is the largest gap here. We did not obtain the 2022 paper's findings, only its setup, its proposed mechanism and its characterisation of earlier work, so we report what it proposes and tests rather than what it concluded. We did not obtain any of the three earlier exception papers, and every statement about value discounting and comparison inhibition reaches this article through the 2022 paper's one-sentence descriptions of them. We did not obtain the dragging-down paper, and report a single summary sentence about four studies. The tension between the incidental-cost exception and the pseudo-free finding is one we raise and do not resolve, because resolving it would require reading both. All arithmetic is ours and every commercial parameter in it is invented: the fee, the meeting lengths, the hourly rates, the average order, the shipping cost and the margins. The $33.33 crossover is exact given the two specific offers we constructed and moves with any change to either. And the free shipping arithmetic assumes the customer crosses the threshold by buying more rather than by abandoning the order, which the dragging-down finding gives reason to doubt and which we could not evaluate.

Related Reading

No. 81 Nobody Has To Be Wrong Against five bidders with unbiased estimates, the winner overestimates by 1.16 standard deviations. No bias is required. Winning selects for the... Behavioural Finance The Rule Your Customers Already Know What customers infer about a price and what it is supposed to reflect. Behavioural Finance When The Conditions Are Essentially Replicated A pricing finding whose conditions were narrower than its reputation.

Frequently Asked Questions

What is the zero price effect?

The proposal that people do not simply subtract costs from benefits, but perceive the benefits of free products as higher. It was tested by holding the price difference between two chocolates constant at 14 cents while moving the cheaper one from one cent to zero, so the economic trade-off was identical and only the zero changed.

How strong is the effect?

We could not obtain a number. The result reaches this article as dramatically more participants choosing the cheaper option and dramatically fewer choosing the more expensive one. The design is excellent and the magnitude is a gap in our sourcing rather than in the paper.

When does free stop working?

Four documented exceptions. Consumers may infer low production cost and lower what they will pay later; a zero price may inhibit comparison against the regular price; incidental nonmonetary costs are borne regardless of price; and a related finding reports fewer units purchased when a discount requires more units.

Why does a free consultation cost the client money?

Because only the monetary component was set to zero. On our own arithmetic, a free two-hour meeting costs a client more than a paid thirty-minute call once their time is worth more than $33.33 an hour, and every client a professional firm would want is above that.

Where should a free shipping threshold sit?

On our own arithmetic, break-even uplift equals the shipping cost divided by your gross margin. At a 40 percent margin and $8 shipping that is $20 of uplift; at 20 percent it is $40. A threshold a third above your average order pays only above about a 53 percent margin on the figures we used.

What should I measure?

Margin and total annual spend per customer, not average order value. Order value is the number the device is designed to move, so watching it tells you the mechanism worked rather than whether it paid. And ask whether the uplift was new business or the same business earlier.

Should I ever offer something free?

Where the incidental cost is genuinely low, yes: a free sample handed to someone already present is close to the condition the original experiment tested. Where accepting the offer demands hours of the customer's time, the useful lever is shortening it rather than pricing it at zero.

