A data expert, an academic, a project manager and a lawyer combined on 17 September to discuss how individuals and organisations working on complex construction projects might exploit more data-driven approaches. Recurring themes included taking a whole-life approach, and the need for organisations to nurture data as something that delivered long-term strategic value. Entrenched industry behaviours and attitudes might need to be overcome, but advances in project data analytics and early efforts to share pan-project show the potential to improve the sector’s performance.
Oakland Group’s Richard Corderoy set the scene. Major projects are, by definition, large, complex, involve large sums of money, and – importantly – attract huge stakeholder interest in both the delivery and the outcome.
Historically, many UK organisations have under-performed on such projects. They could do better, he said. He highlighted four areas: improved real-time or near real-time reporting, more insightful answers to questions, better forecasting of schedules and budgets, and use of new data analytics capabilities (AI, machine learning).
This may sound straightforward, but organisations frequently have disparate, disconnected information sources; manually inputted data often does not reflect reality; skills needs have evolved but people’s capabilities haven’t always kept pace; and many organisations struggle to find a ‘silver bullet’ approach that can be applied across their projects.
The solution? Corderoy gave four pointers:
- Begin with the end in mind. – Think about the indicators of distressed projects so that assurance activities are well-directed. Use more of the data that is collected during a project to understand the underlying dynamics. Make reporting faster, more accurate, and empirically based on the underlying data.
- Develop an end-to-end appreciation of the journey. – Harness the power of the high quality data that is collected, often in trying circumstances, and make sure it has real-world impact on project outcomes.
- Assemble the skills required. – Simply recruiting a data scientist is seldom enough; other roles will also be needed. And executives will need to buy-in to the data-driven project experience.
- Deliver value quickly and use it to reinforce the business case. – Demonstrating value early is invaluable. Confidence flows when systems are reliably forecasting where projects might be heading.
The state of the art and science of project data analytics
Professor Naomi Brookes from the University of Warwick talked about the state of the art and science of project management, drawing on research undertaken with the Association of Project Management. At its simplest, she defined project data analytics as
“the use of past and current project data to enable effective decisions on project delivery. This includes:
- Descriptive analytics presenting data in the most effective format
- Predictive analytics using past data to predict future performance”
Project data analytics interfaces with other concepts, including Big Data, machine learning and artificial intelligence, she said, underlining that AI will (eventually) extend beyond helping people understand where they currently are to enabling autonomous systems to make decisions.
The APM-funded research addresses two questions: What are we doing with project data analytics now, and what will we doing next (both the immediate next steps, and steps indicated by research but not yet used in practice).
Descriptive analytics is pretty much state of the art within most organisations, Brookes said. However, there can still be ‘geeks meet business’ barriers, the right data for decisions may not be identified, or it may not be presented effectively. Across organisations, particularly oil and gas, data is being shared across organisations, but (excepting examples such as Projecting Success’s Construction Data Trust) there are very few data-sharing initiatives in other areas, partly due to concerns about sharing data especially in traditionally adversarial industries.
Predictive analytics is where many individual organisations are taking their next steps (Network Rail working with Oakland Group, for example), though progress is hampered by the comparatively small number of projects available. But use of predictive analytics across organisations is still mainly the realm of academic research.
Brookes said progress in effective use of data will require more use of Big Data across portfolios; considerable pre-analytical challenges still need to be addressed for machine learning; and AI is a long way off. She summed up:
- On descriptive analytics: “Use your dashboards wisely”
- On predictive analytics: “Proceed with caution!”
Lessons from the front-line of data
Consultant project manager Rachel Heywood of Gleeds has a strong background in implementing and managing building information modelling (BIM) processes and data. Accordingly, she endorsed Corderoy’s point about starting with the end in mind, stressing that today’s projects are increasingly about managing the whole life of clients’ built assets and meeting end-users’ long-term information requirements.
Heywood says data has been part of professionals’ day-to-day working lives for years. Budgets, programmes, business cases, cost plans, tenders, risk registers and KPIs are just some ways in which data sources are routinely created and consumed. With digital transformation ongoing, some data is electronic, some has been digitised, and some is truly digital, and it is accessed across multiple internal and external systems. Data is also arranged in different ways, with different software used by different disciplines who – ‘geeks meet business’, again – may not speak the same language, or use the same classification systems or project stage descriptions.
The Gleeds Insight & Analytics team provides a range of services (database collection, cost and fee profiling, benchmarking, market reporting, etc), and – Heywood says – is leading industry change in reporting and creating dashboards. She likened Gleeds’ digital journey to a car.
- The core chassis is provided by its databases and data collection systems.
- The interiors relate to how Gleeds people collect, use and learn to trust the data.
