Tag Archives: analytics

What’s so special about the IBM + Twitter announcement?

I confess that when the IBM + Twitter partnership was announced a few weeks ago I wasn’t quite sure what was new. We’ve been talking to clients for a while now about the value of social media data and using Twitter as a data source. But after a call with the IBM lead for said partnership it’s all a lot clearer.

A brief discussion on social media analytics

Many organisations, including IBM, will talk about social media maturity in the context of analysing social media data.  (There’s also maturity in terms of sending and replying but that’s a different subject.)  The starting point is to listen: looking for mentions in social media about brand, competitors, products, and so on. That’s the sort of thing that perhaps something like Hootsuite or even Tweetdeck can be useful for.

Next there’s thinking: analysing the data you captured in listening.  And for some this will be purely understanding sentiment about brand, product and service.  And there are lots of tools out there that can help you with this, although – perhaps unsurprisingly – I believe the IBM set is probably the most advanced, especially when you consider the sophistication of our analytics, and the ability to find insight that is statistically relevant. (If you have time take a look at IBM Social Media Analytics.)

This leads nicely to the last phase of acting/doing: using your thinking to define actions such as changing product or services, or perhaps marketing strategy as a result.   For me it’s the application of advanced analytics technologies – such as Hadoop (IBM BigInsights), predictive analytics, and so on – that uncovers some very interesting insight, and identify necessary actions.   I’ve used a lot of buzzwords there, let me make it real.  So, for example, we worked with one client to help them understand how to grow their food attach rates and coffee sales.  We helped one client understand that to keep their investors happy they had to focus on their R&D mix, not their stock price as they had expected.  Another client was able to increase their cross- and up-sell opportunities by understanding upcoming life events such as marriage, birth and retirement.

So, why IBM + Twitter?

Our technologies have been able to take social media data feeds from Twitter and many other networks, blogs and forums for a while.  In a way there’s nothing entirely new there.

This partnership is different because of what’s available to test our theories out.  That is, not everyone is sure that social media data really can be a useful source of information to them.  Hopefully some of the examples I’ve given suggest to you that it does have a variety of uses that lead to financial benefit – and customer satisfaction and loyalty and so on – but I suspect this blog is rarely enough to convince!  So, IBM will usually start with running a proof of concept (POC) project together with a client, to prove the value of the analysis, likely with the analytics technologies set up as a cloud service.   In this agreement with Twitter IBM has access to the full firehose of Twitter data, there is no limitation on what IBM will get, and it will include new tweets, as well as old ones.  This ensures that IBM can more accurately demonstrate value of the analytics to our clients.  There’s no guessing or caveats about what we found because of a restricted data set, or old data.  When we run such a POC we, of course, leave the insight with the client.  (But not the Twitter data.)

This is the only such agreement that has been made with Twitter and means IBM will also be training up an army* of consultants to be experts on the Twitter platform.

Lastly, Twitter data will be offered in IBM Watson Analytics, the new cognitive service that brings intuitive visualisation and predictive analytics to every business user, and Twitter data will be available to integrate with IBM DataWorks.

If you want to know more the IBM press release is a good place to start.

*10,000 apparently.

Advertisements

2 Comments

Filed under Analytics, Social Business

It was in the keys

Well, Andy Murray going out was a bit of a shock. I’m still quite sad about it. Having said, that, both Andy and Nadal getting knocked out earlier than expected does make it rather exciting too.

When I heard what happened I immediately went to the Wimbledon Slamtracker to see what the keys to the match had to say about it.

The IBM Keys to the Match system – which is part of SlamTracker – runs an analysis of both competitors’ historical head-to-head match-ups, as well as statistics against comparable player styles.  This allows it to determine what the data indicates each player must do to do well in the match. It does this using predictive analytics based on 8 years of grand slam data and 41 million data points.

SlamTracker said that Andy had to win more than 29% of first serve return points; and he didn’t.

It’s all in the keys.

(There’s a good article in Yahoo about it here.)

Leave a comment

Filed under Analytics

Making Wimbledon Relevant

In a previous post I was thinking about some of the exciting things we do with Wimbledon, and I thought it might be useful to add some practical applications of social and analytics in other industries.

I’m not going to talk about Big Data. If I’m honest I don’t really like the term. To me there’s just data. And there’s an abundance of it, some of it we own, some of it we don’t but we do have access to, some of it is highly accurate, some more questionable. There are many types we can make use of, from a variety of sources, in many shapes and sizes.  And a lot of that data can be from social media and from social business platforms – that is, from systems of engagement.

When that data is analysed it can allow you to do something you were already doing but do it better – because you have a better understanding.

It can allow you do something you weren’t doing but is related to a strategic objective such as understanding customer sentiment to become more customer centric.

It can even allow you to do something truly transformative such as real time traffic flow optimisation, as is  done in Dublin.

