There is so much noise and confusion with big data I thought I would look at the types of data available in an organisation.

One of the big challenges I see is people talking about the CMO outspending the CIO dating back to 2012.

But is this really the case? In Graham Ruddick’s recent article he argues the CIO and the CMO battleground is a fallacy: 

From the conversations I have had with CMOs from very large automotive businesses through to online grocery and even a pub group, is the CMO just wants the technology to work where data is an integral part, but it is less concerned about the how and more about how they do it.

The Questions I DO NOT Get Asked Are: 

• What technology platform should I use?
• What scale out architecture could we take advantage of?
• How can clustering or multivariate analysis help profile customers better?

However We Regularly Talk About How Data Can...

• Help me please answer the questions of my business

OR

• Help me better understand my audience, so I can reach, convert and retain customers, in the most cost effective way.

To bring this to light I took a couple of photos whilst out and about in London.

The first one I saw was this HP advert on the London Underground, which was eye catching but as a marketer didn’t really tell me anything!

I saw this IBM advert on the South Bank in London, which was fun, but again didn’t really help me understand how big data would make my life easier.

And The Final One Was From Bluesheep Group On Social Media:

I wonder who the target audience really is? If it is the CMO who spends more on IT than the CIO then I think I might have missed the mark!

I am no copywriter but I am surprised, given the target audience this is not aimed more at CMOs not CIOs. Why wouldn’t they have lines such as:

• “We know more about your customers that you do”
• “We can help you reach your audience like never before”
• “Price, product, place, promotion, people, tick! What’s next?”

Goals

Ideally I really want to answer those questions which the business needs answering but just doesn’t have the answers or information available. Sometimes this information is readily available, but not within the business. So what do I need to do next?

Types of Data

Before going through a protracted research process of defining big data I thought I would look at the data, which is used by an organisation, and try to classify it.

First Classification

The first classification is about ownership of the data itself.

Owned:

The first type of data is managed and owned by the business. It is data that the business does not have to pay an external party to provide. There is certainly an overhead for maintaining and accessing this data, but the business does not licence the data.

This information is valuable as it is the truth about the business. It is not what customers feel about the organisation but what they actually do.

This Would Include: 

• All transactional information,
• Prospects who have opted in for communications,
• Web analytics data,
• Arguably data insights from their own Facebook, twitter and other social channels
• Call centre data such as as call volume and call reasons
• Retail in store data such as footfall information
• FAQ databases around the business
• Customer reviews and feedback
• Survey information if the surveys were delivered by the organisation

External:

The second type of data is owned by a third party and accessible by the organisation through a free/public access or a paid agreement.

This information can be hugely important as it gives the organisation visibility into areas where they might have lacked insight.

Some of The Sources Might Include: 

• Home mover data
• Flood data
• Footfall data
• Weather information
• Search traffic on a given moment of time
• Trending topics on social media
• Competitor performance looking at website visits and media spend/advertising activity
• Crime information
• UK Health data
• Vehicle MOT failure information
• Vehicle licensing information

Now some of this data will be relevant whilst others will be totally irrelevant to some businesses.

Second Classification

The second classification is something I have written about before in the “single customer view is dead long live the single individual view” but it still plays a strong part in defining the types of data.

Static

This data is the historical information, which has a very low frequency of change, ideally less than once per week, if at all.

This historical data provides the business an accurate picture of the way the business is trending, whilst building an accurate picture of customer value and behaviour.

However this data cannot be kept up to date with the vast amount of data, which is being generated through the interactive and digital channels of today, such as website traffic or social information. Furthermore this data will not be able to accurately predict when a customer will come back to your website, call the call centre and their need or reason.

Examples of This Type of Data Include: 

• Name
• Gender
• Transaction date, time, type, value
• Address
• Email address
• Mobile number
• Segment name or value they are linked to
• Historical communications
• Customer service interactions

Dynamic:

This data is continually changing. The nature of this data means it is very difficult to be able to consume and action this data using traditional methods purely down to the volume, but also due to the expectation from consumers. Imagine processing this data to provide customers a relevant message by passing it through the existing data warehouse. We all know website response times are critical so a 1 second response is too slow let alone a 1 minute processing time.
 

