How To Use Ratings Data to Find Product Trends on Amazon

Available Data

It’s not simple to estimate how much a product sells on Amazon. Unlike AliExpress, unfortunately Amazon does not disclose the number of orders for the products available on their marketplace.

Still, we can try to extrapolate the popularity of a product using the number of ratings it receives.

Let’s assume that a product with 5000 ratings sells more than a product with 1000 ratings.

But what if you want to dig further? What if you want to know sales trend of a product? Whether it’s sales are going up or down? And how it compares with other products in its category?

Well, roughly 0.5% to 1% of sales result in a review. So we can use the number of reviews a product receives as a proxy for how much it is selling.

Extrapolate Sales Trend

And the frequency with which a product receives reviews can give us an indicative sales trend.
We scraped online reviews of a few of the most popular wireless earbuds. And here’s a plot of the number of reviews received by these products per month:

Product Review Count vs. Month

The above plot was made using about 5,000 most recent reviews of the following products: Apple Watch 6, Apple Watch SE, Fitbit Versa 2, Fitbit Versa 3, and Samsung Galaxy Active 2. We used each review’s publishing date to count the number of reviews the products received per month. The dataset used is available for download at the bottom of this post.

The above plot uses about 5,000 most recent reviews of the following products: Apple, Bose, Samsung, Jabra, and Powerbeats. We used each review’s publishing date to count the number of reviews the products received per month.

What We Learn

We can make the following observations using this plot:

  1. Smartwatch sales are on the rise. The curve is going upwards even for the oldest product, i.e. Samsung Galaxy Active 2.
  2. The launch of newer products affects their sales trend. Sales for Fitbit Versa 2 seem to have been affected by Fitbit Versa 3.
  3. The more expensive Apple Watch 6 is more popular than the Apple Watch SE. They both were available first on 15th September, 2020.

Let’s take a couple of more examples of monthly review count vs. time. This time for popular body wash products ranging between $10 and $20. Here are links to each of the products featured: Tea Tree, CeraVe, Dove, Everyone, Puracy and Remedy.

Product Review Count vs Month

Here’s another plot with review counts for the following toilet paper products: Charmin, Cottonelle, Quilted, 7th Gen and Presto.

From the above plot, we learn that:

  • There is a larger pattern here which is likely dependent on promotions and ad spend on Amazon
  • Toilet paper sales were likely higher than average for a year after the Covid 19 pandemic began to spread.
  • From the difference in the review counts pre month, we learn that online sales for these FMCG goods is likely lower than sales for consumer electronics.

Further Possibilities

We’ve only scratched the surface. With reviews and ratings data, one can learn a lot about brands and their audiences:

  • Sales trend
  • Customer sentiment
  • Competitive performance

If you suspect that some products might have a lot of fake reviews, you can choose to only plot the 1-3 star reviews since they might be a truer indicator of sale volume.

Extracting Relevant Data

How can you get ratings & review data for a category of your interest? We used our online reviews API to generate the datasets used for the above analysis.

You can get ratings and review data by scraping the web since this data is public.

Unwrangle’s Online Reviews API scrapes e-commerce data feeds from platforms like Amazon, Walmart, Costco, eBay and more.

Along with reviews, our e-commerce Data Feeds also include pricing information and other useful meta data.

Resources & Sample Datasets

We would like to share the resources we used to write this blog post. Here’s a link to the script used to plot the above graphs. It generates the review count vs. month plot using the response files generated by using our Online Reviews API.

The online reviews datasets used by us are available for download at the bottom of this page.

Lastly, if you want to find trends on Amazon and other e-commerce sites to help your brand succeed. Sign up for a free trial of Unwrangle and extract up to 10,000 online reviews for free. Use the script linked above to learn about the trends of different products.

We also have a new API product that can extract each reviewer’s past purchases, contact us to learn how to use it (documentation is coming soon). You can learn a lot about your target audience using public data. And it’s our mission to make this public data accessible.

Download Dataset

Hotel Reviews Dataset from Yelp

Free hotel reviews dataset with 20k+ customer reviews for 5 star hotels in Las Vegas sourced from Yelp.com. Download link is at the bottom of this page.

Unwrangle’s Review Extractor API now supports scraping English customer reviews from Yelp.com.

Collecting review data from Yelp is now as simple as POSTing the page’s URL to our servers and waiting for the scraped reviews to be posted on your webhook URL. Learn more about how you can easily scrape customer reviews and create your own tailored dataset on our docs page.

