Phil Moss, Chief Technology Officer, takes us through a typical day at Procurement for Housing. I’m more of an early morning person, so my day starts around 6am when I make a quick coffee and leave my home in Lancashire. Depending on my schedule, I’ll either turn left and drive to our office in Manchester where our software development team is based – they are the ‘techies’ working on the software behind our procurement frameworks. But if I’m meeting the people who use our technology – PfH’s sales team and account managers – then I turn right and head to our main office in Warrington. Where ever I am, everything stops at 10.30am. The technology team (that’s six of us) takes part in a daily ‘scrum’ call where everyone across all of our sites answers three questions: we want to know what they did yesterday; what their plans are for today; and what blockers they’re facing. Today’s scrum focused on automating categorisation and how we can ensure that lowest-level transactional data is routinely classified to detect if a member is spending too much on a certain product. As the largest spend aggregator in the UK social housing sector, PfH collects a lot of this transactional data. We manage over a million invoices a year – that’s £250 million of spending. Our aim is to use this unique position to provide housing-specific insights to landlords. Technology, in particular machine learning, is a key part of achieving this. Machine learning is a type of artificial intelligence. AI uses computer programmes to think and learn like humans and machine learning is one of those programmes. It’s all about identifying patterns in historical data – algorithms learn those patterns and then forecast future trends. Historically, the housing sector hasn’t been great at managing data or categorising products – and this has led to a delay in the adoption of AI and smart analytics. There is very little granular detail in the sector, so it’s been hard to introduce machine learning for predictive analysis, or to compare spending data with public data sets to see if a housing provider is paying too much for, for example, kitchen refurbishments. The scrum meeting today focused on tackling this problem. Since I began at PfH around a year ago, I’ve used technology to put data at the heart of everything we do. Every single line of pricing, transactional and CRM information goes into our data warehouse where it is used for analysis and reporting. On a monthly basis, this data warehouse examines 300,000 lines of spending data from more than 900 housing providers. For the last six months, our team has been refining the quality of this data and developing technologies, bespoke to the social housing sector, which we can use to categorise information and provide members with intuition around their procurement spending. This data analysis was taken to a new level when PfH bought Valueworks. The system was specially designed for the social housing sector and it provides a collaborative, real-time view across all spending data so that housing providers can more effectively track prices, control costs and improve quality. The software enables us to group our members’ transactional data into specific programmes of work and then identify whether they are spending too much on certain products compared with their peers, whether they are purchasing several types of one product unnecessarily or whether there are better value alternatives available. After today’s scrum meeting, I meet with PfH’s six-strong account management team who liaise directly with our members on a day-to-day basis. One of the digital initiatives we’ve introduced recently is VFM reporting. This analytics service gives members insight into their spending over a particular period, highlighting saving opportunities. Reports are created using data dissected by our Microsoft PowerBI software and account managers present the reports to members. I’m meeting the team to take them through the latest capabilities of the system. Today, we talk about how the reports can tackle ‘product drift’, when an organisation spends less on the core products that PfH has negotiated reduced rates for, leading to larger bills. I also explained that VfM reports can pinpoint patterns such as a member that is spending more on a particular product compared with its expenditure last year. After lunch at my desk, I rush across the M62 to a meeting at Liverpool University’s School of Electrical Engineering, Electronics and Computer Science. PfH runs a knowledge transfer partnership (KTP) with the university to explore how machine learning can help housing providers’ procurement activities and I’m meeting our KTP associate, Dr David Hamilton. KTPs help businesses like PfH to innovate by linking them with research organisations like Liverpool University. They enable companies to bring in the latest skills (David has a PhD in distributed algorithms and machine learning is his specialist area) to deliver a specific, strategic innovation project. For us, that innovation project is showing housing providers the potential of their data and helping them to manage and use it in the right way. Our KTP is coming to an end and today I’m speaking to David about the next steps, particularly around using predictive analysis to learn from asset management invoice data and how we can link this data to price indices to show housing providers the best time to buy certain products. Back in Warrington, I meet Steve Malone, PfH’s managing director. We are discussing the latest smart procurement technologies to use with our members. Data from technologies such as IoT thermostats, window sensors or smart boiler parts is already recognising ‘failure in advance’ and this could help housing providers switch from reactive repairs to planned maintenance. Machine learning could be used to recommend comparison products, such as a boiler that is cheaper, has a longer warranty and a smaller carbon footprint. Housing providers could also use ‘emotional AI’ to analyse social media mentions about suppliers and combine this with data on contract performance, legal disputes or redundancies to build risk profiles. My day finishes around 5.30pm when I head home. If there is enough daylight remaining, I might jump on the bike and enjoy the Lancashire countryside before an early night. In reality, I probably spend much of the evening negotiating with my two children to convince them it’s bedtime! Phil Moss is the chief technology officer for Procurement for Housing. http://procurementforhousing.co.uk/ If you'd like to share your housing sector experience, all you need to do is get in touch at firstname.lastname@example.org.