The mobile phone is the drill to extract the data

Last week I wrote a blog post entitled “If data is the new oil, we need a bigger drill“, where I complained that we weren’t making enough use of the potential data available to us.

That post was in relation to online research. But on reflection, the opportunity is far greater elsewhere.

On the mobile phone.

Introduction

And, as far as I am aware, it is an area even more underexploited than online data capture. Aside from the odd application (such as Everyday Lives – which looked very similar to Evernote last time I looked), mobile survey panels (such as One Point) or academic experiment (Contextphone in Helsinki), I’m not aware of any innovations in mobile.

Which is a shame, since it is arguably the most powerful media platform for data capture. The Wikipedia page on the 7th mass media lists the eight unique benefits mobile has. Of most relevance are that the mobile is

  • Always on
  • Always (well, usually) carried on the person
  • Available at the point of creative inspiration
  • Highly personal, and personalised

The unique benefits of mobile make it an ideal instrument for both active and passive data capture – for explicit answers and for implicit inferences.

Forms of data capture

I’ve drawn an arrow below of the five primary means a mobile can capture information. It is very much an early draft, so feedback or criticism is very much appreciated.


Ways in which data and information can be extracted from the mobile phone

Background capture

As mobile technology advances, devices incorporate more features that produce information on the location of the phone, and thus the user. These include:

  • Time – the time itself, and the time it takes to do something via a clock and stopwatch
  • Date – via a calendar
  • Space – via GPS
  • Proximity – to people, objects or events via GPS, bluetooth or RFID chips
  • Movement – in three dimensions via an accelerometer, or inferred through GPS and clock
  • Environmental factors – through thermometers, altitude readers and so forth

In addition to location, the following can also be determined through past or current behaviour:

  • Spend – via the in-built payment mechanism
  • Social graph – via the address book
  • History – via cookies or memory
  • Broad character traits – by how the phone has been customised or used

While in future, these will be augmented with innovations such as voice and face recognition (a Google Goggles type of service).

Either on their own or in combination, these features facilitate some extremely powerful data capture. They effectively allow us to understand the “where” and “when”, and potentially the “with”.

But it is only the first level of information capture.

Activity Capture / Activity Follow-up / Prompted Activity

I’ve grouped these three stages together, as they are essentially variations on a theme.

The mobile phone has a large number of features and services that can be used for data capture. These include

  • Voice calls
  • Text messaging
  • Voice recorder
  • Note taker
  • Calendar
  • Bluetooth
  • Games
  • Camera/scanner
  • Video camera/editor
  • Music player
  • Web browser
  • Email/social network use
  • Application downloads/use
  • Shopping/purchasing

The most passive form of data capture is in recording the functions that a person uses their phone for. Forms of analysis this facilitates include

  • Combining activities with the data dimensions outlined in the first section for understanding of individual uses
  • Utilising path analysis across all feature uses to understand how the phone is used as a single device, rather than as a collection of services
  • Converting phone calls to text, and then using sentiment analysis to infer meaning across all forms of communication.

These aspects augment the “where”, “when” and “with” with the “what” and “how” – at least in terms of mobile phone behaviour.

A slightly more active form of data capture would move closer to capturing the “why”.

For instance, a push notification could be triggered when a certain activity is undertaken. This could request a simple answer to a question.

For instance, if I were to use my camera to take a picture, I would know

  • Where it was taken
  • When it was taken
  • What it was taken with
  • How it was taken (landscape or portrait, flash or natural, first attempt or fifth)
  • Potentially who/what it was taken of
  • Potentially who the person was with at the time

But it wouldn’t be known why the photo was taken, or whether the person was happy with the photo taken. A simple question or two would solve that.

An even more active version of data capture would be to explicitly ask the person to use their phone for a particular person. For instance, they could be asked to use the camera to scan each item they buy on the high street or to use the voice recorder or note taker each time they spot a certain advertising campaign. These methods are used by a couple of organisations – MESH spring to mind – but have little noticeable traction to date.

This manual mechanism may eventually be superseded, as technology allows us to automate more of the data capture. Its only real relevance would be in forcing someone to participate in a behaviour where they naturally wouldn’t.

Direct questioning

As should be obvious, the more explicit forms of data capture are those that are most prevalent – primarily because their implementation is independent of technological advancement. For instance, we’ve always been able to interview people over the phone. As technology improves, the interfaces underpinning this method will also improve – we will move from SMS surveys to java to html to html5 or native applications, with touch screen drag and drop functionality.

Benefits

As I mentioned in my previous post, we aren’t close to reaching the level of data capture that is possible. We need to augment explicit questioning with the context that can be inferred from the situational data collected. The mobile phone, moving across space and time and with its unique benefits, offers even more scope for collecting meaningful data.

Potential uses for the data capture include

  • Calculating sleep quality/efficiency (an iPhone app already does this, to a degree)
  • Monitoring movement, speed and proximity of people across an environment could be used for town planning
  • Alternatively, it could be used to plot the efficiency of layouts in supermarkets. If the phone could calculate eye line (it would probably need to be attached to a necklace), it could even inform how the shelves are stacked
  • Providing an understanding of people’s lives – when using and not using their phones.
  • Exploring how things spread across mobile phones. For instance, one person could undertake a type of behaviour, come into contact with someone else, and then the second person undertakes the behaviour. Network effects could be used to identify the mythical influencers
  • Tracking spend can be used for financial management
  • A networked calendar/diary could become predictive e.g. rescheduling a meeting to take place 15 minutes later due to traffic
  • Tracking movement can improve the measurement of exposure to outdoor advertising
  • Sound recognition could be used for radio or TV exposure, and improve out-of-home consumption measurements
  • Inefficiencies of usage could be explored e.g. the time it takes to connect a phone call can be compared across devices and networks

Practical obstacles

Evidently, the previous section was quite speculative and fantastical, but I hope it underlines the potential. Nevertheless, several obstacles need to be overcome before this point is reached

  • What is the best way to collect such information? Within the operating system? The network/SIM? Via the web or an application? The O/S with control of the API would appear to be best placed, but do they have the inclination?
  • Although phones are always on and regularly used, they are also regularly upgraded. Information collected would need to be portable for long-term tracking
  • Similarly, a phone is more susceptible to breakage, theft or loss
  • Background data capture would be a tremendous drain on the battery
  • Effective data capture would require an entire network of people using it – this is highly unlikely, not least because there will always be a significant proportion of people for whom a mobile phone will just be a device to make and receive emergency calls
  • More behaviour will be transferred to the mobile, but it will only ever capture a small proportion of our lives
  • Coverage and connectivity isn’t good enough (in the UK) for full capture – unless information can be stored natively before it is uploaded to a central server
  • Massive issues of data protection and privacy. Some people (such as Nicholas Felton) would enjoy tracking their movements, but I suspect – outside of paid-for testing – few would appreciate it. Particularly since the mobile is the most personal of devices Imagine if large corporations were able to track the movements and social graph of its employees through mobile phone usage?

Conclusion

This post seems to have been sidetracked into future gazing, but my underlying point remains. The technology is available for us to capture far more information – and thus understanding – then we currently do. Organisations should look to harness and utilise this data, to provide contextual meaning to what people are doing.

Thoughts on how we could do this – or on how people are already doing this – would be much appreciated

sk

Image credits: http://www.flickr.com/photos/kioan/3011984637/ and http://www.flickr.com/photos/_parrish_/2575256484/

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: