[Idea Forge] Honest Ratings from Verified Buyers
Conventional online reviews are fraught with peril, but there's a better way.
Have you ever been to a restaurant or hired a contractor whose service wasn’t as good as expected? It happens with 4.5-star products on Amazon too. I like to check reviews and online ratings before major purchases, but I’ve still been surprised plenty of times.
Something isn’t right with a large chunk of online ratings and reviews. Item purchases aren’t always verified; bad ratings are removed when merchants report them to platform administrators; and sometimes paid career reviewers offer touching fictional stories from multiple fake accounts about products and services they never bought.
Unreliable ratings are more than a minor inconvenience. Along with search ranking manipulation, they are the result of an industry that is still in its infant stages. Without accurate reviews and trustworthy certifications, consumers can’t find products and services that solve their problems effectively at a reasonable price.
I am reminded of the time Amazon took down my 1-star review of a pouch of Essiac tea about a year ago, which I learned after purchasing contains a large amount of toxic oxalic acid that needs to be reduced by roasting or other means before consumption. That unusual tea is still being sold to people with health issues, so it’s a serious matter to ensure that buyers have reliable information.
How many paid, inauthentic reviews are promoted within product listing pages on major e-commerce sites and business listing directories? Google, Amazon, Yelp, and others may have behind-the-scenes mechanisms to protect rating integrity, but they would be more convincing if they were to show high-value, high-quality items first, not just the items that come from the largest consolidated brands.
Reviews and ratings would be more useful in a new generation of social commerce with slightly more advanced features:
Large sample size (i.e. high participation rate)
Micropayments for reviews
Utility class filtering
Verified purchases
Cross-item aggregation
Aged reviews
Up, down, neutral
Clear graphic display
Large sample size (i.e. high participation rate)
In order for reviews to represent the crowd’s true opinion of a product, the reviews cannot come from a biased segment of the purchaser pool. In scientific studies and polls, a sample is considered representative of the population being studied if it is sufficiently large. Ratings also become more accurate when a large percentage of a merchant’s patrons pitch in to provide their honest opinion.
For example, if you want to know the average color hue of marine iguanas on the Galapagos Islands, you could take a sample by photographing lizards in the shadows of the dark volcanic rocks near the forest’s edge and you would see that their color might be predominantly dark grey. Try again along the water’s edge near the colorful reefs and you might find iguanas with red and green coloration. A study that only samples from one ecological niche would be biased toward a certain color pattern.
Micropayments for reviews
People regularly pay for news, plus Consumer Reports and expert opinions about big-ticket items like houses and cars. Shoppers benefit greatly from knowing if their peers like a product. Without knowing which product to buy, I might end up buying three different water filtering systems before I know which one actually works for my needs. Accurate ratings would save me the lost time and expense of driving back and forth to the store or going to the shipping office to make returns.
With advancements in online payment technology, a market platform could pay buyers a few cents or fractions of a cent for their opinion about a recently purchased product. This would boost the review participation rate to a threshold where it constitutes a representative sample. These incentive rewards are more effective when offered by the market platform rather than the merchant to keep a level playing field.
Utility class filtering
If I buy a soda to use it for an experiment in cleaning off rusty tools, it’s a lot different than buying soda pop to drink. Gross. Let’s use a different example: vinegar for cleaning vs. using on a salad. The quality of the two items are probably very different, and I still might upvote them both times. Or I buy balsamic vinegar for salad, realize it’s actually just white vinegar, and give it a downvote. But with a more advanced review system, I could change my utility class from “eating” to “cleaning” and give it a neutral vote. Shoppers could also browse reviews by utility class to find the product that rates best for their particular needs.
Verified purchases
In order for reviews and ratings to be authentic, they must come from actual buyers. If a purchase is made outside a platform, the seller and buyer could independently report date, time, items, and purchase volumes. If both transactions match, it might be considered a verified transaction, but the most verifiable way to be sure the transaction happened as reported is to host the transaction within the platform providing the ratings or to collect data directly from the payment processor.
Purchase size weighting
A buyer’s large purchase should be weighted more heavily than a smaller purchase in the aggregate rating. For example, if a restaurant regularly buys tomatoes from a local farm and all goes well, they can offer their upvotes. If the restaurant needs an emergency tomato that isn’t usually on the delivery schedule, the tomato might be bad. But because it was a smaller purchase, the rating should not be weighted as heavily as the regular tomato deliveries.
Cross-item aggregation
Buyers can leave one rating per transaction to save time, but these ratings can be separated out based on the seller and the item reviewed. If many reviewers buy similar items from different sellers, those related ratings can be shown in aggregate for each item. Higher general ratings for an item than a seller-specific item rating would indicate that the seller has a worse version of the item or offers lower-quality service.
Aged reviews
Useful lifetime varies by product. A house lasts a lot longer than a sheet of paper, so it’s easier to rate a pack of paper a week after purchase than it is to rate a house. Reviews made longer after a purchase generally contain more information than more recent reviews because the customer has had more time to test out the product or service and find its potential flaws or perks. Micropayment incentives can be sent at different lengths of time after a purchase to collect more diverse data to present to users.
Up, down, neutral
Ebay has a relatively good, simple system with transparent data about previous reviews. It’s easy to decide if something was good, bad, or neither, whereas five-star ratings take a bit more time to think about. Those few seconds it takes for someone to decide between four stars and five stars could make or break the review because buyers might completely skip the rating if it takes too long. People who do not leave a rating should be treated equivalently to someone who leaves an explicit neutral review.
Clear graphic display
YouTube shows a clear graph of upvotes, no-votes, and downvotes to give viewers an idea of how appealing a video might be. A marketplace with high review participation rates could emphasize the positive review percentage out of total purchase volume. Many platforms ignore the importance of neutral reviews, but they matter. Assuming there’s no viewership manipulation, a YouTube video with 20% of viewers upvoting is statistically likely to be more satisfying than a video with 1% of viewers upvoting, even if the ratio of upvotes to downvotes is much lower for the first video. Sorting algorithms can more effectively use upvote percentages when participation rates are relatively constant.
I’m excited for online shopping to be more certain, less confusing, and more decentralized. As online marketplaces become ever larger, we could use a few basic tools to make buying choices easier. Small businesses would have the opportunity to show their true quality with authentic ratings systems.
In the local economy, I would be glad to buy more sweet potatoes from local vendors when I know with certainty that they’re offering a tastier, healthier product than their grocery store competitors. As their sales grow, they can probably afford to lower their prices a bit too.
Graphic icon by Freepik via Flaticon.