Peter Barker, CTO, Rufus Leonard explains how cloud technology can deliver powerful new insights for business leaders, and shares the steps to successful implementation – and explains why he believes the Azure platform offers the easiest way to do this:
Delivering on being an insight-led business requires a lot of data foundation work – the intelligence is just the tip of the iceberg. Fortunately cloud insight technologies have arrived and are finally allowing us to make use of all the ‘big data’ we have been collecting for years – but until now have struggled to use effectively. Often this has been because data sets and the data points we need are stuck in old or segregated systems and the investment for consolidation is often seen as a barrier.
If this is a new area and your business is asking you to provision an insight solution, then here are four keys steps:
1. Sort out your data estate
This can be hard as data quality can hinder progress; but in essence we now have solutions which ingest and orchestrate data from various sources, be that older systems, applications and increasingly sensors and devices. This then allows us to connect and analyse that data upon which we can then draw insight and serve that into tangible actions.
Fully connected elastic platforms which do this can now be rented with relative ease – such as Microsoft Azure’s suite of capabilities and services; initially we need to orchestrate an ingestion with tools like Azure Data Factory. This allows an automated data integration solution, code free, via a drag and drop UI. Alternatively, it may be necessary to use the technology to perform knowledge mining, where we can again use cloud services from Azure to perform ‘document cracking’, where we can use document element extraction tools available through the Cognitive Services suite to process data. This can then be stored through one of a number of methods such as a Data Lake store – this enriched data set is then ready to be used for delivering intelligence and insight relevant to the business use case.
2. Work out where to activate
The insight itself can be ‘generated’ in a number of ways, plug in Power BI and start to explore – often how you generate the insight will be based on the specific use case relevant to the business and this is where machine learning or artificial intelligence tools which are often forms of pretrained machine learning models come in.
3. Do something with the insight
Many models are fundamentally similar and the availability of pretrained models is increasing e.g. if you wish to understand a customer’s propensity to churn (a common use case) and identify who will do this, then you will need to understand the triggers, model them, test them on a representative data set and then probably keep iterating. Now cloud insight technologies don’t replace some key thinking – you still need to apply some data science to your specific data; to have a hypotheses; but using or creating a model can be relatively easy using Azure Machine Learning studio which is full or easy to use tools to help you choose and test the decision tree algorithm with your data and see the success (or not usually to start with) of your hypothesis. So, tweak, iterate and retrain the model and see where you go…
4. Operationalise
Once you have proven the value of the solution there are a number of tools to operationalise the solution and allow it to grow e.g. feed in more data, add more models – build customer experiences driven through insight and intelligence.
The Azure platform is the perfect environment to do all these things – start small renting the tools, prove success and ROI and grow the estate and capabilities that will ensure your business remains relevant.
