Slashdot Mirror
People Are Complete Suckers For Online Reviews (nypost.com)
schwit1 shared an article from the New York Post:
No reviews, no revenue. That's the key takeaway from a new study published in Psychological Science, which finds that if two similar products have the same rating, online shoppers will buy the one with more reviews... "[When] faced with a choice between two low-scoring products, one with many reviews and one with few, the statistics say we should actually go for the product with few reviews, since there's more of a chance it's not really so bad," wrote researcher Derek Powell of Stanford University, lead author of the report. In other words, when there's only a handful of reviews, a few bad ones break the curve and bring down the overall rating. "But participants in our studies did just the opposite: They went for the more popular product, despite the fact that they should've been even more certain it was of low quality," he wrote.
Matt Moog, CEO of PowerReviews, previously conducted a study with Northwestern University [PDF] that drew from an even larger data pool of 400 million consumers, which also found that the more reviews there are of a product, the more likely it is that a customer will purchase that product... He has also found that customers who read reviews often click the bad ones first. "They want to read what's the worst thing people have to say about this," he said... Most online shoppers (97 percent to be exact) say reviews influence their buying decisions, according to Fan & Fuel Digital Marketing Group, which also found that 92 percent of consumers will hesitate to buy something if it has no customer reviews at all.
Matt Moog, CEO of PowerReviews, previously conducted a study with Northwestern University [PDF] that drew from an even larger data pool of 400 million consumers, which also found that the more reviews there are of a product, the more likely it is that a customer will purchase that product... He has also found that customers who read reviews often click the bad ones first. "They want to read what's the worst thing people have to say about this," he said... Most online shoppers (97 percent to be exact) say reviews influence their buying decisions, according to Fan & Fuel Digital Marketing Group, which also found that 92 percent of consumers will hesitate to buy something if it has no customer reviews at all.
I think it's a bit more complex than that. Oftentimes, you can determine from poor reviews exactly what the shortcomings are, and decide if those shortcomings affect your intended use of the product. If a competing product has no reviews, then you have no way of knowing what the shortcomings are.
People who say "sheeple" have about as much sophistication as an AOL user, and in fact are probably actually AOL users.
If one product has 12 reviews and one 45k reviews, the 12 review has 5* and the 45k review has 4* I'd be more likely to buy the 45k review one. It's simply significantly harder to astroturf and bot 45k reviews than 12. Online typically all you have to go on is a shitty picture, product details that are nearly always incomplete and exaggerated, and reviews. The reviews are the least shitty way to not get ripped off. For the record, yes I know that many reviews on most every site are fakes, paid for the review, or bots.
Plus sometimes the reviews are pure comedy gold and are worth reading on their own which can inspire some sales on their own.
If something has 1000 reviews, I know that the bulk of those are likely to be legitimate. If someone has 3 reviews, I have no idea how accurate the reviews are. More reviews = more sample data of user experience, and more data means that "wrong" reviews (reviews that don't reflect the user experience) are obscured. If you were presented with 2 studies where one used a sample of 5 people and the other used a sample of 5000 people, which results would you trust more? Reviews are just a less controlled study.
I always check the bad reviews and only the bad reviews. Good reviews generally aren't helpful, they're either fake and posted by the seller or genuine but not useful. Seeing twenty 5 star reviews that say "it worked great for me" is great, but what I really want to know is how it failed for people and if I care about those failures.
If all the bad reviews are by people who are clearly crazy or doing something stupid, I can be fairly confident in the product. If they instead reveal significant flaws, I may want to reconsider.
As always, XKCD has a relevant cartoon about this.
At one point in the 1970s it was my misfortune to be employed as a Filter Queen vacuum cleaner salesman (regrets, I've had a few). Anyways at the time I was taught that some people "have to be told to buy" something, even if it's from a perfect stranger. They can't make the decision on their own as most people are told what to do their whole lives. That and "Advertising is the best way to sell something, especially if it's no fucking good".
Exactly. Furthermore there is a problem with the methodology:
This bias was so strong that they often favored the more-reviewed phone case even when both of the options had low ratings, effectively choosing the product that was, in statistical terms, more likely to be low quality.
That is incorrect.
Say you have product A and product B, each with a score of 3 stars out of 5.
Product A has 6 reviews, 3 of which are 1-star and the other 3 are 5-star - the average is obviously 3.
Product A has 600 reviews, 3 of which are 1-star, 594 reviews are 3-star and the other 3 are 5-star - the average is obviously 3, again.
Based on that information alone, you can NOT determine which product is more likely to be bad. You could however determine which product is more popular - but that's it.
Furthermore, if both products have a low score, it is actually better to go for the more popular one simply because more people bought it therefore more troubleshooting, more info about how to overcome its shortcomings and more workarounds would be available online for that product. Shortly put, it would be more likely to find someone else who had a similar problem with the product and posted a solution online.
...gis sdrawkcab (usually not responding to ACs; don't bother posting as AC)
And I suppose it's my turn to be the guy that posts the related xkcd :
https://www.xkcd.com/937/