The paradox of choice is the idea that giving people more options can make them less likely to choose anything and less happy with what they pick.
The research behind it is mixed: a 2010 meta-analysis found an average effect of virtually zero, so fewer options only sell more when the choice is hard, the options are hard to compare, or shoppers don't know what they want.
Our A/B tests show the same split: cutting options has won big and lost big.
The phrase comes from psychologist Barry Schwartz's 2004 book The Paradox of Choice: Why More Is Less.
Two separate ideas usually get bundled under it: Hick's law, which is about how fast people decide, and choice overload, which is about whether they decide at all.
They make different predictions, so this page treats them separately.
For the wider family of decision biases, see our psychology hub.
Hick's law: more options, slower decisions
Hick's law says the time it takes to choose grows with the number of options, but only logarithmically.
It comes from two lab studies: W. E. Hick's 1952 paper "On the rate of gain of information" and Ray Hyman's 1953 follow-up, which tied reaction time to the amount of information in the choice.
The formula
For equally likely options, the usual form is:
T = a + b × log2(n + 1)
| Term | Meaning |
|---|---|
| T | Average time to make the choice |
| a | Base time that doesn't depend on the number of options (seeing, moving) |
| b | Time per bit of information, measured for a given task and person |
| n | Number of equally likely options; the +1 covers the choice of whether to respond at all |
A worked example
The log term is what matters for design.
Going from 3 options to 7 changes log2(n + 1) from log2(4) = 2 to log2(8) = 3: one extra bit.
Going from 7 to 15 adds just one more bit, log2(16) = 4.
So each doubling of options adds the same fixed amount of time, not double the time.
Trimming a menu from 10 items to 8 changes log2(11) ≈ 3.46 to log2(9) ≈ 3.17, a cut of about 8% in the variable part of the decision.
Hick's law alone predicts small gains from small cuts.
Where Hick's law is the wrong model
The law was measured on simple reaction tasks: a light comes on, press the matching key.
It describes people who already know every option and just have to pick one. Most web pages don't work like that.
- Unfamiliar or unordered lists. When people have to read each item to find the one they want, search time grows roughly in a straight line with list length, and Hick's law does not apply. Clear grouping and labels matter more than the raw count.
- Familiar options. Studies of verbal responses to familiar stimuli found little or no slowdown as options increased. Returning visitors who know your nav are less affected.
- Speed isn't the goal. A purchase decision is slow for reasons the formula doesn't model: price, trust, fit. Saving 100 milliseconds of decision time rarely moves revenue.
Hick's law is better at explaining navigation and menus than buying decisions, and our navigation tests fit that.
Harvard Business Services cut its homepage nav from 10 tabs, removing Blog, About Us and Contact Us, renaming two tabs and adding a "How to Incorporate" tab.
Completed orders rose 15.68% across about 32,000 visitors.
But several changes shipped at once, and the clearer labels probably did as much work as the lower count.
Hick's law sits beside related ideas in our cognitive load guide: what slows people down is usually the effort each option demands, not the count itself.
Choice overload: when more options mean fewer sales
Choice overload is the claim that big assortments make people put off the decision, choose nothing, or feel less satisfied afterward.
It is the part of the paradox of choice that would actually cost you sales, and it is the part with the shakiest evidence.
What the jam study found
The famous evidence is Sheena Iyengar and Mark Lepper's 2000 study, run at Draeger's, an upscale grocery store in Menlo Park, California.
A tasting booth showed either 24 jams or 6. According to the original paper:
| 24 jams | 6 jams | |
|---|---|---|
| Passersby who stopped | 60% (145 of 242) | 40% (104 of 260) |
| Stoppers who bought a jar | 3% (4) | About 30% (31) |
The big display drew more people in but sold far fewer jars.
The same paper reported that students were more likely to write an optional essay when given 6 topics rather than 30 (74% vs 60%).
The replication problem
The jam study was a single field test with 4 versus 31 purchases.
Later work did not find a reliable general effect.
Scheibehenne, Greifeneder and Todd's 2010 meta-analysis pooled 63 conditions from 50 experiments (N = 5,036) and found a mean effect size of virtually zero, with wide variation between studies.
Some studies found overload, some found that more choice helped, and many found nothing.
That variation turned out to be the useful part.
Chernev, Böckenholt and Goodman's 2015 meta-analysis of 99 observations (N = 7,202) found four factors that make overload more likely:
- Choice set complexity: options that are hard to compare, or no clear best option.
- Decision task difficulty: time pressure, many attributes, or a confusing layout.
- Preference uncertainty: people who don't know what they want or know little about the category.
- Decision goal: people trying to decide with as little effort as possible, rather than browsing.
The fair summary: choice overload is real under some conditions and absent under others.
It is not a rule that fewer options sell more.
What our A/B tests show
Tests in the LiftArchive database line up with the meta-analyses.
Cutting options wins when the extra options were noise, and loses when they were options people actually needed.
DRIP Agency
Conversion Fanatics
MarketingExperimentsFewer options won. DRIP Agency cut the New Shades collection page to about a quarter of its products, and revenue rose 14%.
Blinds are a category where shoppers often don't know what they want, which is the preference-uncertainty case.
The case study gives only the headline figure, with no sample size or significance.
In apps, Vahe Baghdasaryan's redesign of a party game's paywall moved the monthly plan behind "View All Plans" and shortened the page.
It reported +31% install-to-trial and +64% revenue, though the headline, button and trial toggle changed at the same time.
Fewer visible options won, but removing them lost. Conversion Fanatics ran two versions of the same mobile product page.
Showing only the key option groups, with easier-to-tap choices, gave +36.6% mobile conversions.
Expanding every option group gave about -4% after leading early.
The options didn't change; how many showed at once did.
Fewer options lost. MarketingExperiments removed a subscription-length dropdown so visitors were locked into monthly, and conversion fell about 40%.
Patrick McKenzie tested Stripe-only checkout against three payment methods on Bingo Card Creator, and three options won, weakly, at about 90% significance.
Plan pickers show both sides. Stardust led with yearly and hid the other plans in a "More Pricing" drawer, and it lost on conversion and proceeds per user.
Rash ID went the other way, from two plans to three with yearly marked as the best offer, and saw about +20% ARPU and over +50% conversion to paid.
A labeled recommendation helped more than a shorter list.
Guiding choice instead of cutting it
If overload comes from people not knowing what they want, the fix can be to help them choose rather than to remove products.
ConvertPolo added a short scent quiz to Theater's perfume category page and got +24% conversion rate and +20.44% revenue per visitor.
More examples are in our product quiz tests.
The same thinking applies to pricing tables: a highlighted default or "most popular" plan reduces effort without taking options away.
See the tests on our pricing plans page.
How to test it on your own site
- Check the four conditions first. Overload is most likely when options are hard to compare, the category is unfamiliar, and visitors want a fast answer. If your buyers are experts who know the exact SKU, a big range is more likely to help.
- Hide before you delete. Collapse secondary options, lead with best sellers, or add a recommendation. Removing products or plans outright cuts off buyers who needed them.
- Measure revenue, not clicks. A shorter list can raise conversion while lowering order value, or the reverse. Track revenue per visitor.
- Change one thing if you can. Many option-cutting wins shipped with new copy and layout, so the count may not be what worked.
- Don't stop early. The Conversion Fanatics loser led by 40% at first and finished below control.