Improving at its “Job to be done”

Aditi Priya
4 min readFeb 15, 2017


I am a reader. I read anything from long novels to short stories to business articles to the news. I am a reader, Product Manager, and a recent Kindle convert — I now carry my library with me at all times, and there has been a sudden spike in my activity on Goodreads along with my new-found Kindle.

And as much as I like Goodreads, I know it can serve me much better!

Please bear with me for the next few paragraphs and I promise you a good conclusion.

A consumer, I use Goodreads to find my next book, most likely a novel on which I am likely to spend at least a week on, and I would want the most accurate recommendation; something that will suite my pallet at the point of time.

How do I go about this book search?

The obvious answer is to go to the Recommendations button hidden deep under the “Browse” dropdown, and I can see many recommendations based on my choice of genres. I am sure Goodreads would have gotten some great insights based on my reading pattern and used sophisticated data analysis algorithms to predict my next choices. Assuming one of the recommended books is what I am looking for, how do I now go ahead and select the best among the multitude — problem of the plenty at play!

The answer again seems very obvious — I read the ratings of the book, and if it is good and I want confirm the hypothesis that the ratings are correct, then I spend some time going through the reviews by other readers.

To illustrate better, I would pick up the instance of one of my favorite books — Atlas Shrugged by Ayn Rand. I am just using this as an example, and I was not recommended this via Goodreads but my friend.

The rating of the book is 3.68 — good? Not good enough? I can’t say. I do not know who gave the ratings and what they are based on. I have been on the rating end of the product side for too long, and I know how we rate products — sometimes it may appeal to me but not to others. Sometimes my ratings are influenced by my friends, and other times a whole host of further factors, but I know that effectively, ratings convey nothing to me.

So I dismiss the whole rating system now, and scroll down to read the reviews.

And voila! I get two very different reviews!

Person A says it made his eyes hurt, while person B simply seems to love it!

Now who do I trust and why?

So now I have reached the conclusion part of the story. Can we make the recommendation system better and more useful to the readers of Goodreads? I think we can.

So, here are my ideas to create a better recommendation engine:

1. Goodreads selects best reviews for users!

Allow the users to choose whether they want to read all the reviews, or read the reviews of people whose profile is similar to theirs. I am sure Goodreads would already have real good algorithms and data analytics tools to figure out similar interest readers based on factors like — genre interests, past reading history and many other parameters.

To illustrate better, below is what I am talking about.

2. Users make their choices of review

This option is to guide users to select their own reviews that they might want to refer to, and others they want to reject — so whose opinion to consider? Person A or B?

Again, to illustrate better, below is my perceived design!

It is now easier for me to correlate my expectations with the person who has written the review and choose the appropriate one.

Every book reader relates with this problem and would love and appreciate a better recommendation engine!



Aditi Priya

Product Management @ServiceNow | Talk about Products, AI, and more | Read more @