Lifting the lid on a live project: Bringing behavioural insights into NHS procurement

By Harriet Boulding, Saba Hinrichs, Hannah Burd, Mark Egan and Michael Hallsworth

In his February 2016 independent report, Lord Carter announced the creation of a ‘model hospital’. The purpose of this imaginary hospital is to show ‘what good looks like’ by providing a set of metrics, best practice checklists and benchmarks for other hospitals to build into their procurement processes.

This is a much needed endeavour. Lord Carter noted that ‘after analysing the variation in non-clinical resource costs we believe there is at least a £2 billion opportunity across the areas of procurement, estates and facilities and administration (back-office) costs.’


This is a challenge where the Behavioural Insights Team (BIT) and the Policy Institute at King’s College London are beginning to make some inroads in our work with a London hospital to hardwire behavioural insights into the development of an innovative NHS procurement system.

The fundamental problem with current purchasing processes, mentioned in Lord Carter’s report, is that most hospital trusts ‘don’t know what they buy, how much they buy, and what they pay for goods and services.’ This poses serious challenges for researchers, like us, trying to help – and especially if attempting to trial changes robustly, before cementing them into the new procurement infrastructure.

This is where BIT’s research partnership and trial plans are truly innovative and exciting. Traditional procurement platforms used commonly across the NHS are not set up at the moment for adaptation or experimentation. They are the researcher’s equivalent of a breeze block – difficult to remould or adapt. In contrast, Virtual Stock’s new Amazon-style procurement platform, that we’re using in our study, is more akin to putty or clay. Its developers update the system in an agile way, constantly responding to change. We’ve been able to make use of this opportunity as behavioural insights researchers. When they plan to run A/B tests on different versions of new webpages, we can ascertain which is better in terms of designing out wasteful procurement practices.


We initially planned to take a data-driven approach to identify hotspots of excess spending (what Lord Carter calls ‘unwarranted variation’) that can lead to waste. However, a challenge that BIT have identified is that the purchasing data we have access to is often not very fine-grained. For example, it might record a purchase of ‘surgical tapes’, but not specify their size (eg whether they were 7cm or 10cm) or quantity (whether they came in boxes of 50 or 200). Such data limitations make it challenging for both Trusts and researchers to identify wasteful spending. What looks like someone spending too much on surgical tapes might ultimately be cheaper per unit if they are buying in bulk.

Purchased datasets are also difficult to clean or crunch. BIT deals with challenges like this typically by being flexible and pragmatic when planning interventions. This might mean beginning a project with an idea of what the intervention will look like, but then modifying the idea if the data is very different to what we expected. In this case, we expect the introduction of Virtual Stock to have added benefits in terms of data quality. The current system collects data via a list of poorly entered codes and free-text boxes – so often subject to human error. Virtual Stock will automatically capture much richer and more accurate datasets. This in turn should enable better targeting of research activities once the system is live.


Another approach we have taken has been to place an ethnographer from the Policy Institute in the procurement offices. This allows us to keep abreast of developments in the hospital’s procurement and more broadly throughout the Trust. It’s already meant we were alerted to the introduction of a computer replacement project that could influence how the new procurement platform is received and understood by staff across the Trust.

This ethnographic approach has also highlighted the importance of understanding the existing procurement context in detail. This has taken BIT to new parts of the procurement pathway. Originally we started by focusing purely on purchasing behaviour: what happens when a member of staff tries to buy a product in the first place. But we are now discovering a diversity of behaviours across the almost 2,500 members of staff who are tasked with stocking the hospital in different departments. One area we have explored was where those products end up at the other end of the procurement chain (inventory cabinets). This has opened a number of new behaviours to investigate: why people habitually order new items even if they have a stockpile sitting in storage down the corridor; or why they struggle to accurately record the removal of stock items. We will blog more about our findings on that score in a later update.

This just gives a flavour of our work. In the coming months, BIT will continue to explore new, promising avenues of research with our procurement partners (the hospital and the platform) in creative, innovative ways. Watch this space for more announcements on this later in 2017.

Hannah Burd, Mark Egan and Michael Hallsworth are from the Behavioural Insights Team and work on this NHS procurement project with Harriet Boulding and Saba Hinrichs from the Policy Institute at King’s.

Leave a Reply

Your email address will not be published. Required fields are marked *