The sample dataset can be found at the bottom of this post. The dataset consists of more than 20k reviews. Starting from the first few reviews, including all reviews up until 22nd April, 2021.

For all you data scientists and research enthusiasts out there, we hope that this freshly collected dataset could give you a little kickstart with your research project.

The fields available in this reviews dataset are shown in the screenshot below.

dataset schema
Sample Review Object Schema

This dataset can be used for the following applications and more:

  1. Analyzing trends

    Just as an example, you can see estimate how room occupancy must have been affected by the Covid 19 pandemic.

  2. Sentiment Analysis / Opinion Mining

    Using NLP techniques one can find out what the average user’s sentiment is towards each of the featured hotels in this dataset.

  3. Topic / Aspect Extraction

    Using categorization techniques one can quickly figure out how each of the hotels featured in this dataset fairs on attributes such as room quality, staff, food, check-in process, etc.

  4. Competitor Analysis

    If you would like to find out what customers think about your competitors, a tailored dataset like the one featured in this blog post can enable you to do so with simple data analysis or visualization techniques.

Use the link below to download our free hotel reviews dataset for your research*

Download Dataset

If you want a sneak-peak, a truncated version (with 500 reviews) of the dataset is also available on Kaggle.

If you’re interested in generating tailored datasets like the one featured in this post on-demand with an API call, you can book a meeting with me here.

*Please note that the download link takes a few seconds to load and requires an email address where the dataset will be sent immediately. Your email address would be stored in Paddle.com’s servers (who is GDPR compliant). We will only send you a single email introducing you to our service at max and will not SPAM you at all.

7 Tips to Help You Respond to Negative Customer Reviews

“I will never shop there again! I’ve waited more than 5 weeks for my delivery and the customer service is totally incompetent!”

Unhappy customer

True or not: If your company receives online reviews like these, it’s not only embarrassing. One or two stars on Yelp, Tripadvisor or Google put customers off in the long term. They are bad for your business’ public image. If these comments pile up, it will even affect your bottom line at some point.

How can I react correctly to a negative review – or should I even have it deleted immediately?

Beware that the Internet is an important, maybe even THE most important acquisition opportunity for you. So good reviews from satisfied customers are a great advertisement for you and your company. Still, a bad review is not the end of the world. You yourself decide through your response how customer-oriented and professional your perception is. Bad customer experiences can even help your company to achieve an authentic external image by signaling a high level of service readiness through your fast and solution-oriented reaction. Everyone makes mistakes – and honestly admitting them is a pleasant surprise for the customer.

Incidentally, having a few bad reviews give a company more credibility than having only positive ones. Because solely positive comments could give the impression that the owner himself wrote or commissioned comments that were praiseworthy.

So refrain from wanting to prematurely delete bad reviews. A really angry customer then finds other channels to vent their frustration out. And he might even share his negative experience through these channels. This then becomes embarrassing for the external reputation of your company. And it could also scare off potential new customers.

The following tips can help you answer:

1) Fast response


Reacting to a 3 week old review seems very unprofessional. To prevent this, you have to instruct employees to check rating portals and social media channels for new ratings. Every. Single. Day. Or, much more efficient and reliable – use reputation management software like Unwrangle for being updated on your online reviews.

2) Take customer criticism seriously


First of all, assume that your customer is not a “bad” person and does not want to harm you. He only describes his personal perception and bad experience. In fact, a customer who is taking time out to write honest feedback is doing you a favor. By keeping this in mind, you can prevent impulsive responses and remain composed.

3) Always stay friendly


Answer factually. And always be friendly. This will show your customers that you care about them and are responsive to their feedback. An unfriendly response would be perceived as abusive and could permanently damage your company’s reputation. A good start would be to say thank you very much for pointing this out. Always remember that your comment is publicly visible to all other potential customers.

4) Short and sweet


Never write long justifications as to why everything went exactly the way it did on that day. Because that would just sound like excuses. Your customers will feel like they are not being taken seriously. And if your answer is too long, most users will not even read it. Instead, you should redirect the dissatisfied buyer in a friendly manner to private communication channels. This way the external effect would be less. Please note that your answer can still be visible to everyone – if the extremely angry customer should post personal mails online.

5) Offer a solution to the problem


The aim should be to fix the problem. Perhaps you cannot follow the assessment yourself or the effort to rectify it is not appropriate. Nevertheless, it is important to approach your customers and sympathize with them. If treated fairly, the annoyed person sitting on the other side could be more accommodating with your company.

6) No mass processing with standard reply


Nothing is more annoying than seeing the same response under each negative rating. You do not want to weaken your image with the copy & paste method. Cookie cutter replies make customers feel like their criticism is not being taken seriously. Giving potential customers the impression that the company doesn’t care about them. It is better, however, to use personalized formulations and respond to each bad evaluation individually.

7) Admit mistakes, apologize for inconvenience


According to a study by economists from the Universities of Bonn and Nottingham, 45% of all disgruntled customers lifted their criticism after a personal apology. If there were errors in the process, you must of course admit them. But nothing more. If you otherwise disagree as a company, you should describe it in an equally friendly manner.

A rating example from a disappointed visitor:

“The food was really bad. How can you call yourself a cook when you can’t even use spices? Everything there just tasted totally bland, never again!! “

Another dissatisfied visitor

So here we have a dissatisfied restaurant guest describing a specific service. It is important that you do not want to give the impression that your cook would prepare food without spices in your restaurant. So one possible reply would be:

“We are really sorry that you were not satisfied with our performance. Thank you very much for the specific points of criticism. We will immediately look into it and improve our offer. We would be happy if you gave our kitchen a second chance. It is important for us to convince you of the quality of our dishes. Our chefs are happy to prepare your order according to your wishes. To do this, give the waitress a brief note about your spice preferences. So that we can take your instructions or allergies into account during preparation. “

Responsible manager

If you run a practice and a patient gives unspecific and destructive feedback to your practice – you respond in a friendly manner and respond to the patient’s needs. Also try to direct the further contact to a private channel. One possible answer would be:

“We are very sorry that you are disappointed in us. We would like to find a solution and improve. So we would be happy if you could tell us what exactly bothered you during your visit. Please use our contact form on the website. Thank you very much in advance. “

A patient practitioner

As you can see, it’s not that difficult to adequately respond to the comments of dissatisfied customers. The most important thing here is to keep your own composure and of course: to keep an appropriately short reaction time for your response.

Why Customer Satisfaction Is More Important Than Ever Before?

Customer satisfaction is a contruct in which the subjective expectations of the customer are compared with their actual experience after making a purchase.

What are the factors that influence customer satisfaction and what measures can a company take to adapt its offering to better match the needs of the customer. To find that out, we go on a search for traces on the big issue of customer satisfaction.

Every customer has individual expectations, wishes and demands for their purchase as well as for their provider. Let’s call the sum of these expectations ‘the target state’. The customer subconsciously compares this with the actual state of the experience, i.e. the comprehensive performance of the product provider as perceived by them. There are different ways this comparison can turn out. If the actual and target states are close, the customer is usually satisfied. On the other hand if they are far apart, the customer is either disappointed with the experience or in the better case: they get an unexpected WOW experience.

If everything meets our expectations, we are satisfied. When the experience exceeds our expectations however, the possibe effect is much greater. Such a pleasantly surprised customer will buy from us again and again and even recommend our product to their acquaintances. Every business should aim to exceed their buyer’s expectations. This will lead to customer loyalty, or better still, creation of brand evangelists. Trying to deliver a wow customer experience goes a long way in building a following for your brand.

customer satisfaction wow experience

Never before has high customer satisfaction been more important than in today’s digital age. Where every potential guest or buyer can get an idea of the expected performance before visiting your store even once.

Did you know that 90% of all customers consult online reviews to make a purchase decision before buying a product?

A very satisfied customer will likely come back to your business and become a regular customer. Repeat customers are demonstrably more likely to pay higher prices, make their shopping carts fuller and even shop spontaneously.

In the best of cases, some customers publicly express their satisfaction on rating portals. There is no advertising as authentic and unadulterated as such reviews for a company. These free endorsements are more effective than paid ads, and do not deduct anything from your advertisement budget.

A dissatisfied customer could migrate to competition as an unvoiced complainer. In worse cases, however, he shares his dissatisfaction before that as mouth to mouth propaganda with other potential customers. In this way, our dissatisfied customer makes himself heard about his frustration, about the subjectively inappropriate product behavior or a lack of its quality.

The worst case for us as an entrepreneur would be for their bad experience to become a widely shared review and grab attention of prospects. This is where quick and flexible action counts; this is the only way to reduce the negative perception created from the documented bad experience.

A personalised response to a dissatisfied customer’s review conveys that your company cares about improving its customer experience. This course of action will have a positive effect on buying decisions of potential customers and hence also on the financial wellbeing of your company.