Does poor asset data quality undermine advances in analytics?

The short answer is yes. Of course, it does. You will never be able to do the more advanced analytics, including machine learning unless your data is up to scratch. The analytical potential that IOT and AI has begins and ends with proper data management.

So, how do you go about getting that asset data ready?

One approach I’ve seen across asset intense sectors is to build out pockets of work that focus on increasing data quality across a small set of data, or business unit, before expanding it out to the wider business. This approach can give stakeholders confidence around improved DQ results while giving the chance for first-time adopters of good data management principles to implement quickly, without the need to draw up a lengthy business case focusing on ROI.

Another approach is to set up a dedicated centre of excellence that promotes data management throughout the business. This can be challenging to implement in the first place and comes with high expectations from senior management following an upfront investment. However, if successful it should be able to penetrate multiple business areas increasing the data quality maturity level enterprise-wide.

Setting up a data management best practices group within an organisation is also a good step. This approach is less formal than the aforementioned centre of excellence but can look to identify who is putting data at the heart of their business unit. Look to agree on what good data management looks like, agree on a way to measure it and agree to report back frequently on the progress of maturity. Oh, and be harsh when assessing your own data. This isn’t a tick box exercise and you’re only fooling yourself if you rate it a 10/10 when it’s actually a 2!

Whatever you do, make sure you get buy-in from upstairs. Without senior sponsors getting involved, pushing through any data transformation becomes very tricky. One key development in this area over the last 18 months is the introduction of the Chief Data Officer. The role is still being defined by many organisations but what you can be sure of is that any CDO will want to know who in the business can push a positive data agenda.

Line-of-sight

Effectively managed asset data provides asset managers with valuable insights into factors such as asset conditions, spatial coordinates, failures, etc. However, if data producers don’t have line-of-sight as to why accurate data affects their day to day jobs, little is likely to change. There needs to be processes and rules in place that help monitor and asses to ensure they integrate accurate data gathering into their usual routines. Equally, you must take the time to listen to them and harness their knowledge. Instilling solid data management principles is about taking people on a journey so they can see the benefits. It’s not about forcing KPI’s, unnecessary rules and processes on them.

The future

That being said, the IIoT is advancing by leaps and bounds, meaning that the data gathering process may not even be necessarily a human job, in the long run.

It is entirely possible to automate some asset data quality measurement and monitoring activity. Considering the potentially intensive and long-standing issues to be ironed out, automation through connected devices combined with machine learning has the potential to clean up large quantities of legacy data.

Conclusion

I would recommend getting out to as many data management events, conferences, and meetups as possible. There are some superb success stories out there and an equal number of nightmares! Educating yourself on both is important for obvious reasons.

And remember, it’s a marathon, not a sprint! You’re not going to change a traditional asset management company into a data-driven one overnight.

P.S. This is not an exhausted list on how to deal with data management! Just a few approaches and views I have witnessed over the years. Comments and feedback are very welcome.

If you’re interested in this topic I recently wrote about the benefits of applying machine learning to asset-intensive environments, read it here.