putting data to work: finding answers before you know the question
What if it were possible to find out what is really going on in your business? What if you could solve issues before the normal stage at which they’d flag as even being an issue? What if you could align how your customers think with exactly how you communicate with them? Where would you even start?
The gathering, storage and analysis of data has been making headlines. It’s been described as the most valuable resource on earth. The rise of data as an asset and data-driven insights as a way to get ahead, in business or in elections, has been fast, furious and more than a little scary.
But the reason such techniques are hitting the headlines is, that however you feel about the way they have been applied, done right they work. Yet the gap between simply gathering and holding a myriad of data and turning it into deep insights that catalyse change is still vast for many organisations.
At Egremont Group we are using data analysis software with our clients in the water industry to help them glean insights from data sets too vast to be analysed with human eyes alone. Once we have these insights we work hand in hand with them to develop solutions and enable the organisation to change.
To do this we are working with Relative Insight, a company founded by PhD students from Lancaster University specialising in linguistics. The team has coded every word in the English language and assigned them into topics, and also into emotions where relevant. The main innovation in this type of analysis is that it allows us to analyse qualitative data or free text in the same way that we can analyse numbers.
Originally developed for Child Protection Services, the programme can detect adults masquerading as children in chat rooms simply by studying the language used by verified children in a school intranet and comparing that with other general conversation (unconfirmed children / adults) in an internet chat room. With the ability to analyse five million words in 30 seconds it is easy to see the benefits.
In traditional data analysis the Analyst is looking for something. In the work we do with Relative Insight the interesting words and phrases that are discovered may mean nothing to the Analyst at first glance, but with some context, further exploration and discussion with relevant teams, patterns start to emerge and deeper, actionable insights can be gleaned. The Analyst is literally finding answers to questions they didn’t know to ask.
In the water industry we are using this type of data analysis to evaluate free text logs across the organisation to see what is really lurking beneath the surface.
Maintenance data – the devil in the detail
Operational data is often made up of a tick box, or standardised responses on a drop-down list. But these are frequently accompanied by ‘free text’ where an Engineer can add commentary. For years this information has been collected and filed away, creating vast and wide-ranging data sets containing insights and trends. It would take literally thousands of man hours for human eyes to scan and process every form. At Egremont Group we are combining Relative Insights’ cutting- edge tools with our experience of the water industry and hands on approach to transformation to uncover these trends and patterns in operational data that have remained hidden, and then thinking through how to change ways of working to realise benefits.
For example, in analysis of a three-month sample of operational data ‘Comms’ was a topic found to be more prevalent in one region than the others. Upon further exploration, we highlighted that a specific comms issue was causing this regional difference.
Through our experience in the water industry, we knew that this issue has been at the root cause of pollution events and believed our client should therefore proactively rectify the problem. By implementing a back-up system to warn the control centre when a communications problem has occurred Engineers can be on site to restore it quickly before there is a major incident rather than playing catch up. By collecting and analysing Engineers’ commentary, a data set that the organisation already had but had previously not known how to leverage, we can identify points of unidentified operational risk and make recommendations to address them before the risk is realised.
In a separate experiment we compared the commentary of Engineers’ jobs completed right first time in full, to those that were cancelled, or required a follow-on rework. The Relative Insight tool was able to detect that the word ‘valve’ was three times more likely to be included in the comments for jobs that required rework. To draw this type of insight from traditional analysis would require lengthy, detailed analysis as well as a preconception that there could be a relationship between a generic term like ‘valves’ and an inability to complete jobs right first time.
The real value to the organisation comes when the Engineers understand that the words they use in these free text forms are important and can be used to spot hidden issues and trends. Part of our role is to work with the business leaders to change behaviour within the organisation and encourage the Engineers to be appropriately expressive when writing up descriptions of jobs and their work. Using this approach emotion, like frustration, can be tracked as a lead indicator of underlying issues. Ultimately, we are changing the mind-sets and behaviours of Engineers and their managers so they understand the value of a rich problem description to predicting future asset failures.
Communicating with customers – are you speaking the same language?
How often do water companies send out a variety of different customer communications only to find that none of them ‘land’. With language analysis it is possible to scan the transcripts of all customer service calls over a set time period and order them by outcome i.e. positive customer engagement/ satisfaction scores. The words and phrases used by the customer and by the customer service operative can then be analysed to see which words are used to produce a positive outcome. Equally if the customers of a certain business are using much more informal or colloquial language and the business itself is always very formal there is a clear difference between the two. Meeting halfway can make the business appear more engaged and approachable to its core customer base and therefore drive up customer satisfaction scores in the long term. These findings can then be fed back into the organisation for training purposes and to inform the language style the organisation adopts in communications.
Of course, spotting the disconnect in language is only the first part of a longer-term behavioural change that will need to take place across the organisation. Matching your customer’s style of communicating is possible, but if you are going to do it then consistency will be needed across all communications channels.
Data as a valued partner
The examples above are just the start of how a data-based approach could benefit the industry in the future given the right tools. Language analysis is just one – how about using AI to review photos to spot hidden insights about failed assets?
The connections and patterns that can be discovered are only one part of the story. What we do then is up to us. When we work with water companies we have a host of tools at our disposal, but experience tells us that real sustainable change happens when we harness an organisation’s people as a part of the initiative. This will continue to be true, but as water companies are asked to do more, with less, taking a new approach to data analysis could help you identify potentially surprising barriers to success. Then you can start focusing in the right places and asking how you can break through those barriers.
So, are you asking the right questions to unlock your potential? Or could you do with some help from your data?
This article first appeared in Institute of Water on 30th August, 2019.
To find out more please get in touch / +44 (0) 20 7298 7878