References

1. Shampanier, K., Mazar, N., & Ariely, D. (2007). Zero as a Special Price: The True Value of Free Products. Marketing Science, 26(6), 742–757, November, DOI 10.1287/mksc.1060.0254. Publisher record reproducing the abstract: on people, when faced with a choice of selecting one of several available products or possibly buying nothing, being expected by standard theoretical perspectives to choose the option with the highest cost-benefit difference; on the authors proposing that decisions about free zero price products differ, in that people do not simply subtract costs from benefits but instead perceive the benefits associated with free products as higher; and on the authors testing this by contrasting demand for two products across conditions that maintain the price difference between the goods but vary the prices such that the cheaper good in the set is priced at either a low positive or zero price. The same record carries the paper's reference list, and a university research profile independently confirms the citation, volume, issue, pages, DOI and publication date. Note: the publisher's record. We obtained the abstract verbatim and not the paper's results, so no magnitude for the central effect is reported here. pubsonline.informs.org
2. Copy of the same paper hosted by the third author's own university, reproducing method and discussion passages: that three hundred ninety-eight subjects took part in the experiment; that the authors used a Hershey's as the low-value product and a Lindt truffle as the high-value product; that the experiment included a free condition with prices of zero and fourteen cents, a cost condition with prices of one and fifteen cents, and a second free condition with prices of zero and ten cents; that in the two free conditions the price of the Hershey's was zero cents and the prices of the Lindt were fourteen and ten cents respectively, while in the cost condition the Hershey's was one cent and the Lindt fifteen; and that the results were similar to the hypothetical choices in the paper's first experiment. Note: a copy hosted by one of the paper's authors at his own university. Our source for the design and sample size; we did not obtain the results tables. people.duke.edu
3. A second copy of the same paper hosted by another of the third author's institutional pages, reproducing discussion passages in which the authors consider alternative explanations for their own finding: that a free Hershey's involves benefits and no costs while a Lindt at any positive price involves both benefits and costs, so it is possible that options with only benefits create more positive affect than options involving both; and that alternatively, much like the disutility of paying while consuming, citing Prelec and Loewenstein (1998), it is possible that options which involve, at which point our reproduction is cut off. Note: a second institutional copy of the same paper. Recorded because it preserves the authors' own alternative explanations, which retellings of this finding omit; our reproduction of the second alternative is truncated. web.mit.edu
4. Publisher record for Fan, X., Cai, F. C., & Bodenhausen, G. V. (2022), The boomerang effect of zero pricing: when and why a zero price is less effective than a low price for enhancing consumer demand, Journal of the Academy of Marketing Science, 50(3), 521–537, May, DOI 10.1007/s11747-022-00842-1, reproducing body text: that consumers' positive reactions extend to pseudo-free offers, those presented as free but requiring certain concessions such as completing a customer survey, providing personal information or watching an advertisement, citing Dallas and Morwitz (2018); that the research investigates whether there are conditions in which a zero price is less effective than a nonzero price at penetrating the market, which would constitute a boomerang effect; that the authors propose the effect of zero versus low nonzero pricing depends on incidental costs, being nonmonetary costs inherently involved in acquiring or using the product such as the time required to commute to a store or wait in line, which consumers bear regardless of the product's monetary price; that three papers provide exceptions to the general rule that zero pricing boosts consumer demand; that Raghubir (2004) and Kamins and colleagues (2009) tested a value-discounting hypothesis in which consumers infer that a discounted or free product has a low production cost and lower their willingness to pay once the promotion ends; and that a study by Mao (2016) showed a zero price, but not a low nonzero price, inhibits comparisons against the regular price and thus reduces the attractiveness of the offer. Note: the publisher's record. Our source for the boundary conditions; we obtained the setup, mechanism and characterisation of earlier work, and not the paper's findings. Every statement here about the three earlier exceptions reaches us through this paper's one-sentence descriptions of them. link.springer.com
5. Bibliographic service record for the 2007 paper, reproducing the abstract and carrying summaries of related work: recording that in contrast with a standard cost-benefit perspective, in the zero-price condition dramatically more participants chose the cheaper option whereas dramatically fewer chose the more expensive option; and summarising a related paper by the same authors reporting that four studies across a range of domains find a dragging-down effect in which consumers purchase fewer units of a product when a discount applies to more units. Note: a bibliographic service record. Our only source for the direction of the original result, which it gives without magnitude, and for the dragging-down companion finding, which reaches us as a single summary sentence about four studies we did not obtain. semanticscholar.org
6. Economics database record for a paper on the zero-price effect in a multicomponent product context, DOI 10.1016/j.ijresmar.2016.01.009, whose reference list confirms Shampanier, Mazar and Ariely (2007), Marketing Science, 26(6), 742–757, and whose citation record confirms Fan, X., Cai, F. C., and Bodenhausen, G. V. (2022), Journal of the Academy of Marketing Science, 50(3), 521–537, May, and Mazar, N., Shampanier, K., and Ariely, D. (2017), When Retailing and Las Vegas Meet: Probabilistic Free Price Promotions, Management Science, 63(1), 250–266, January. Note: an economics database record; citations only. Used to confirm the 2022 citation independently and to identify the companion papers. We obtained none of the works named here. ideas.repec.org

This article discusses research on pricing and is not pricing, marketing or commercial advice. The 2007 paper's results were not obtained, so the magnitude of the central effect is reported as the word used by a summarising source rather than as a number. The 2022 paper's findings were not obtained, only its setup and proposed mechanism; the three earlier exceptions reach this article through that paper's one-sentence descriptions. All arithmetic is the authors' own and every commercial parameter in it is invented.