Although this data might have little or no historical information it does provide an understanding of the consumer need at the time of the interaction. For example what their needs are now. What they are looking to achieve now. We do not need to ask questions like what channel are you on? when do you want to do this? and why would you like to do it? As these questions are already answered for us by the consumer implicitly.

The Type of Information Generated By Dynamic Data Includes: 

• Phone call and reason for call
• Last click on the website
• Referral from the search engine
• Click on the advertising banner
• Conversation at the till
• Live chat conversation
• Mobile app download
• Form completion online
• Social media comment

Visualising the Types of Data
In the diagram below I have highlighted how these two different data classifications might be captured within a business.

What can I do with this, now I understand what data I have in my business? 

So now the data has been mapped it is possible to start to answer the challenging business questions.

These are questions which the business asks, but sometimes is unsure of the answer, as the data is unavailable.

As an example an automotive company might want to answer these questions: 

• What impact has website outages had on my business?
• Has the recent recall affected my business sales?
• Does weather have a positive or negative impact on my business?
• Do in market competitor offers affect conversion?
• Do MOT failures affect customer loyalty or present an opportunity for acquisition?
• Do areas with good transport links affect customer purchases?
• What affect have warranty claims had on customer loyalty for servicing their car with us?

Some of this data will be in the business whilst others might not be available. These questions above are certainly a challenge when we look at “Owned” and “Static” data only, the typical sources for an analytics team within a business.

Business Question 

“Do MOT failures affect customer loyalty or present an opportunity for acquisition?”

Firstly we need to identify the data the business needs to do the analysis (with the items in red currently not within the business).

Owned and Static:

• Service bookings and value
• New car purchases
• Website test drive bookings
• Website service bookings
• Extended Warranty claims
• Car breakdowns

Owned and Dynamic:

• Website visits
• Website conversion rates
• Website service booking and test drive page visits
• Call centre volume and call reason
• Referral data from, MOT advice and car evaluation websites

External and Static:

• Car trade-ins
• Car finance transactions
• MOT failure data
• Non owned warranty claims
• Non owned car breakdowns

External and Dynamic:

• Social media chatter on failed MOT
• Search volume on MOT
• Competitor traffic

Example: Extracting External Data

Now we have identified above the data we need to determine which information will help answer this question from the External sources. The MOT failure data will be used in this example. Using a simple snapshot below with data compiled by Honest John and data provided by http://data.gov.uk/. To simplify the process, I have selected the top 3 sample of the data to highlight areas of the UK and makes & models of cars with the lowest MOT pass rate. This has been filtered by their first MOT, with car being at 4 years old. Ideally we would take other external data to add further insight and context to our research.


This illustrates the potential audience who might be in the market for a replacement car by region and make & model. A simple acquisition campaign might be a local door drop or list rental communication to non customers. Or a retention or renewal mailing to existing customers.

By combining this with trade in data by region, by trade in vehicle (or by capturing the car the customer changed from in the post purchase survey) we can start to build up a picture of customer loyalty or acquisition depending on which manufacturer is doing the analysis.

An initial trial acquisition or retention campaign (depending on the manufacturer) might be in Dundee or customers who own a Chrysler Grand Voyager, 300C or Peugeot 807.

Defining the Metrics

Next we have to identify the metrics that are important for our analysis. The metrics are car sales (acquisition) and service bookings (retention). These two metrics are clear and allow us to accurately understand the effect the other sources of data have had on sales or service bookings.

Building the Picture

We can now analyse the data using our analytical team to build a picture on the relationship between the different data sources with our goals of increasing sales or service bookings.

This will show the correlation or likelihood each source has had on an improvement in sales and service bookings.

This allows us to make decisions on which sources of data were closely linked to service bookings or car sales.

Finally this will then enable the organisation to build out a marketing plan and the purchasing of media/licencing of external data to try and reach these audiences and validate more accurately what has and has not had an affect on sales or service bookings.

Summary 

With so much data out there is it becoming harder and harder to actually know how to take action on the information available.

By doing the following we can try to answer some of those complex business questions: 

• Identify the business questions in the organisation which need answering
• Confirm the types of data within the organisation
• Select the data which is needed to answer the question
• Use the metrics to help steer the analysis in the right direction
• Perform the analysis to identify the relationship between the data sources and the key metrics
• Take action on the insights to validate the data hypothesis 

Original Article

Read More on Digital Doughnut