Good or Bad ? Predicting Sales of a New Product

A lot of consumer-oriented companies nowadays tend to rely on one or two flagship products. For example, one of the first things that will come to mind when you think about Apple would be the iPhone. Sales of such products are decisive for the success of the company.

With the rise of Big Data Analytics, capturing crowd opinion has been a lot easier. Comments on social media and forums, articles from specialized blogs, consumer reviews and ratings from E-commerce websites… the list of data sources that are now available to get feedbacks and evaluations on any product is pretty vast. If one has the expertise to extract relevant information ahead of time and analyze it with powerful Artificial Intelligence algorithms, it becomes possible to anticipate and estimate the future sales of these products and therefore the impact on the company’s revenues.

Through two use cases in the Tech sector, we will see how QuantCube Technology has managed in the past to deliver profitable trade decisions thanks to sales prediction.

Blackberry Z10

The BlackBerry Z10 was a high-end touchscreen-based smartphone developed by the Canadian manufacturer Blackberry and presented at a launch event on January 30, 2013. This new smartphone was highly anticipated by observers and experts, as it was supposed to embody the new strategy of Blackberry, transitioning from their trademarked trackpad to a touchscreen interface.

In anticipation of the launch event, we developed one of our very first application of our Data Analytics algorithm. Extracting and analyzing comments from Twitter in real-time, we computed a Sentiment Index, which is a score that ranges from 0 to 100 measuring the sentiment expressed by Twitter users on a specific product, brand or company. The closer it is to 100, the more positive about the product the public is (conversely, the closer to 0, the more negative it is).

Before the official launch of the smartphone, the Sentiment Index on the BlackBerry Z10 had a value of 75, indicating that consumers were pretty hyped about what seemed to be a promising new product. During the presentation of the product specifications and by the end of the event, the Sentiment Index rapidly dropped to 22, which triggered the issuance by our systems of a SELL recommendation for the BlackBerry stock. One hour later, BlackBerry shares plunged 15%, reflecting the disappointment of consumers and experts that we had anticipated through our analytics. By issuing our recommendation ahead of time, we managed to make a substantial profit.

Apple Watch

The Apple Watch is a smartwatch developed by Apple and released in April 2015. As it was the first new product to be launched since the death of Steve Jobs, the watch has been much anticipated, meaning that disappointing sales could send a bad signal regarding Apple’s ability to innovate and reach new markets, and directly impact revenues.

Despite the hype surrounding the launch of the product, the first results of our analytics on Social Media comments revealed that there were five times fewer posts on the Apple Watch than for the launch of the iPhone 6 and iPhone 6 Plus six months earlier. Plus, the Sentiment Index on the Apple Watch before the official launch (45) was pretty average, whereas it reached a value of 60 for the iPhone 6. These preliminary analytics showed that this new product seemed to generate less interest from consumers than historical flagship products.

In order to confirm this intuition, we also extracted and analyzed systematically – through Natural Language Processing algorithms – thousands of consumer reviews posted on major e-commerce websites and tech blogs regarding the Apple Watch, after its official release. Similarly, we found evidence from our analytics that consumers were overall not impressed by the watch, indicating therefore that sales could be worse than expected. In anticipation of the Apple’s earnings release that were scheduled to happen on July 21, 2015, we then issued a SELL recommendation for the Apple stock, based on these analytics. By the end of the earnings conference call that day, shares of Apple went down 6.6% in after-hours trade, despite beating profit and revenue consensus. One of the key information from this first earnings release since Apple Watch debut was that the watch sales, as we had successfully anticipated, had got off to a slow start since launching in April, allowing us once again to make some profit out of our analytics.

What comes next?

These two uses cases confirmed that the combination of new data sources and Artificial Intelligence could generate profitable investment decisions and forward-looking insights on product sales for consumer-oriented companies. Possible applications seem endless, as the amount of comments, ratings and evaluations posted online grows bigger and bigger each day. One of our next challenges in the upcoming months will be to predict the success of a new highly-anticipated flagship product released by another tech giant: the Nintendo Switch.