Unlocking the value of SMS data with machine learning

Here at Chatterbox Labs we excel at monetizing real-time and historical short text datasets with our machine learning technology. These datasets may be real-time feeds from social networks such as Twitter, Facebook, Sina Weibo or Tencent WeChat, but also SMS. Over 350 billion text messages are sent per month worldwide and on top of being a chat technology they are used as an effective customer engagement mechanism. For a full run down of the tech please visit our main site at http://www.chatterbox.co.

Our statistical machine learning and natural language processing technology learns the slang & niche language often found within a text message, across multiple languages. As a business model, our technology is licensed to global technology partners who use our NLP Framework to build new classification products in days, and our Path 2 Purchase product to pinpoint buyers in noisy data.

Security & Privacy

Chatterbox technology is always deployed inside a partner firewall. This is important because no information or data is sent across the public web to our servers - it all stays inside the partner firewall. This is critical both for latency and privacy concerns.

Use Cases

Partners that integrate our Path 2 Purchase technology into their software stack often use this for driving sales, targeted advertising and audience segmentation by pinpointing in-moment buyers. As the technology is embedded into their offering and data centers, the action taken is entirely up to them and bespoke to their businesses. Those partners that also take our NLP Framework are able to rapidly build new classification products in days without needing an army of data scientists.

Path 2 Purchase. Path 2 Purchase can analyse customer SMS interactions with a brand in real-time to pinpoint in-moment buyers, allowing the brand to respond in the moment with an appropriate action. The actions will be bespoke to each brand, however they may consider the following:

  1. Pre Purchase - Interest. If we detect that the user is very early on their path to purchase - and there will be a lot of these as this is the top of the funnel - we can send them a link to our e-commerce store to move them down the purchase path whilst engaging at scale.
  2. Pre Purchase - Consideration. If we detect that the author of the message is mid journey - in the consideration phase where they may be comparing products, or unsure of the purchase - we can send them a link to some information on our site, perhaps with some information about the pros and cons of our offering vs the competitors.
  3. Pre Purchase - Lead Validation (Buy). If we detect that the customer is ready to buy a product, we'll notify them that a sales rep will be calling them shortly.

It's important to note that these actions are, of course, mock. In live scenarios the execution mechanism is bespoke to the partner organisation.

New NLP Products

Spam Filtering. Using the NLP framework, new classification products can be created very quickly. At the cellular network level messages being sent to the network's customers can be screened for spam or adult content, ensuring that this does not reach the end customer.

Interest Based Segmentation. Using the NLP framework, classifiers relevant to your segmentation needs can be created. For example, our fictious cellular network provider offers a product to their customers that gives them access to discount deals including retail purchases, event tickets etc. The network doesn't want to bombard their customers however, they want to ensure that those interested in Music get the deals for gig tickets and those interested in Technology get the discount laptop deals. The custom classifiers can run in real-time or on batch datasets to segment the audience based on their actual behaviours. The decision of what segmentation and classification to build with the framework depends upon your business goals - the opportunities are endless!

For more information, please contact Andrew Watson, VP Strategic Alliances andrew@chatterbox.co or visit our web site www.chatterbox.co