Data is Critical for Effective Construction Management

Effective management and use of data is critical in driving optimal results on construction projects, a technology leader says.
 
Speaking with Sourceable at his company’s Construction Technology Summit held in Melbourne last month, Karthik Venkatasubramanian, VP, Data Science & Analytics at Oracle Construction & Engineering, talked about the use of data on construction projects.
 
According to Venkatasubramanian, there are several misconceptions about data use on projects.
 
First, many contractors feel that they cannot use data as they either do not possess such data or do not have it in a way which is integrated and accessible.
 
This, Venkatasubramanian, says, is incorrect. Almost all contractors have some records from previous projects and have produced some form of reports based on these. These could be used, for example, to look at causes of delays on previous projects and how these might be mitigated on future jobs.
 
Second, Venkatasubramanian says many overestimate the investment needed to derive value from data. Whilst flagship projects often showcase expensive data systems, he advises contractors to instead pilot small initiatives which deliver immediate value and to then seek feedback prior to scaling up.
 
On a related note, Venkatasubramanian stresses the need to focus on business problems which impact the success of your organisation. A common mistake, he says, is to focus on IT concepts such as machine learning or linear regression. Whilst these are fine technologies, the focus must remain on underlying business challenges which you are trying to solve.
 
A final misconception is that data is for the ‘big-end’ of town. Whilst smaller contractors and subcontractors might lack the scale to invest in AI or machine learning, they can still sort data on a spreadsheet. This can help to identify the best and worst paying contractors or the most and least reliable suppliers. This can be used to identify the contractors from whom they prefer to bid, the price at which they will bid and the suppliers they will preference. Applying linear regression, meanwhile, they can predict how their business will perform overall based on current trends.
 
According to Venkatasubramanian, the focus of data use in construction is evolving.
 
At the moment, most data use revolves around solving immediate problems and improving processes on the management of issues such as RFIs or variations. Much data being used, meanwhile, is historical and provides at best lagging or current indicators of performance.
 
Going forward, he says data use will evolve in several ways.
 
First, contractors are now seeking more integrated ways of using data. Where variations occur, for instance, they increasingly want visibility about how these will impact budgets and costs.
 
Beyond that, there will be greater use of data to identify and address future problems and challenges.
 
Speaking particularly of Oracle and its Oracle Aconex platform, Venkatasubramanian says much focus in the past has involved addressing common challenges such as ensuring that project team members are working with the most current set of documents. Nowadays, he says the platform provides dashboards which facilitate identification of issues such as outstanding RFIs, pending change requests and review approvals.
 
Going forward, he says the software will provide greater ability to make future predictions and to enable contractors to anticipate, for example, the number of variations which are likely in a given month.
 
Asked how effectively data is being used at the moment, Venkatasubramanian says the picture is mixed.
 
Whilst some use data effectively, there are other cases where potential uses for data are missed. Architects, for example, might use generative design in the design phase but may not have thought about using that data generated during this phase to price tenders or to conduct review and approval processes.
 
“I think it is varying degrees in terms of different options in how people are using data,” Venkatasubramanian said.
 
“Some do exceptionally well in some of the use cases.”