- The body of the car is the external user interface, helping clients access what they need, when they need it, in the form they need it.
- The engine for this comes from AI and machine learning, supported by industry professionals able to fine-tune the engine through their knowledge of data nuances.
- Finally, at the heart of the system is the fuel: data – which has to be of the right type and input in the right way.
Heywood said every project manager needs to think not about outputs but about the outcomes they seek (and about the show-stoppers they want to avoid). Learning from past projects and identifying best practice is also vital, and Heywood cited use of data to support decision-making on a healthcare project.
She described work by Gleeds on a complex project at Grange University Hospital in Cwmbran. Based on previous project evidence, the project team proposed an approach incorporating Modern Methods of Construction. This, while marginally more expensive (less than 0.5%), delivered value for money, a 23% programme saving (cutting a 179-week programme by 42 weeks), cashflow certainty, better safety, and overcame skills shortages.
“We need to use data to inform,” Heywood said, shunning standardised, one-size-fits-all type approaches. “Our data outputs need to be useful and usable to the client and to the project, and to help deliver the required outcomes and success factors.”
Legal aspects of complex project delivery
Leeds-based Clarion construction lawyer Phil Morrison stressed the fundamental need for information to manage construction project relationships and associated commercial and information transactions. “As well as payments, the vast majority of disputes can be put down to the lack of information that has been shared between parties,” he said. “Where there is doubt, that’s where disputes creep in.”
Information needs therefore must be properly set out in construction contracts, he said. He described common characteristics of effective construction contracts (price, change management, programmes of work, provisions for delays, notices, dispute and payment processes, insurance arrangements, termination provisions). Like Heywood, he talked about BIM. From a lawyer’s perspective, he said that conventional contracts did not support the more collaborative design and construction processes involved in BIM, nor did they govern new data ownership issues. Greater clarity was also needed about outputs and outcomes, he said.
What process or behavioural changes are needed for effective data implementation projects?
Heywood stressed the critical distinction between outputs and outcomes: “As an industry, we have often focused on outputs: deliverables. We need to change our mindset, wear a more strategic hat, and understand that, for our clients, success is about outcomes. If we are only going to get the data we’ve always had, we are going to be guilty of collecting data about the wrong things.”
Brookes says we need to take data seriously. “Cavalier” approaches to progress reporting need to change and be based on good data. “A bit of data discipline is needed.”
Corderoy highlighted businesses who take data seriously are often the ones that have turned themselves into data businesses: “it becomes part of their DNA”. Avoiding garbage in / garbage out situations is fundamental: “don’t blame the people doing data entry; this is often because the organisation hasn’t valued data collection as an important activity.”
Should project data analytics be a basic discipline?
Brookes highlighted the need for improvement given that so many project fail. “Project analytics has been able to uncover real relationships between project performance and some very unusual things.” From experience on nuclear decommissioning projects, for example, she said “predictive analytics exposed to us things that we didn’t know mattered.”
Morrison talked about the constant need to save time and money: “If these are always going to be the goal, there will always be a race to the bottom. Why not use these tools to do these projects ‘properly’?”
Heywood agreed, adding that industry was also terrible at sharing knowledge, both in understanding successes but also understand where things have gone wrong. “If we are going to use predictive analytics, we are going to need a rich data source and the detail behind it, but also historical datasets too. And there must be value in sharing those historical datasets.”
Corderoy talked about Oakland’s work in the rail sector. “Individual project managers are often promoted for doing their bits of projects well, even if others did badly. We need to be looking really deeply at behaviours and processes within an organisation. Can we build systems where people work for the collective good, better outcomes for society and so on? Certainly, we aren’t set up to work that way at the moment.”
Do cultural issues affect our approaches to data?
Heywood felt the conversation showed that the APM and some people in the project management world are engaging with the challenges. “The construction industry can be very backward, doing things the way we’ve always done them. … We need to make a lot of behavioural change. The Construction Data Trust talking about sharing open data shows we are taking some baby steps, but we are a long way from solving the problem yet.”
Despite the continued rise of construction disputes, Morrison agreed there was room for optimism if the use of data became more prevalent, particularly if industry is less preoccupied with time and cost issues, and more focused on delivering wonderful end-products, with delivery and operation enhancing and benefitting all of the UK. “And if you don’t want to spend time with lawyers, do your contracts and your data properly!”
“Fundamentally, we have never taken project management seriously – that is, management of the whole project,” Brookes said, “There have been very few attempts to optimise the project as a whole, so we end up with lots of sub-optimisations along the way. And thinking about data shows up these sub-optimisations. Perhaps with data, we can think holistically and rise to the challenge.”
Corderoy agreed, adding that the development of data-driven approaches means that UK construction had never been in a better position to improve its infrastructure delivery performance.
