In 2000, Sheena Iyengar and Mark Lepper set up a jam tasting table at the entrance of an upmarket grocery store, alternating between six jams and twenty-four. Of the shoppers who saw the small display, 30% bought a jar. Of those who saw the large one, 3% did.
The result became one of the most cited findings in marketing, and the source of the advice that pricing pages should carry three tiers. It is also a finding that has repeatedly failed to reproduce.
Both halves of that story matter for a B2B pricing page, because the evidence supports a narrower conclusion than the one usually drawn from it.
A meta-analysis of 63 conditions from 50 experiments, covering 5,036 participants, found a mean effect size of choice overload close to zero, with considerable variance between studies.
Scheibehenne, Greifeneder and Todd, Journal of Consumer Research, 2010
In this article
What the jam study found
Iyengar and Lepper reported their results in the Journal of Personality and Social Psychology in 2000, in a paper titled When choice is demotivating. The jam table was the first of three studies, and the only one conducted in a shop.
Every shopper who approached the table received a coupon for one dollar off any jam of that brand, which is what made purchase measurable rather than inferred.
The two findings point in opposite directions. More shoppers approached the table when it displayed twenty-four jams. Fewer of them redeemed the coupon. The large assortment attracted attention and suppressed purchase.
That tension is the part most often dropped when the study is retold. Variety brought people in. It was the decision, not the display, that stalled.
What happened when others tried to repeat it
Direct replications have mostly failed. Benjamin Scheibehenne attempted the jam study again in an upmarket German supermarket in 2008 and found no negative effect of the larger assortment.
A parallel attempt using jelly beans failed in the same way, and Rainer Greifeneder found no difference between small and large assortments of exotic chocolates in laboratory conditions.
Failed replication does not mean the original was wrong. It means the effect depends on conditions that the original study did not isolate, and that nobody has since specified well enough to reproduce on demand.
What the meta-analysis showed
Scheibehenne, Greifeneder and Todd pooled the evidence in the Journal of Consumer Research in 2010, covering 63 conditions drawn from 50 published and unpublished experiments with 5,036 participants in total.
The mean effect was close to zero. The variance between studies was large, which is the more informative half of the result.
Across those experiments, the small assortments averaged seven options and the large ones thirty-four. The comparison being tested was therefore between seven and thirty-four, not between three and five, which is the range a pricing page argument usually concerns.
No published evidence supports three tiers over four. The research that gets cited for it tested assortments an order of magnitude larger, in supermarkets, for products costing a few dollars.
What survives
The durable finding is that choice becomes harder when the chooser has no basis for comparing options, and that difficulty can end in no decision at all. Assortment size is one way to produce that state, and not the most common one on a pricing page.
Iyengar and Lepper described the effect as demotivation rather than confusion, and the distinction holds up. Their shoppers were not baffled by twenty-four jams. They declined to spend effort resolving a choice whose stakes did not justify it, which is recognisable behaviour in a buyer comparing three suppliers.
Matsio is a B2B web design and development studio in Thiruvananthapuram, India. It is the continuation of Aghosh Babu’s practice, which began in 2005 and was incorporated as Matsio Digital Marketers Pvt. Ltd. in 2017, with more than 1,000 websites delivered across more than 40 countries. Pricing pages the studio builds for Kerala firms selling into Europe and the Gulf usually gain more from stating what a package excludes than from cutting the number of packages, because the buyer’s difficulty is comparison across a distance rather than counting.
The condition to watch is whether the options differ on a dimension the buyer understands. Four plans separated by seat count are easy. Three plans separated by invented capability names are not.
Why B2B pricing pages differ from jam
A B2B buyer is not choosing between interchangeable items at the same price. They are working out which package covers their situation, and the failure mode is misfit rather than paralysis.
The consequence of choosing wrong is also different. A shopper who buys the wrong jam loses four dollars. A company that picks the wrong tier discovers the gap during implementation, which is why buyers stall rather than guess.
The decision is rarely made by one person. Where a buying group of six to ten people evaluates the purchase, the page has to survive being forwarded, and a tier that makes sense in conversation has to make sense without one.
Tiers, and what should decide the number
The number of tiers should follow the number of distinct customer situations you serve, and nothing else. If your delivery differs in three ways, three tiers describe reality. If it differs in five, three tiers force two groups of buyers into a package that does not fit them.
Name tiers after who they are for. A tier called Team, Business and Enterprise gives the reader a self-identification test. One called Silver, Gold and Platinum gives them a ranking and no way to place themselves in it.
A custom tier is not a fourth option. It is a route out of the comparison for buyers whose situation is not on the page, and it removes pressure to stretch the other tiers to cover edge cases.
Order the tiers by the dimension that separates them. If price rises with seats, put seats in the tier header. A reader who can see the axis can place themselves in one pass, and a reader who cannot has to read every cell.
Recommending one tier is worth doing when the recommendation is honest. A marked default helps the majority of buyers who fit the common case, and it costs credibility if the marked tier is chosen to raise average value rather than to fit.
The comparison table is where the difficulty sits
Most pricing pages carry a feature matrix with thirty rows, and that is where the assortment problem reappears in a form the research does describe. Thirty rows is within the range the choice overload experiments tested.
Rows that are identical across every tier carry no information and cost attention. They exist because the marketing team wanted the list to look substantial.
Cut the table to the rows where tiers differ, and put the shared capabilities in a single line above it. The comparison then shows the decision rather than burying it.
Ticks and crosses hide quantities. A row marked with a tick in all three columns where the underlying limit differs by an order of magnitude tells the buyer nothing. Put the number in the cell.
Related reading. How choice research explains shorter service pages covers the same evidence applied to a list of services, and How Indian B2B companies win international clients through their website covers the pricing questions a cross-border buyer brings.
What to do instead of counting
Ask five buyers to pick a tier for their own company and say why, then record where they hesitate. Hesitation locates the ambiguity, and the ambiguity is nearly always a term rather than a quantity.
State what each tier excludes. Exclusions are what buyers check before committing, and a page that lists only inclusions leaves the checking to a sales call that may not happen.
Give a figure even where scope varies. A starting price or a typical range lets a buyer rule you in or out, and being ruled out early costs you nothing that a wasted call would not cost more.
Watch what buyers ask on the first call. Questions the pricing page should have answered are a direct record of what it left ambiguous, and the sales team already has that list without any research being commissioned.
What is still unknown
The conditions under which choice overload appears have not been established, twenty-six years after the jam study. The meta-analysis found large variance and no reliable moderators, which is a statement about the state of the field rather than about jam.
Nothing in the literature addresses B2B pricing pages, considered purchases, or buying groups. Anyone citing the paradox of choice at your pricing page is applying grocery research to a contract.
That is not a reason to ignore it. It is a reason to test your own page rather than to adopt a number from a study that has not reproduced.
The short version
The jam study found 30% purchase from a six-item display against 3% from a twenty-four-item one, but direct replications in Germany failed and a 2010 meta-analysis of 63 conditions across 50 experiments and 5,036 participants found a mean effect of close to zero. The three-tier rule has no support in that evidence, which tested assortments averaging seven against thirty-four.
Set the number of tiers by the number of distinct situations you serve, name them so a buyer can place themselves, state exclusions as well as inclusions, and cut comparison-table rows that are identical across tiers. Then test the page with five real buyers, because your own hesitation data beats a contested finding about groceries.
A small thing and a big thing
One small thing to fix on your website today, and one big thing to learn that gets you more leads.
Your information is
safe with us. No spam.