The serial position effect is the tendency to remember the first and last items in a list better than the ones in the middle.
The early-item advantage is called the primacy effect and the late-item advantage the recency effect.
Its close relative, the Von Restorff effect, says that one item that looks different from the rest is remembered better than its look-alike neighbors.
All three come from memory research, where they hold up well.
Whether they help on a website is a separate question, and our test database gives a mixed but useful answer: putting the right thing first and making one thing stand out often wins, but only when the item deserves the spot and nothing else competes for it.
This page covers the research first, then what real A/B tests show, then how to test position and distinctiveness on your own pages.
For the wider family of mental shortcuts, see our cognitive biases guide.
Where the serial position effect comes from
The classic curve comes from Bennet Murdock's 1962 free-recall study, "The serial position effect of free recall" in the Journal of Experimental Psychology.
People heard word lists of different lengths and wrote down every word they could remember, in any order.
Recall was high for the first few words, dipped through the middle, and rose sharply for the last few.
Plotted, it looks like a U with a tall right arm.
Murray Glanzer and Anita Cunitz then showed the two ends behave differently.
In their 1966 paper, a slower presentation rate improved recall of early items but not late ones, while a delay before recall removed the recency advantage: after 30 seconds of counting, the end of the curve flattened.
Their reading was that early items get rehearsed into long-term memory, while the last items are still sitting in short-term memory when the test begins.
That second finding matters for web design. Recency in the lab is fragile.
A visitor who reads your pricing table, scrolls past reviews, and opens a new tab is not in a lab recalling a list 2 seconds after hearing it.
The Von Restorff effect: one thing that stands out
The Von Restorff effect, also called the isolation effect, is named after Hedwig von Restorff, whose 1933 paper in Psychologische Forschung showed that an item which differs from the rest of a list (a number among syllables, for example) is recalled better than the same item placed among similar ones.
The detail designers usually miss is that the effect depends on the other items being alike.
Robert Hunt's review "The subtlety of distinctiveness: What von Restorff really did" (1995) argues that von Restorff's own account put weight on what happened to the non-isolated items, not just on the isolated one.
Distinctiveness is relative. If three things on a page are bright and badged, none of them is isolated anymore.
What our A/B tests show about standing out
The cleanest example in the database is a badge test from Dave Diederen.
A "Bestseller" badge on 3-4 products on a collection page lost: conversion rate fell 4.72%.
The team then put the badge on only the single #1 product, and that version won with +6.03% conversion rate and +8.89% revenue per visitor.
Same badge, same page. The only change was that one product was now different from its neighbors.
Contrast on the main action works the same way.
Conversion Fanatics ran a test on a black and grey apparel store where the black add button blended into the page; switching the add button and path-to-purchase elements to a bright pink accent raised conversion rate 57%.
The case study gives no sample size, so treat the size of the lift with caution.
The pattern behind it is the point: the button was the one thing that no longer matched.
Our button color page explains why the specific color matters much less than the contrast.
Navigation can be isolated too. Rightpoint visually highlighted the mobile hamburger menu for High Mowing Organic Seeds and reported +15% mobile conversion rate and +22% add-to-cart rate over 4 weeks.
It does not always work. Clean Commit added 'New' and 'Best Seller' badges over the main product image and got -0.5% conversion rate and -9.1% revenue per visitor on 20,113 visitors, a flat result with a 46.9% probability to beat control.
On a product page the visitor has already picked the product, so a badge that helps them choose between products has little left to do.
Primacy and recency effects on web pages
In memory research, primacy and recency are about recall.
On a website, the version that matters is about choice: which option people pick, and which section they actually see.
There is real research on order effects in choice, outside the lab word list.
- Ballots. Joshua Koppell and Jennifer Steen studied the 1998 New York City Democratic primary, where candidate order rotated by precinct. In their 2004 paper in The Journal of Politics, candidates got a larger share of votes when listed first in 71 of 79 contests, and in 7 of those 71 the first-position advantage was larger than the winner's margin (summary at JSTOR Daily).