There the city uses data to identify and solve the root causes of traffic congestion in its public transport network. This means they improve traffic flow and provide better mobility for commuters. Data is taken from a citywide network of sensors, bus timetables, cctv and combined with geospatial data and the gps updates transmitted by the city’s 1000 buses every 20 seconds. Using this, the traffic can be monitored and managed in real time by those who have the responsibility in the city.

Based on the success we are now working on projects with Dublin and our Research organisation to add meteorological data into the traffic control centre so actions can be taken to reduce the impact of severe weather on commuters. We are also developing a predictive analytics solution which will combine the city’s tram network with electronic docks for Dublin’s free bicycle scheme.

We tend to divide analytics into three categories although there are other ways to do it. Those are descriptive – what happened, predictive – what is likely to happen – and prescriptive which not only anticipates what will happen and when it will happen, but also why it will happen, and suggests decision options to take advantage of the predictions.

I see a lot of organisations do the descriptive analytics, whether using more intuitive and interactive dashbords or just, dare I say, excel spreadsheets. Fewer are taking advantage of predictive, and even fewer prescriptive.

So, with the right type of analytics there all sorts of things one could do:

  • We can predict and act on the intent to purchase. It’s possible to identify what customers are researching and send this information to human and online channels. The SlamTracker keys to the Wimbledon game are based on prior player performance, and we can similarly understand customer behaviour and predict likely purchases.
  • We can truly personalise our interactions with the customer.  System U within Watson – needs just 200 tweets to understand an individual’s wants, needs, psychological profile, emotional style, and so on, and this – combined with any other data we may have about a customer – can allow us to tailor the right message for the right customer at the right time.  I talked about this at the TEDxUniversityofStrathclyde recently.
  • IBM helps Thames Water analyse a range of social media channels including blogs, online forums and Twitter to create real-time public opinion snapshots, identifying trends and usage behaviour while understanding how consumers feel towards the brand. But we are taking that analysis one step further and working with other water companies around the world to determine where there is a leak in the infrastructure using social media as a feed.
  • In Toulouse they use social media analytics <French site> to understand where they have a problem with their road infrastructure – pot holes to you and me – and they’ve cut response times down from 15 days to 1.
  • In the Netherlands and the US we’ve applied analytics to social media to understand the likely success of programme and film launches, and to take direct action to change the outcomes.

For Wimbledon data and insight is crucial to the fan experience.  The same can be said of all business, replacing the word “fan” appropriately – “employee”, “consumer”, “citizen”, and so on.

IBMslamtracker

Leave a comment

Filed under Social Business

Apps in business: bandwagon or reality?

These days everyone wants to access the function they need to do their job in the shape of apps for their smartphones and tablets don’t they?  More and more app stores are available whether iTunes, Windows Store, Google Play, IBM PureSystems Centre, and so on.

A conversation I had recently got me thinking more about how we consume IT and the changes IT delivery organisations will experience.  I also wonder about the hype cycle and where we may be on it these days.

Right now an organisation may find that consumers of its IT want access to just one or two functions via an app, and perhaps 10 apps will be created for 10 different functions.  (By function I do mean one type of interaction with some back end technology, whatever that may be, perhaps searching for client information.  I don’t mean the “Sales” function or other such organisation.)  And that sort of thing is proliferating, so perhaps we’re somewhere between the “Technology Trigger” and “Peak of Inflated Expectations” with lots of these new apps being developed.

But where do we draw the line?  That is, is it realistic or unmanageable to have to navigate between 10 or 20 apps to do one’s job?  As we move more and more to this new model driven by the consumerisation of IT I think we will hit that “Trough of Disillusionment” when it starts to get hard.  As it is I have well over 100 apps on my smart phone, and while very few of them are to do my job,  I expect that to change increasingly.  Management of that is going to be very hard.

apphypecycle

So, I’m thinking about what is next.   Will we ditch our smartphones, tablets and their apps, and go back to a desktop/laptop world to access enterprise applications?  Back to green-screens anyone?  I seriously doubt it.   There will clearly continue to be a place for both.

We’ll mature into a world with more feature-rich apps to allow us to do more from our smart devices in a sensible manner, and with better models for identifying which instruments are best for which tasks.

So, what changes?  Our architects must be able to design technology which has the flexibility to support a number of interaction models, and a variety of performance models, and our requirements gatherers (business analysts, system analysts, etc.) must understand our companies’ business models to better define who needs access to what function and information in what form.   The “business” must get closer to “IT”, DevOps must become BusDevOps, and whilst a technical person may think their business counterparts need to listen more perhaps technical leaders must learn to become trusted advisors.

Analytics will become increasingly important to allow us to understand the non-functional characteristics and apply that knowledge to future developments.  Is it possible for security to be any more important than it already is?  Perhaps not, but we may see more organisations begin to take it more seriously, and we certainly need to adapt our security measures more quickly.

And as these are just my initial thoughts they’ll evolve and mature themselves!

Leave a comment

Filed under Uncategorized