← Back to the principles list

Confirmation bias in ecommerce

Next principle →

People look hardest for proof of what they already hope is true.

A shopper who hopes a moisturizer is gentle on her rosacea scans the reviews for exactly that. When a review in her own words is easy to find, she can feel sure enough to buy the bigger jar.

Confirmation bias
Made-up example: a review filter lets a hopeful shopper find the answer she came for.

Confirmation bias, sometimes called myside bias, is the tendency to look for, read and remember information in a way that supports what you already believe or hope.

Help shoppers find honest answers to the question they came with, using review filters and plain product facts. Keep bad reviews in those results, because a filter that hides complaints only fools people. Store owners have this bias too, so check your favorite ideas against real orders.

Confirmation bias example: finding a rosacea review for a moisturizer

Here is a made-up skincare store. A shopper with rosacea hopes its soothing moisturizer will be gentle on her skin, and she can buy a trial size at $14 or a full jar at $38. With reviews sorted newest first, the top one says “Smells lovely,” which tells her nothing, so most shoppers take the trial size and each order is worth $14. The store adds a “Rosacea skin” filter. The top review now says “Gentle on my rosacea,” her own words back, and most shoppers pick the full jar, so each order is worth $38. Let shoppers find the answer they came for, and keep the complaints in that filter too.

How confirmation bias works with other principles: 4 store examples

Confirmation bias catches store owners as well as shoppers. Three pairs below separate it from similar ideas and one shows how it keeps a favorite tactic alive. Every store below is made up to show the idea.

Confirmation bias vs hindsight bias

Often confused with: Confirmation bias filters evidence to fit a current belief, while hindsight bias rewrites the past so the result looks as if you saw it coming.

Illustrative store example: A toy store owner believes bright banners work, so she quotes only the weeks when one did well, and that is confirmation bias. When a plain page then wins and she says she always knew it would, that is hindsight bias, so write your guess down before each change.

Hindsight bias: why results look obvious once you know them →

Confirmation bias vs cognitive dissonance

Often confused with: Cognitive dissonance is an uneasy feeling after a choice, while confirmation bias is seeking proof you were right, which can ease that feeling but also happens without it.

Illustrative store example: A home cinema shopper, already set on a $1,500 projector, skims the specs only for points in its favor, which is confirmation bias with no discomfort at all. After paying she feels uneasy, which is cognitive dissonance, so send a setup guide and honest tips that give her real facts instead of more glowing reviews.

Cognitive dissonance: why buyers feel uneasy right after paying →

Confirmation bias vs survivorship bias

Often confused with: Confirmation bias favors the evidence you like, while survivorship bias leaves out the people who never reached your data.

Illustrative store example: A phone case store counts only repeat customers and calls its new design a hit, because shoppers who disliked it never came back, and that is survivorship bias. If the owner then ignores the few unhappy emails she did get, that is confirmation bias, so read those first.

Survivorship bias: why you only hear from shoppers who stayed →

Confirmation bias and the law of the instrument

Works with: The law of the instrument is reaching for a favorite tactic every time, and confirmation bias can make you notice only the proof that it works.

Illustrative store example: A pet supply store owner loves popups and points to the sale one brought in last week. She misses the shoppers who closed the tab because of it, so test the page with and without the popup before adding another.

Law of the instrument: why a favorite tactic gets overused →

Put it into practice

How do I set up review filters that really help?

Read your support chats and list what shoppers ask before buying: skin type, height, dog breed. Ask reviewers for those details when they write a review, then turn the answers into filters with plain names like “Rosacea skin.” Show every rating inside a filter, low ones included. The filter should answer the shopper's question, whichever way the answer goes.

Should I hide bad reviews to keep shoppers hopeful?

No. Hiding them is dishonest, it can land a store in legal trouble in some countries and it sends the wrong people to checkout. Someone who hopes a cream is gentle needs to read about the person it stung. If the complaints are minor, like slow delivery, they can even settle her nerves. If they aren't minor, you have a product problem to fix.

Can confirmation bias work against my store?

Yes. A shopper who already suspects a store is shady will go looking for proof, and small things qualify: a blurry photo, no contact address, shipping that appears only at the end. One find can settle it for them. The same bias that helps a hopeful shopper hurts you with a wary one, so fix those small signs first.

How is confirmation bias different from the halo effect?

The halo effect starts with an impression: one good thing makes other things seem good too. Confirmation bias starts with a belief or hope, and then the person hunts for evidence that fits. They can run one after the other. A tidy homepage gives a shopper a good feeling about the store, and then she goes looking for reviews that back that feeling up.

How do I check whether a review filter helped without fooling myself?

You probably want the filter to work, so decide before you start what counts as a win and what would show it failed. Turn the filter on for half your shoppers and leave it off for the rest. Compare what they pick, then check returns a few weeks later. Read the whole result, including the numbers you dislike, or ask someone with no stake in it to read it first.