I heard from a reputable source that on e-commerce websites, users find it easier to choose a product when there is an odd number of items in row, as compared to an even number.

In their experience, 3 or 5 products tested better than 4, for example. Sure enough - I can find plenty of sites with 3 or 5 products per row, but I can also find plenty of sites with an even number too.

So given that everything else remains the same:

  • What actual evidence is there for an optimal number of products per row being related to the odd/even nature?

  • How/why does this affect the ease with which customers can make a decision?

enter image description here

Examples Charles Tyrwhitt Ted Baker John Lewis

  • 1
    Amazon don't seem to worried about it, searching harry potter in clothes gives 4 columns, film & tv gives 3 columns, books gives 6 columns. Admittedly my kudos is much reduced by searching for harry potter rather than nice shirts.
    – icc97
    Commented Feb 1, 2013 at 13:24
  • 1
    This sounds like a use case for A,B,...N testing.
    – zzzzBov
    Commented Feb 1, 2013 at 17:52

3 Answers 3


An example of this claim can be found in Smashing UX Design: Foundations for Designing Online User Experiences - page 300, point 2. It says that an odd number of products:

  • can be easier to scan,
  • makes the choice easier to make,
  • keeps the eye moving across a row because the items cannot be grouped easily.

It suggests that this might be backed up by the "rule of odds" (see on Wikipedia and here, and on photoSE). The claim is that an odd number of objects is more interesting and natural, or even more pleasing and attractive.

I think the problem might also be connected with the issue of 5-star ratings (from 1 to 5). Some claim that people tend to choose the neutral, middle option because it's easier and therefore there should only be an even number of stars to force a more meaningful choice (others disagree). Perhaps you do want to make the choice easier on an e-commerce site.

  • Good answer - although...I was hoping to find at least another piece of evidence, or a reference to research or test results, because it just happens to be one of the authors of this book that mentioned this to me in the first place! Commented Feb 1, 2013 at 23:54
  • Just bought the book!
    – Tom
    Commented Aug 14, 2013 at 19:00

Here's my take on your second question. Long story short, it is naturally simpler to "divide & conquer" an odd group:

In an odd grouping, the center item likely becomes the natural focal point, and so you'll get a mental image of that item in your mind as you look at it. Next, your eye will likely gravitate to one of the side groups to compare your mental image against items from that grouping. If you like an item from that grouping better, then your mental image is replaced with that item. If you don't, your mental image will be refreshed when you glance back to compare it against the items from the other group. When you get to the other side, you've already developed a mental strategy for comparing groups of that size, so you reuse or improve upon your strategy for comparing items in that group, which lessens your mental burden. All in all, the biggest sub-group size you had to consider during this process was (N-1)/2, or for example 1, 2, and 3 items per group for 3, 5, and 7 items total respectively.

In an even grouping, there is no clear center (except whitespace), so the left or right item will probably become your focal point, likely depending on your reading language. Again, you will get a mental picture of that image in your mind and then compare it to the remaining items. At this point, the two obvious choices are a linear or a grouped comparison. If you linearly compare items one-by-one, there will likely be a lot of back and forth eye-movement to refresh the mental picture of the item you like and then return to the item you are considering, and so your eyes and brain will get an extra workout. If you comparing items as a group, though the remaining group is odd, it will be difficult to compare it in the same way as you did before. This is because you already have one extra thing you are thinking about (the first item), and you will also not have the visual symmetry in your groups, because the first item will throw off that balance and be a distraction. Either way, you will lose the ability to re-use/improve upon your mental strategy for comparing groups. And last but not least, your maximum sub-group size will be (N-1) instead of (N-1)/2, so you will be dealing with larger than necessary sub-groups.


This sounds like an unordered list problem, not a visual focal starting point concern. Regardless of where the eye naturally falls at first (possibly the center, is there research to show this behavior for list evaluation?), the user is going to evaluate items one at a time in either rows or columns, based on the proximity of items or other visual cues directing them. Even or odd, they will start top-left (or top-right) and scan.

Evaluation is largely based on the Uncertainty Principle/Hick's Law and there was some interesting research I've read by Yaliang Chuang and Lin-Lin Chen titled "How to Rate 100 Visual Stimuli Efficiently" that may be enlightening as well. Another interesting article on Smashing Magazine about Redefining Hick's Law touches on some of the broader issues of simplifying the decision making process for users.

For the example you gave, color strikes me as something that the user would key off of the most, so maybe the visual focal point is directed by that or some other Visual Variable and overrides any tendency for any divide and conquer tactics? I couldn't find any research to back up the tendency of users to evaluate a list of options more easily with odd numbers of columns or rows either. Research titled "Rating: How Difficult is It?" by Shilad Sen and E. Isaac Sparling reviews the various rating systems (unary, binary, 5 star, and slider) and how the number of options affects speed and satisfaction, but that seems only loosely connected to your question.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.