Does variety mean greater innovation for Big Data?

How Big Data enables the relationship between data types to be explored

In my previous article in this series exploring the application of Big Data to marketing, I looked at how the Velocity component of Big Data provides greater understanding of your customers, improving marketing performance. In this post, I will explore the third of the 3 Vs, the potential impact of Variety on Marketing

Variety covers the multiple types of data that are now available from the established structured data of tables and columns, with defined elements and values, to the many unstructured free format types of data.

In recent years the amount of unstructured data and machine generated data is increasing at an exponential rate, being the major proportion (>80%) of the vast volumes of data now in existence.

So what does this variety of data mean for small to medium-sized businesses (SMEs) and does this diversity of information formats drive innovation opportunities for Marketing?

The machine data is structured and the main challenge is the sheer volume and velocity of data as discussed in my earlier articles. With unstructured data there is deluge of new information being provided in non-relation structures containing formats such as video, sound, images, free text, etc, with the volume and number of unstructured data types ever increasing.

Are all unstructured data types used by your business?

Of the core unstructured data types (text, video, image, sound) you probably use all when your business delivers messages to your customers, the key question is how many do your customers use when interacting with you?

In most SMEs this will cover three types of data:

• Textual Data – Textual unstructured data exists in many formats from email messages, facebook, twitter, complaint detail, call centre notes, etc. Although classified as unstructured due to the free format of data entry, the information does follow rules based on language even text language (e.g. LOL = Laugh out Loud). This characteristic of textual data enables meaning and sentiment to be determined from the text.
• Audio Data – As with text, speech follows the same language rules with the addition of timing, tone and volume to provide emphasis and additional meaning.
• Image Data – Static Images provides details of colour, brightness and saturation, enabling images to be classified and tagged with key descriptors (e.g. Google Image Search). This identification allows customer generated images to be used to enhance your message, as well as providing pointers on sentiment of customers.

I have not included video data in this list currently as the majority of interactions are covered by the three types identified, but in coming years this will grow and will provide a further source of information similar to the image data, but providing the added element of sequenced pictures.

Therefore unstructured date has actionable content, it is just harder to access than classically structured data where a defined field contains known values or ranges. For example, a defined field rating your holiday from 1-5 (Poor to Excellent) is easy to understand and interrogate, but a series of Tweets expressing ‘I had a great time on holiday x’ offers similar value once the key information indicators are identified.

How might unstructured data drive innovation for Marketing?

1) Greater connection

In today’s ‘always on' world, customers interact through multiple channels continuously creating a constantly connected marketplace where the ability to capture, store, analyse and drive insight is key to building and maintaining a customer relationship.

This interaction is being driven through social media channels such as Facebook, Twitter, Instagram, Pinterest, etc. The information provided from these channels covers two of the types of data mentioned, namely textual and image data, and can be a key source of information to understand:

1.1 Sentiment - Using sentiment analysis to determine the attitude and emotional level of communication:

• Attitude - positive, negative or neutral.
• Emotional Level – angry, happy or calm.

This identification sentiment can be used in many ways, of which two are:

• Current Pulse - provides a view of current and historic customer’s opinion/sentiment allowing you to better understand and therefore improve engagement with your customers.
• Customer Service – identify key positive and negative interactions, enabling a quick and appropriate response to customers, enhancing positive experiences and neutralising negative ones.

1.2 Consistent Message – Classically, customers are broken into key groups with different offers and messages presented, but the connected nature of customers means these messages/offers will be shared across wider audiences than initially targeted. This will result in the dilution of personalisation & contextualisation and the creation of misconstrued messages, which drives the need to:

• Deliver consistent messages across channels and groups, providing meaning for differing value groups. Help the customer understand the offer criterion. For example, Boden send my wife a £10 voucher for spend over £30 every now and then, but have never sent my mother in-law the same deal. Now this is probably due to spend patterns and recency of last purchase, but from my mother in-law’s viewpoint she is being treated differently and does not know why. Now imagine this over social media with the message shared across multiple customers/prospects and you can see the potential damage this could cause. In actuality, my wife buys products for Mum to take advantage of the offer and hence compounds the cycle of offers.
• Respond to hijacked/misunderstood messages to ensure a simple misinterpretation does not snowball into a major negative impact on your revenues or brand image.

1.3. Product Usage – Machine data will be a major future source of information to drive improvements in products, but this will provide actual usage and not imagined or aspirational usage. With the ability to monitor and determine insight from unstructured data, the opportunity to identify product improvement or alternate usages will increase. This provides the opportunity for the customer to be involved indirectly or directly, through follow up engagement, in the future of companies products/services.

2) Improved messaging

When creating a programme of communications, there are many elements to consider, from who to communicate with, what message to use, how to segment the data, which channels, how to test the communication, etc. With the introduction of Big Data and in particular Big Variety this would seem to add to an ever growing list of things to think about and take action on. So what does Big Variety add to the time pressured activity of marketing communications?

2.1 Enhanced Testing – Access and use of unstructured data will enable additional testing of communications beyond the traditional A/B testing of response/conversion rates to start looking at social buzz caused and spread of communication. Is your message being received positively/negatively is your message spreading?

2.2 Improved Content – As well as providing content to help your customer understand and engage with your product/service, you also want your customers becoming brand advocates spreading the positive message and ideally creating their own positive content. Monitoring of unstructured data will not only identify these advocates, but also customer generated content which could then be used to drive future communications.

From this last in four articles looking at the value to Marketing of Big Data I have looked at how Variety can drive innovation, with opportunities to take advantage of untapped data sources.

This completes my short review of the potential impact of Big Data on marketing and I have explored how the increase of available, ever-growing, ever-changing data can be utilised in the marketing arena. For details of earlier articles in the series please click on the following links:

• An introduction to using Big Data for Marketing.
• Does Volume mean Value for Big Data in Marketing?
• Does Velocity mean increased performance for Big Data?