On Saturday, Scott Brave of The E-Commerce Times posted a great article about seven deadly consumer biases that impact customer ratings and reviews on etailer websites.
- Personal bias. When using a small sample to predict what the population as a whole will do, personal bias is a real problem.
- Squeaky wheel bias. When it comes to product reviews, there are certain groups of people who are most vocal for various reasons. Usually, these are the people who have had negative experiences.
Contextual bias. People use products for different purposes and their feedback on those products can be distorted by this.
The E-Commerce Times uses the example of two different customers reviewing a digital camera. One says the “resolution is incredible” and another says that the “resolution sucks.”
- Emotional bias. The emotional state of a person can have a dramatic impact on the feedback that is provided at any given moment.
- Gaming bias. Feedback is often provided by people who are “gaming” the system. While this is sometimes done maliciously, sometimes it’s done with the best of intentions (i.e. you are an employee of the company that makes the product being reviewed).
- Time delay bias. Products and services change for better or worse and feedback rarely shows this as well as it should.
- Multiplier bias. New feedback is often influenced by existing feedback, creating a “herd mentality” effect.
So how can we deal with these biases? Brave suggests that we look at the behavior of the “silent majority” by collecting data about the customers who will never bother to leave explicitly feedback.
I agree wholeheartedly with this “analytics” approach.
At the same time, I think there are ways to deal with these biases by implementing feedback systems that take them into account. Here are some simple suggestions:
- Personal bias. Expand the number of feedback sources customers have access to by pulling in feedback data from third-parties (or by linking out to feedback available through third-parties). Professional reviews can also be useful here too.
Squeaky wheel bias. This is a tough one to deal with. A friend of mine who runs a small network of ecommerce websites offers a potential solution: he emails customers a month after their purchase and reminds them to leave feedback.
His emails encourage those who have had good experiences to let other customers know about it. He tells me that once he implemented these emails, the number of feedback left increased notably, as did the number of positive reviews.
- Contextual bias. Ask your customers to provide contextual details with their reviews. This could be as simple as having your customers classify themselves (i.e. “Power User” vs. “Casual User“) or by having them detail how many products of this kind they’ve owned in the past (i.e. if somebody has owned 5 digital cameras, other customers will likely assume that he is more demanding).
- Emotional bias. This is another tough bias to deal with. Reminding customers that their reviews may have an impact on other customers’ decisions can help ensure that they understand the importance of providing an objective review.
- Gaming bias. While there’s no realistic way to eliminate reviews from those with agendas, I like what NewEgg.com does – it tells you when a reviewer actually purchased the product through NewEgg.com. I personally trust these reviews more since I know the person actually purchased the product in the first place.
- Time delay bias. Show customers how feedback has “trended” by breaking reviews down into definable periods (i.e. this product was rated 3 out of 5 by 100 people all-time but was rated 4.5 out of 5 by 10 people in the past month).
- Multiplier bias. Try encouraging feedback on terms that help mitigate the “herd mentality.” If you email customers asking them to provide feedback and direct them to a page on which existing feedback is not visible, you may help blunt this bias.
There is no doubt that feedback systems are a must-have for etailers today but making sure that the feedback is collected in a thoughtful manner that addresses bias is equally important.