Phone House case study: How to generate 3,000 mobile descriptions in seconds

At Narrativa, we often speak about automatic content generation for e-commerce and how this can free up companies from the most repetitive tasks. But how do we do it?

Using this real-life case study, we’ll show you how we can digitally transform companies with our artificial intelligence system.

Would you want to write 3,000 manual descriptions?

The company Phone House, one of the largest mobile phone retail chains in Spain, displays the product characteristics of all their mobile phones on their website to help their customers understand and differentiate between the devices. Up until now, these product descriptions have been written manually.

However, considering that Phone House has a catalog of more than 3,000 mobile phones for sale, creating such a large number of descriptions manually is arduous and time-consuming work. In addition, these descriptions need to be constantly updated when new products or product lines are introduced.

Gabriele gets to work

Companies need the descriptions of their products to integrate SEO best practices in order to score highly in Internet search engine rankings and writing all the descriptions by hand requires a lot of effort. So, what is the solution?

This is where Gabriele comes in. Narrativa’s artificial intelligence system can make this type of work easier for companies by following these three ‘simple’ steps:

1. Phone House provides us with a database of characteristics of all 3,000 phones in its catalog.
2. The original hand-written descriptions give us a body of data which we can use as a basis to train our AI. The NLP (natural language processing) lets us find and extract relevant information such as themes, expressions and writing style.
3. Gabriele elaborates on the descriptions and uses Machine Learning (ML) algorithms to identify the weight of the mobile, the screen size, memory capacity, etc.

This makes the descriptions more natural sounding, providing useful information to the reader without using technical terms. The client generally wants to know whether their phone is large or small or whether they can fit enough photos or applications on it, etc.

Our ‘expert’ Gabriele classifies 3,000 mobile phones

One of the biggest concerns about using a machine rather than a human to generate these descriptions is making sure they sound natural, but actually Gabriele is able to provide information that is even more user-friendly. Our AI does this by identifying characteristics and adding opinions to them, classifying the data in various groups. But how does Gabriele know what a big or small screen is?

Machine Learning (ML) is a classification tool. In this instance it creates three groups in which to classify the more than 3,000 phones and assigns each one an adjective. For example, if the device has a 4-inch screen it will be put into the ‘small’ group, while those with an 8-inch screen will go into the ‘large’ category.

Artificial intelligence performs this classification automatically, using expert-level assessment and adding subjectivity and natural language. All this can be done with training; no need for an actual expert nor do you need to be one yourself.

Let’s take the example of the iPhone 7:

With this Apple phone not only can you take high-quality snapshots with its high-definition camera, but you can also enjoy them on its large screen.

It has a high resolution 12 MP main camera, with an f/1.8 aperture and a video resolution of 4K (3840×2160) pixels with an f/1.8 aperture; It also has a 7 MP front camera, full HD (1920×1080) pixels for video and f/2.2 aperture, perfect for your selfies and video calls. Inside we can find a 4-core Apple microprocessor. The large-format screen measures 5.5-inch (13.97 cm) diagonally, and with full HD (1920×1080) resolution it’s perfect for browsing your favorite content.

Connect to Wi-Fi networks, or 4G (LTE), 3G (UMTS) and 2G (GSM) mobile networks. Use the Bluetooth v4.2 connection to synchronize your mobile with compatible accessories as well as Bluetooth and Lightning-type audio connections. It has space for 2 SIM cards (dual-SIM).

It has been designed with more features such as NFC connectivity, integrated biometric sensor for fingerprints, A-GPS and GLONASS geolocation systems to get you to wherever you need to go, and IP67 certification.

Would you say that this was written by a machine? Would a machine refer to ‘your favorite content’? Well it did, and it was. Let’s take another example of this Samsung:

Do you want a high-performance phone? Well you won’t want to miss this Samsung, featuring a high-definition camera and large-format screen. You’re going to fall in love with it.

It has a high resolution 12 MP main camera, with an f/1.7 aperture and a video resolution of 4K (3840×2160) pixels with an f/1.7 aperture; it also has a 5 MP front camera with f/1.7 aperture for brilliant self-portraits. Inside we can find an 8-core processor developed by Samsung. Ready for multimedia content with a wide screen measuring 5.5 inches (13.97 cm) on the diagonal, and a resolution of QHD (2560×1440) for superior viewing. Its 3600mAh capacity battery, which you can charge with its wireless charging technology, a Micro USB or Qi.

Connect your terminal using Bluetooth and Stereo 3.5mm audio connections, and Bluetooth v4.2 which lets you connect your device to headphones. You can stay connected all day thanks to its compatibility with 4G (LTE), 3G (UMTS) and 2G (GSM) mobile networks and, of course, WiFi. It has space for two SIM cards (dual-SIM).

It has more features such as NFC connectivity, a biometric fingerprint sensor which you can use to unlock your phone, A-GPS, BeiDou (BDS), GLONASS and positioning GPS systems so you don’t get lost and it is dust- and water-resistant up to IP68 certificate standards. It weighs 157 grams, not bad for a device with such a large screen size. The device dimensions are: 15.09cm high, 7.26cm wide and 0.77cm thick.

Are you sure you will fall in love? Yes, Gabriele really wrote this.

Gabriele is not alone

Do not forget that an automation editor supervises and generates the text so that everything is perfect. It is also important to note that mobile phones are constantly being updated, so new phones will be added in the future and algorithms to update the ratings will be redesigned.

You can read this case study here.