Collating and analysing vast amounts of project data efficiently can cut construction project costs by 20-25%. It can also prevent the loss of initial project data, of which 30% is typically lost by completion.
How we did it
A successful example of this is Linesight’s life sciences benchmarking initiative, which allows 17 participating life sciences companies operating globally – including AstraZeneca, Merck and Pfizer – to safely share and benchmark data from $26bn worth of construction projects.
Linesight collates cost and schedule data from participants on their completed projects. The data is contextualised and anonymized, allowing for analysis and interpretation around early-stage construction cost forecast via an interactive dashboard. This work allows us to streamline approaches to cost estimation, which will mean clients can make a stronger business case for a project and increase the level of cost certainty around future projects.
Guiding principles
To deliver the step-change in efficiency needed, predictive analytics should be brought in at an early stage. By collecting and, most importantly, understanding historical data to forecast future projections, projects can work within much more robust programmes.
Used and facilitated correctly, predictive analytics offers functional capacity metrics to schedule capacity during the design phase and can benchmark how many people will work in an office, or the total number of hotel keys against the total project cost. Crucially, predictive analytics can cut days and weeks taken to establish an output down to just a matter of minutes.
As our benchmarking initiative shows, there are several ingredients necessary for successful predictive analytics. The basis of a robust predictive model is accurate, current and auditable data. This data then requires consistent, relevant additions to enhance the ability of the model to make more accurate predictions, considering economic and geo-political factors. To include these non-numerical elements, we apply uplift factors, which are established based on market intelligence and past project data – we add cost escalation and adjust the data for geoparity using location factors. All stakeholders must be aware of the value accumulating such data adds and the subsequent analytical activity it provides.
Making this happen takes hard work and collaboration with the party who provided the data.
We, therefore, work with the data provider to understand the data and collect contextual information about projects that helps us structure the data in the correct way.
Main obstacles
While collecting good quality data is the main obstacle for clients, organisations often need to improve how the data is structured to enhance and facilitate it rather than just capture it.
Here is what companies need to do. First, consider data capture from the outset and integrate it into your processes by creating a golden thread through your project planning and execution processes, to avoid it being an added burden. Technology can then be used to connect disparate data sources together using APIs, speeding up the process and enhancing the data quality by being directly linked to the data sources. Investing in an AI platform without the resources to implement an adequate data structure is futile.
Wider organisational education is then required to ensure all staff, beyond the data analysts, understand how to review data sets and articulate results and appreciate the significance data holds when supporting costing and scheduling. Once all stakeholders are aligned, trust in the analysts and the data sets they provide increases.
Project supply chain collaboration and data-informed decision making will be fundamental when embracing the changes required to leverage the power of predictive analytics. The industry needs to ensure it is working with contractors and subcontractors who are willing to invest the time and resources into data collection and manage expectations during procurement, to align on data priority and consistency.
The source of this hardhatNEWS article is Placetech
Comments
Post a Comment