Why is productivity an intriguing topic for this post? Because it is at the heart of profitability. If your workforce and machines were more productive, you would decrease variable costs and improve business profitability. If you could understand how productivity may have varied over time or specific activities, then you would potentially be able to control productivity and, therefore, profitability.

Keytree and Costain, a leading government main contractor, have been working together on a problem that has challenged economies and business for years, if not decades. Namely, how do you get a reliable, insightful and actionable view of productivity in a business? The first challenge in meeting the ambition to control productivity is to answer the question: What do you mean by productivity? The economic definition of productivity is:

Productivity (noun): The state or quality of being productive. The effectiveness of productive effort, especially in industry, as measured in terms of the rate of output per unit of input. “Workers have boosted productivity by 30%.”

Richard Benson

Head of Data Science

Not judging on face value

On the face of it, this seems very clear, but when you pause for a second to think about how you would measure it in a real-world situation, it becomes infuriatingly vague. As we are working closely with Costain, the construction industry is the first vertical we are considering, so following this vein of thought, let us consider building a wall versus the above definition:

Challenge #1: How do we define the rate of output?

Is it how many m2 or m3 of bricks you lay in a day? But what about the laying of foundations – how can you comparably measure that? Then you add value to the wall by adding windows or a door – how are these outputs measured?

Challenge #2: How do we measure the unit of input?

Is this measured by how many people are employed doing the task? What if you use, for example, the AcmeMagicWall machine that only needs one operator but costs the same as three people per day and digs foundations at the same time as laying bricks, but can only be used for some types of wall? Do we measure the productivity of a person or the cost of operating the whole team?

The classic way economies measure productivity at a national scale is explained by the World Confederation Productivity of Science as follows: “In terms of Gross Domestic Product (GDP) per capita, per employed person or hour worked. It is viewed by many as a key indicator of the economic health of the country. ”Therefore, given that economies have diverse activities, the solution to measure at a national level has to be an economic view of output and input. This gives us a potentially interesting way we can solve our problem of looking at diverse activities within a project. However, this then opens up new cans of worms.

Measuring cost

Should we measure the cost to the end customer, or the amount paid by the main contractor? Or should we drill deeper and measure the profitability of the project to the subcontractors and their sub-sub-contractors? You can see how it quickly results in a big headache.

Finally, so what! Even if you decide on the approach for the above, then what are you going to do with the final magic figure for productivity? It needs to be actionable. How will knowing the number affect your plans going forward, and what can you change to improve productivity? We wanted to approach the above problems in a way that would:

  • Be transferable across many different types of projects, both within the construction industry and beyond that to other industries
  • Be consistent, so you are not just measuring the productivity of building one type of thing, be it a motorway, a railway, or an office block
  • Have a link to causality, so we could understand why, on the face of it, the productivity of building a wall is different from foundation digging
  • Be actionable, so we could see that there were risks of a benchmark anticipated productivity of drilling piles next January, not being met

To do this, we designed a tool we call the Intelligent Productivity Engine that brought together many data sources coming from different teams in sufficient quantity to enable correlation with potentially causal influences that may affect productivity. To normalise across activities, we devised an econometric method to measure the contribution of any given task to the overall project output and measured the input by looking at raw costs.

This approach gave us a way of breaking down different types of activity and scoring productivity against time, activity specifics, supplier and environment. In normalising, at this level, we were also able to get a standardised score across different types of projects that then enabled us to correlate against a flexibly defined set of influencing factors. We have built an initial set of:

  • Environmental: Weather, ground conditions, geology and traffic
  • Project-specific: Site access times and site location
  • Economic: Local, national and worldwide macroeconomic conditions
  • Health: Time of year versus (e.g.) the flu season as well as measuring the effects of black-swan type pandemic events around Covid-19

Managing influencing factors

The list of influencing factors is highly flexible to enable the exploration of single and multi-variable effects. The challenge of joining this disparate set of data sources was alleviated through the use of SAP Data Intelligence. It provided a graphical tool to build a federated data model as well as easing the knowledge sharing with Costain, showing the technical and non-technical communities inside Costain just how we are combining data. This also eased the communication of ideas and ultimately led to fresh insights coming from different experts and stakeholders within the firm. The tool is already proving useful in helping Costain communicate with commissioning agencies so that they can work with the customer, to actively highlight how projects can be specified, to maximise delivered value. The wealth of data and domain expertise that Costain allows us to compare both large numbers of similar projects as well as contrasting against totally different project types, such as road versus rail.

As the Costain team builds out the parent Intelligent Infrastructure Control Centre, more main contractors will join the initiative. The Productivity Engine component will be proven against even more varied project and commercial data, ultimately giving the DfT a tool to enable the smart specification of infrastructure projects, potentially saving billions of pounds in the process. Beyond this, many other industries would also benefit from being able to analyse large projects with complex teams and supply chains, from automotive to pharmaceuticals, energy and utilities to aerospace.

In future blogs, we will report how the tool enables active research into productivity, the detailed findings of how productivity varies due to the above influencing factors while gaining confidence and reassurance as expected causes emerge. Hopefully, we will also gain new insight as evidence builds to highlight less obvious causes of productivity variation.