- Tasting in sequence. Antonia Mantonakis and colleagues gave people 2 to 5 samples of what was actually the same wine and asked them to pick a favorite. In "Order in Choice" (Psychological Science, 2009) the first wine had a large advantage. Participants who knew a lot about wine showed a recency effect in the longer sequences instead.
So the research supports a first-position advantage in choice when people are unsure, and shows that recency can appear for experts or longer sequences.
It does not say "put your best thing last" as a general rule.
What our A/B tests show about going first
YSL Beauty moved the 'Exclusive Offers' tab ahead of 'New Products' in the homepage menu.
The test was proposed by AB Tasty's consulting team and ran for a month on over 150,000 visitors: +9% basket page visits, +13% checkout visits, and +10% transactions.
Elevate A/B Testing listed the biggest pack first in the bundle selector for an underwear brand, and revenue per visitor rose from €3.54 to €3.84 (+8.47%).
This one is not pure primacy. Showing the most expensive option first also sets a reference price, which is anchoring at work, so the two effects can't be separated in this test.
The first slot only helps if it holds what people want.
Dave Diederen's team saw on a heatmap that visitors ignored the "Best Sellers" tab placed first on a homepage and kept clicking one specific category.
They moved that category to the first tab and removed Best Sellers: +14.17% conversion rate and +14.40% revenue per visitor.
The removal was part of the change too, so this is a test of "best first slot," not of position alone.
What our tests show about the end of the page
Recency is the weaker half online. One Conversion Fanatics test is a useful warning.
An urgency message about an intro price, shown both above and below the product customization form, won with +41% conversions.
A second variation in the same test that put the message only right above the order button, at the end, lost: control converted 32% better.
Neither result comes with a sample size, but the direction is clear enough.
A message that appears only at the last step can arrive too late or feel like pressure, not a reminder.
@DaveDiederen@DaveDiederen
@elevate_ab
Conversion FanaticsResearch findings vs A/B test evidence
| Claim | What research shows | What A/B tests show |
|---|---|---|
| First items are remembered better | Strong and long replicated in free recall (Murdock 1962; Glanzer and Cunitz 1966) | Moving the right item first often wins, but most tests change more than position |
| Last items are remembered better | Yes, but the effect disappears after a short delay | Weak; an end-only urgency message lost to control |
| First position wins choices | Ballot and tasting studies find a first-position advantage | Supported in menus, tabs, and bundle order, often mixed with anchoring |
| A distinct item gets noticed | Yes, when the other items are alike (Hunt 1995) | One badge beat four; contrast buttons and highlighted menus won; badges on a single-product page were flat |
The lab effects are among the sturdier findings in memory research.
The weak link is the jump from "remembered a word" to "bought a product."
Treat these effects as reasons to test, not as a guaranteed lift.
How to test position and distinctiveness
- Pick the slot by data, not by habit. Rank menu items, tabs, and collection order by clicks or revenue per session, then test putting the leader first.
- Isolate one thing per view. One badge, one accent color, one highlighted plan. If everything is emphasized, the effect cancels out, which also adds clutter.
- Change one variable when you can. If you reorder and remove at the same time, you learn that the bundle worked, not why.
- Measure revenue, not just clicks. A highlighted option can pull clicks from a better option. Track revenue per visitor and the mix of what sells.
- Don't rely on the end of the page. Repeat key information near the top and next to the decision, as the winning urgency variation did.
- Size the test properly. Order effects are often a few percent. Check your sample size before you call a winner.
Related effects
Order and emphasis rarely act alone. The decoy effect uses a deliberately weak option to make a highlighted one look better, the peak-end rule is about how people remember an experience rather than a list, and the paradox of choice explains why shorter lists leave less of a middle to get lost in.
For the rest of the topic area, start at the psychology hub.