What Good is Your Suggestion?
Modern websites elicit commentary. Ecommerce websites elicit ratings and reviews. Publishing websites elicit feedback and approval. Social media websites are very similar. The value of feedback is something readily available to the end user, while the value of suggestions can be much more questionable. Your approach to providing suggestions to your users can end up being either a valuable insight or a mistrusted, disregarded waste of effort.
Feedback You Can Judge for Yourself
You can't throw a digital rock without hitting a website soliciting a subjective opinion. Facebook, Amazon, Walmart, Twitter and Netflix all feature ways in which the site or the end user can pick winners and losers. Even Rotten Tomatoes, a site that provides other people's ratings offers the viewer a chance to provide their own rating, sometimes of the reviewers themselves.
We all know that sometimes the ratings are fake, sometimes the reviews are real. Either way, we are all analysts and the numbers are provided in order for us to feel that we can comment on, or find others who we trust to comment on, a product, service or person. Much of the numbers are the result of straight averaging, summation and scoring.
If you go to a popular place for publishing articles, like Medium, there is a way to validate someone's ratings through claps, comments, and backlinks. It does not matter whether an article is curated or reviewed directly. You have the option most of the time when someone rates or reviews something to understand their logic or passion. You just view their comments and you understand why they felt that way.
Suggestions Can Be Friendly
Amazon displays multiple types of suggestions if you are shopping for products. First, you get a set of sponsored products related to the item you are viewing. Each of those items shows an image, rating and price so that you can assess for yourself the validity of the comparison. These suggestions are your basic alternatives for what you might purchase, your basic cross-sell. Then, there is a section that shows either customers who bought the item also bought or customers that viewed the item also viewed. Both of these are a suggestion related to someone else's behavior, not strictly to the product you are viewing.
With some products, a grid shows comparables with a set of details about the product itself and competing products. These include objective information like price coupled with subjective information like other customers' aggregate rating. You can view any reviews and ratings for the particular item, along with any questions and answers related to it.
This kind of data is mostly objective. You can regard it or disregard it, but you are aware of its origins. Yes, someone may write a fake review, but still, often the criticisms are fairly realistic and if not, you can ignore them. Amazon's role as a seller is to try and sell you more of what you want. Amazon also tries to move items related to the incentives received, but they notify the end user that an item is sponsored. In other words, you know that they are paid for placement at times and you can factor this into your purchasing decisions.
Suggestions Can Also Be Questionable
I remember as a kid riding in a car in San Francisco and passing a bunch of teenage punk fans. They had on the denim clothing with the colored hair spiked like mohawks or whatever set them apart. Then, in 1983, Nicolas Cage and Deborah Foreman burned up the screen in Valley Girl and they meet to a tune from the Payolas. I had no idea what "payola" even meant.
Payola is a term for the illegal payments record companies made to radio stations to play the music they wanted. It started in the 1950s, but there have been prosecutions and settlements as recently as 2007. The suggestions in this context have nothing to do with reality. You are exposed to recommendations that are there to promote the commercial interests of a product or service that is unrelated to what you are seeking.
If you watch a commercial television network, you see recommendations to watch shows that are made by that same network. You know they own the network and you know they want you to keep watching. The relationship there is clear.
Netflix removed the ability of viewers to rate shows and movies with stars in April of 2017. Prior to that time, you were able to offer your rating of a show using stars. Netflix replaced the old system with a new thumbs up or down option, coupled with a matching system which supposedly tells you the odds you will like a movie.
I perused their matching system and found that there is almost no correlation to my viewing tastes. I also found that there are some very curious cases where Netflix appears to be pushing the content on me where it is obvious from my viewing patterns that there are very low odds of interest. Some of the formulations are very curious.
For example, there is a section called "Recently Added" which has a sub-section entitled "Popular on Netflix". Every single title in that set shows a match of 89-98% for me. IMDB has the movie Mute holding a 5.4 rating. Yet on Netflix, it indicates a 95% match. I don't always subscribe to the opinion of others for my movie tastes, but still, I like to have some idea of how well-received a film is before I make a decision whether to view it.
I am not implying that Netflix is breaking any laws, but I am absolutely implying that their suggestion system could very well be used to promote any type of property for reasons that are not transparent. I don't trust Netflix in terms of their matching system because it is not transparent and does not seem to match at all in a way that it would if it understood my habits. With Amazon, their system may not be perfect, but at least they are transparent that they have sponsors and I get to see what the other shoppers did or did not do.
Suggestions are the output of analytics. You persist data then you query it and tell it to give up insights that are not obvious. Or, in some cases, you are using analytics that provides the highest profit.
The analytics you use can benefit your end user or manipulate it, or perhaps both at the same time. A proferred suggestion from a data scientist is similar to a relative who comes over and tells you "you gotta buy this product" or "you gotta see this movie". You start to form an opinion whether the advice is coming from Mr. Miyagi or your crazy uncle Kent. The companies that emulate your crazy uncle Kent can give you poor suggestions for a while, but eventually you are going to disregard their algorithms altogether. As Dennis Miller used to say "of course that's just my opinion. I could be wrong."