Understand how your store score is generated
Your Company Score Understanding how we calculate the ResellerRatings Score ResellerRatings Team
Dig in more to understand the difference between the RSR Score, the average score, and other details on the store page. The RSR Score goal is to give you an idea of what a current purchase experience might be like, while also taking into consideration the lifetime of reviews in a collection
What the ResellerRatings Score IS and IS NOT
The ResellerRatings score on a page is our attempt to quantify consumer sentiment into a scoring system that's easy to look at and understand quickly. It is also representative of a score in which any reasonable consumer can see and say to themselves 'this should be my current experience if I were to buy from this store today'.
The ResellerRatings Score is
IT IS a specific rating curve based on review count, review recency, and weighted buckets
The score takes into account all reviews across the lifetime of a store, however, a review that is within the first 6 months affects the score more than reviews that are older. The more recent a review is, the more weight it holds.
We have a couple of goals for the RSR Score.
We want it to represent consumer sentiment for their current experience with a retailer (if they were to make a purchase today).
It is based on consumer sentiment
It is fair to retailers who go through various dips or increases in their business operations
Track retailer histories without punishing them for resolved issues or rewarding them for the outdated positive feedback
Eliminate troughs and peaks, ensuring a smooth and easy-to-understand reflection of all reviews
The algorithm we use at ResellerRatings is designed to give consumers a simplified view of retailer ratings based on consumer opinions. The rating aggregate is not a reflection of a straight average (i.e., number of reviews/score), but rather a snapshot that gives more weight to recent reviews. While all active reviews are considered, reviews posted within the last year influence the score more.
The Score is shown on a sellers store page
We show several scores on the store page. Within finer details of the store page, a customer can dig much, much deeper into the scoring system to understand review trends much, much better.
RSR score being the primary score.
The distribution of the scores across the lifetime of the store (you can dig deeper on detailed ratings for more information here)
Detailed benchmark ratings for particular services (shipping, returns, recommendation, would you come back, etc) which are averages on the last 6 month of scoring data
Detailed historical data on the benchmarks themselves
We believe all this data provides a cumulative and holistic view of ratings and scores over time.
How the Score is Calculated
Let's talk about how we calculate our score. The scoring system in ResellerRatings takes several things into account.
time of when the review was written
review counts for a particular time period
the actual rating for overall customer satisfaction
Each star rating is worth a certain number of points. This adds up to a score out of 5.
Review Aging and Value
Reviews age over time so a client's score is always changing regardless if they are collecting reviews or not. For example, a bad review that ages over time may have less impact on a score as it ages. The same goes for a good review. In time, with no collection - all reviews ultimately are considered for the final score with generally even distribution and weighting.
We capture the concept of aging by introducing 'buckets'. Reviews are split into 5 buckets representing date periods of last 30 days, 30 to 90 days, 90-180 days, 180 to 365 days, and older than 1 year. Starting from the first bucket, "last 30 days", we attempt to reach a minimum threshold of reviews before calculating, should the number of reviews in that bucket be below the threshold we'll mix the next bucket with the current bucket to ensure proper consideration is met with review counts. For example, this prevents 1 review that is recent, from completely overpowering 20 reviews that are perhaps aging into bucket 2. The buckets must always meet these minimum counts otherwise they are merged together until they meet this.
Once a threshold has been reached the average rating for the reviews in that bucket is calculated. Once all buckets have had their average calculated those ratings will be combined using a point-weight system. The "last 30 days" bucket for example is worth a possible 60 points. Any buckets that were merged have their weightings added together. Any buckets without any reviews do not contribute to the overall total possible points.
the rating calculation effectively works as: (bucket average * bucket weight) divided by total possible weight for all buckets counted
All active reviews are taken into account during the scoring process, but ultimately reviews past a one year date range are weighted considerably less in the algorithm if there are reviews from customers within the past year.
This ultimately smooths out the curve and ensures a lowering of jumps and score trenches and peaks.
How often is the algorithm changed?
We study the changes and take feedback very seriously. Algorithms must always adapt to how people collect reviews, try to game systems, etc. We rarely change our algorithm, but when we do, we announce incoming changes ahead of time and update accordingly. We have an upcoming algorithm change slated for q4 of 2020, more about it can be read on the ResellerRatings changes blog.
Historical Changes
August 24th 2020 - Change of Rating scale from out of 10 to 5
May 2020 - Change of detailed rating scale and historical ratings from 10 to 5
Did this answer your question?